Pentium vs PowerPC : my personal take.
Dennis O'Connor
The following is indeed purely MY opinion, and I haven't even
checked the official Intel PowerPC* position to write it.
Okay : Pentium(tm) processor versus PowerPC* 601 processor :
Hardware:
Right now, According to Motorola's ads, the integer performance is
slightly in favor of the Pentium processor, and the PPC601 enjoys
about a ~50% advantage in floating point. PPC601 pulls (according
to the Motorola ads) 8.5W worst-case at 3V, while the latest stepping
of the Pentium processor pulls I think less than 14W worst-case at 5V.
Pentium chips cost $965 (I think) in Qty.1K, PPC601 costs ~$450 in Qty.20K.
Now considering the Qty20K pricing for the Pentium will be bit lower
than Qty1K pricing, and that a 8.5W 3V part (drawing 2.8A) will, if
powered via a 5V-to-3V regulator (as many desktop systems might do today,
tho very few will in the future) dissipates a chip+regulator total
of 14W from the 5V supply, well, it really looks like the Pentium
chip and the PPC601 are technologically about even, unless "native"
3V system operation or fast FP are a critical need.
Also consider that the price difference between a PPC and a Pentium
isn't enough to pay for the re-purchase of even two or three major
software packages. That your existing software will run at 486DX33
speeds on a PPC601 isn't a persusive reason to buy a whole new system.
The future in hardware : well, PPC601 has already made the jump
to 3V, And the next Pentium family chip, the P54C, will be 3.3V too.
Not even taking into account the fact that the P54C is on a better
process than P5 (therefor having smaller die size and higher speed),
just the reduction in the power supply voltage from 5V to 3.3V
_should_ (I don't _know_, I'm not involved in it) drop power
to 45% of the 5V version : 14*.45 = 6.3W, a little less than PPC.
PowerPC won't be able to reap the power benefits of "down-volting"
again until 2V or less is accepted : how long till then ?
BTW P54C is _not_ the P6 : it is, according to Intel's announcements,
due out before the end of the year, with volume production in 1Q94.
Will there be a PPC successor in that time frame ? I don't know.
OS:
Right now, the Pentium processor can run either of three vendor's "DOS"s
(MS-DOS*, PC-DOS*, DR-DOS*), any of four vendor-proprietary windowing
OSs ( Windows, WindowsNT*, OS/2*, and NextStep486* ) and any of a large
number of UNIX versions ( Interactive, ESIX, SCO Unix*, Solaris*, ... ).
It's rumored that the MAC OS will be ported to the 486 as well.
It's said that PPC will also run a bunch of OSs. I don't doubt it,
but more than the Intel architecture does now ? I _do_ doubt that.
My conclusion:
People in the past have said that a brand new architecture has to
have at least twice the performance or twice the price/performance
to displace any existing one. Well, PPC601 (according to Motorola)
isn't twice the performance, and when viewed from a system cost
perspective, it's not twice the price/performance either. It still
has a future : the 10 million MACs out there are now end-of-lifed,
and PPC has as good a shot as any other architecture of winning
the big share of that market. But x86 will win some of it, too.
Especially if the MAC OS is ported to 486, as some have said it will.
IMO, technically, MIPS processors would be a bigger threat to Intel's
architecture than PPC, except they don't have the marketting and production
muscle and installed user base IBM, Apple and Motorola bring to the game.
*Trademarks are the property of their respective owners.
--
Dennis O'Connor doco...@sedona.intel.com
Intel i960(R) Microprocessor Division Solely responsible for what I do.
Charlie Kuehmann
doco...@sedona.intel.com (Dennis O'Connor) wrote:
>
> The following is indeed purely MY opinion, and I haven't even
> checked the official Intel PowerPC* position to write it.
>
> Okay : Pentium(tm) processor versus PowerPC* 601 processor :
>
[much comparison's deleted]
Your comparisons were based on assumptions that the PPC 601 was a 3V chip.
My impression was that the first PPC chip the 601 would be a 5V chip and
the second the 603 would be the low power 3V chip. Can anyone who has the
MOTO spec sheets on the PPC confirm this? If the 601 is indeed a 5V chip
then I think the power dissipation numbers would definately favor the PPC.
--
Charles Kuehmann
Northwestern University
Steel Research Group
qm...@casbah.acns.nwu.edu
Steve Kanefsky
Dennis O'Connor <doco...@sedona.intel.com> wrote:
>
>Also consider that the price difference between a PPC and a Pentium
>isn't enough to pay for the re-purchase of even two or three major
>software packages. That your existing software will run at 486DX33
>speeds on a PPC601 isn't a persusive reason to buy a whole new system.
You seem to be aware that quantity can be a big factor in pricing, since
you negate a lot of the price advantage of the PPC over the Pentium
because the PPC price is quantity-20K and the Pentium price is
quantity-1K. But guess what: there's a heck of a lot more difference
between quantity-1 pricing and quantity-1K pricing than between 1K and
20K. How can you compare quantity-1 software prices with quantity-1K or
20K differences in chip prices? Even worse, you seem to think that
you'll have to purchase a brand new copy of the software, rather than
paying a small upgrade fee (which you probably would have paid anyway to
upgrade to the next version -- the processor upgrade is thus effectively
free). You're also ignoring the fact that Pentium owners will have to
upgrade their software as well to take full advantage of their machines,
otherwise you have to give the performance advantage to a recompiled PPC
app over a non-recompiled '486 app running on a Pentium.
Finally, you're only considering differences in the CPU itself. A brand
new Mac designed around the PPC is likely to take better advantage of
the PPC processor than a '486 with a Pentium dropped in. Macs already
have a high-performance bus, disk controller, network, etc. built in, so
it won't add much to make a complete high-performance PPC Mac. '486
machines need major reworking to take advantage of the Pentium chip,
which will add much more than the CPU cost to good Pentium systems.
Cheap Pentium systems (i.e. '486 systems with a Pentium thrown in) may
be almost as cheap as PPC machines, but won't perform well because
they'll have bottlenecks in disk, bus, and network access that prevent
the CPU from realizing its full potential.
--
Steve Kanefsky
Mel Martinez
doco...@sedona.intel.com (Dennis O'Connor) wrote:
>
>
> checked the official Intel PowerPC* position to write it.
>
> Okay : Pentium(tm) processor versus PowerPC* 601 processor :
>
> Hardware:
> Right now, According to Motorola's ads, the integer performance is
> slightly in favor of the Pentium processor, and the PPC601 enjoys
> about a ~50% advantage in floating point. PPC601 pulls (according
> to the Motorola ads) 8.5W worst-case at 3V, while the latest stepping
> of the Pentium processor pulls I think less than 14W worst-case at 5V.
> Pentium chips cost $965 (I think) in Qty.1K, PPC601 costs ~$450 in Qty.20K.
>
Keep in mind that the major vendors of systems might purchase lots
differently. For instance IBM and Apple, the announced primary vendors of
PPC systems will be buying in HUGE lots of perhaps Qty.100k+. On the other
hand "Bibbo The Generic CloneMaker(BTGC)" may only be purchasing in lots of
1000. For a chip that is overall cheap (like the current 486) the
difference in price from the two lot sizes is small enough for the clone
maker to compete with, say Compaq. But if there really is a $400-500
difference in the price that Compaq pays per Pentium versus what BTGC pays,
then BTGC cannot hope to compete for the consumer unless either Compaq
keeps exhorbitant margins (thus all Pentium systems stay high priced) or
BTGC cuts his margins (and quality) to the bone. The reason is simple:
wouldn't you rather buy a Compaq over a BTGC unless the BTGC was LOTS
cheaper? I predict that most cheap clonemakers will wait quite awhile for
the big boys (like Compaq, Tandy and IBM, assuming IBM markets Pentium
systems) to bring the overall chip price down before jumping into the pool.
Most will milk the '486 market as long as possible and I bet a few cast
longing glances at the PPC. Note that many smaller companies probably buy
via co-op so that may help them somewhat.
And also, the fact that IBM itself will be the main manufacturer of the PPC
implies that THEY at least will be getting PPC601 chips at a good rate.
> Now considering the Qty20K pricing for the Pentium will be bit lower
> than Qty1K pricing, and that a 8.5W 3V part (drawing 2.8A) will, if
> powered via a 5V-to-3V regulator (as many desktop systems might do today,
> tho very few will in the future) dissipates a chip+regulator total
> of 14W from the 5V supply, well, it really looks like the Pentium
> chip and the PPC601 are technologically about even, unless "native"
> 3V system operation or fast FP are a critical need.
>
The total power consumption by the system is irrelevant. The problem with
the Pentium (in its current incarnation) is that this power is dissapated
in the chip itself. Localizing the heat to one component means higher
temperatures and increased chance of a component breakdown. If you
dissapate it in an external regulator/transformer system, big deal, that
hurts absolutely nothing since that is easy to locate far from any heat
sensitive components. It also means that transport is easier because the
increased surface area (from dissapating in multiple components) gets the
heat into the air which can be moved easily. So basically the above
paragraph is wrong, at least as far as heat/power handling technology goes.
The PPC and the Pentium are definitely NOT 'technologically about even'.
Your comments about futurelow-power Pentium versions (the P54C) catching up
may or may not be relevant. Since both it and the even lower power PPC603
are being developed on similar schedules, I don't see at the moment that
Intel will be catching up with it (in regards to the low=power consumption
issue), but you may be right.
Low power consumption at the component level is actually a very important
issue since it grows in importance with the number of components. And at a
fundamental level, remember that higher clock speeds through a given
transister mean higher current flows and thus higher heat production. If
you can get a component to work at a given speed with less power consumed,
then you are a closer step towards developing the all-important faster
versions. All those R&D $$$ that Intel, Motorola and others have been
throwing at this issue should indicate to you just how important it is.
>
> OS:
> Right now, the Pentium processor can run either of three vendor's "DOS"s
> (MS-DOS*, PC-DOS*, DR-DOS*), any of four vendor-proprietary windowing
> OSs ( Windows, WindowsNT*, OS/2*, and NextStep486* ) and any of a large
> number of UNIX versions ( Interactive, ESIX, SCO Unix*, Solaris*, ... ).
With the possible exception of NeXTStep and some of the miscelaneous unix
versions, all these will be running on the PPC as well.
> It's rumored that the MAC OS will be ported to the 486 as well.
The recent rumour mill says that this project died with Apple's recent
lay-offs.
> It's said that PPC will also run a bunch of OSs. I don't doubt it,
> but more than the Intel architecture does now ? I _do_ doubt that.
>
Probably right about that. PowerOpen, though, should allow one to run
multiple OS's simultaneously as processes within X that can be simply
switched between with a mouse-click. I don't believe you will see
PowerOpen ported to Pentium.
At any rate, all we can do is watch it all happen. Whoever wins, I am
personally pretty excited at the fact that in less than a year's time the
level of power/$ available for a personal computer will take a fundamental
leap. Pretty cool, I think.
Mel Martinez
The Johns Hopkins University
Dept. of Physics
m...@pha.jhu.edu
Dennis O'Connor
kane...@halcyon.com (Steve Kanefsky) writes:
] In article <DOCONNOR.93...@potato.sedona.intel.com>,
] Dennis O'Connor <doco...@sedona.intel.com> wrote:
] >
] >Also consider that the price difference between a PPC and a Pentium
] >isn't enough to pay for the re-purchase of even two or three major
] >software packages. That your existing software will run at 486DX33
] >speeds on a PPC601 isn't a persusive reason to buy a whole new system.
]
] You seem to be aware that quantity can be a big factor in pricing, since
] you negate a lot of the price advantage of the PPC over the Pentium
] because the PPC price is quantity-20K and the Pentium price is
] quantity-1K. But guess what: there's a heck of a lot more difference
] between quantity-1 pricing and quantity-1K pricing than between 1K and
] 20K. How can you compare quantity-1 software prices with quantity-1K or
] 20K differences in chip prices?
Software prices are usually quoted Qty. 1, regardless of whether
the software is for Pentium(tm) or PowerPC(tm) chips. However, the
only prices in the public domain for Pentium chips is the Qty. 1K
price, and the only price in the public for PC601 is the Qty. 20K
price. Therefor, comparisons of these two chips prices, based on the
public data, have to be fudged to compensate for the high-volume
discount built-in to the PPC price but absent from the Pentium chip price.
Also, The average user buys his system from a manufacturer who buys
processors in large volume. Therefor, the user cost of the processor is
more affected by the high-volume price than the Qty. 1K.
] Even worse, you seem to think that you'll have to purchase a brand new
] copy of the software, rather than paying a small upgrade fee (which you
] probably would have paid anyway to upgrade to the next version
The "upgrade fee" for, say, AutoCAD, is pretty stiff. Also, has any
software maker announced this cross-platform upgrade program ?
Discount upgrade programs exist to generate extra revenue : most
users of V5.1 just plain _will_ _not_ pay full price for 5.2,
so software makers sell it to them at only, say, 2X cost, instead
of 10X, and add some to the bottom line. This is not the situation
someone who moves from host to another is in : they really need
a new version of the software (unless they are willing to tolerate
the performance decrease of emulation, in which case, why switch
hardware?), so they will pay full price.
IMHO, The big software vendors like having One Huge Market for their
software, as opposed to lots os smaller markets, anyway. They
aren't going to sacrifice _their_ profits to encourage diversity
in the installed hardware base. For an example, look at Microsoft :
they aren't willing to port WindowsNT, they are letting Motorola
pay for that.
] You're also ignoring the fact that Pentium owners will have to
] upgrade their software as well to take full advantage of their machines,
] otherwise you have to give the performance advantage to a recompiled PPC
] app over a non-recompiled '486 app running on a Pentium.
Fact1: Pentium chips runs regular 486 programs faster than even a DX2-66.
Fact2: Code recompiled with the Pentium chip inspired optimizations runs
faster on a 486 than code from older compilers does
Fact3: Programs compile for Pentium chips will run on 486s.
Fact4: Because of Facts 2&3, many ISV (Independant Software Vendors) have
announced that all future code will use the new optimizations.
This is a different scenario from PPC 601, which does not run 486 code
faster than a same-price 486, and whose code can not be run on a 486.
] Finally, you're only considering differences in the CPU itself. A brand
] new Mac designed around the PPC is likely to take better advantage of
] the PPC processor than a '486 with a Pentium dropped in.
Brand new "PCs" are being designed for the Pentium chip, too.
Plus, motherboard replacements are very common today. I don't
see where _either_ chip has an advantage in this arena.
Dennis O'Connor
qm...@casbah.acns.nwu.edu (Charlie Kuehmann) writes:
] doco...@sedona.intel.com (Dennis O'Connor) wrote:
] > The following is indeed purely MY opinion, and I haven't even
] > checked the official Intel PowerPC* position to write it.
] >
] > Okay : Pentium(tm) processor versus PowerPC* 601 processor :
] >
] [much comparison's deleted]
]
] Your comparisons were based on assumptions that the PPC 601 was a 3V chip.
] My impression was that the first PPC chip the 601 would be a 5V chip and
] the second the 603 would be the low power 3V chip. Can anyone who has the
] MOTO spec sheets on the PPC confirm this? If the 601 is indeed a 5V chip
] then I think the power dissipation numbers would definately favor the PPC.
It would be nice if someone could independantly confirm, but my
source for the 3V assumption is the Motorola PPC601 literature.
Specifically, page 2 of the April 93 version of
"Technical Summary, PowerPC(tm) 601 RISC Microprocessor"
Copyright 1993 Motorola Inc. Order code MPC601/D
Dennis O'Connor
m...@pha.jhu.edu (Mel Martinez) writes:
] doco...@sedona.intel.com (Dennis O'Connor) wrote:
] >
] >
] > The following is indeed purely MY opinion, and I haven't even
] > checked the official Intel PowerPC* position to write it.
] >
] > Okay : Pentium(tm) processor versus PowerPC* 601 processor :
] >
] > Hardware:
] > Right now, According to Motorola's ads, the integer performance is
] > slightly in favor of the Pentium processor, and the PPC601 enjoys
] > about a ~50% advantage in floating point. PPC601 pulls (according
] > to the Motorola ads) 8.5W worst-case at 3V, while the latest stepping
] > of the Pentium processor pulls I think less than 14W worst-case at 5V.
] > Pentium chips cost $965 (I think) in Qty.1K, PPC601 costs ~$450 in Qty.20K.
] >
]
] Keep in mind that the major vendors of systems might purchase lots
] differently. For instance IBM and Apple, the announced primary vendors of
] PPC systems will be buying in HUGE lots of perhaps Qty.100k+. On the other
] hand "Bibbo The Generic CloneMaker(BTGC)" may only be purchasing in lots of
] 1000.
Most PCs are built by the "big guys" : IBM, Compaq, Dell, etc.
And a lot of companies OEM their PCs from Intel itself ! So
the playing field here is level.
] Your comments about futurelow-power Pentium versions (the P54C) catching up
] may or may not be relevant. Since both it and the even lower power PPC603
] are being developed on similar schedules, I don't see at the moment that
] Intel will be catching up with it (in regards to the low=power consumption
] issue), but you may be right.
Well, I haven't seen anyone say PPC603 would be shipping in volume
in 1Q94, when P54C is supposed to be. And as I said : PPC603 can't
exploit the power saving of the 5V-to-3V conversion, which P54C will.
We'll see what happens.
] > OS:
] > Right now, the Pentium processor can run either of three vendor's "DOS"s
] > (MS-DOS*, PC-DOS*, DR-DOS*), any of four vendor-proprietary windowing
] > OSs ( Windows, WindowsNT*, OS/2*, and NextStep486* ) and any of a large
] > number of UNIX versions ( Interactive, ESIX, SCO Unix*, Solaris*, ... ).
]
] With the possible exception of NeXTStep and some of the miscelaneous unix
] versions, all these will be running on the PPC as well.
]
] > It's rumored that the MAC OS will be ported to the 486 as well.
]
] The recent rumour mill says that this project died with Apple's recent
] lay-offs.
Okay. Thanks for the update.
]
] > It's said that PPC will also run a bunch of OSs. I don't doubt it,
] > but more than the Intel architecture does now ? I _do_ doubt that.
]
] Probably right about that. PowerOpen, though, should allow one to run
] multiple OS's simultaneously as processes within X that can be simply
] switched between with a mouse-click. I don't believe you will see
] PowerOpen ported to Pentium.
I've heard IBM dropped the MAC OS support for PowerOpen recently.
Also, I've always assumed PowerOpen was YAPUI : Yet Another Proprietary
UNIX Implementation. I could be wrong. How does it compare to NextStep ?
] At any rate, all we can do is watch it all happen. Whoever wins, I am
] personally pretty excited at the fact that in less than a year's time the
] level of power/$ available for a personal computer will take a fundamental
] leap. Pretty cool, I think.
Yep. And the intense competition makes things fun (I mean FUN) for
chip designers, so I've got no complaints !
Graham K. Johnson
>In article <DOCONNOR.93...@potato.sedona.intel.com>,
>doco...@sedona.intel.com (Dennis O'Connor) wrote:
>> OS:
>> Right now, the Pentium processor can run either of three vendor's "DOS"s
>> (MS-DOS*, PC-DOS*, DR-DOS*), any of four vendor-proprietary windowing
>> OSs ( Windows, WindowsNT*, OS/2*, and NextStep486* ) and any of a large
>> number of UNIX versions ( Interactive, ESIX, SCO Unix*, Solaris*, ... ).
>
>With the possible exception of NeXTStep and some of the miscelaneous unix
>versions, all these will be running on the PPC as well.
Which leaves you with "all" meaning Windows, WindowsNT, and OS/2. I
had heard about NT, and OS/2 doesn't suprise me, but will Windows
3.1/4.0/Chicago be ported to PPC?
>> It's said that PPC will also run a bunch of OSs. I don't doubt it,
>> but more than the Intel architecture does now ? I _do_ doubt that.
>
>Probably right about that. PowerOpen, though, should allow one to run
>multiple OS's simultaneously as processes within X that can be simply
>switched between with a mouse-click. I don't believe you will see
>PowerOpen ported to Pentium.
The recent rumor mill says that this project died with Apple's recent
lay-offs.
--
___
/__ ___ "Uh, you're not gonna joh...@cps.msu.edu
<__/raham / eat that, are you?" Michigan State University
|<. <_/ohnson --Fish Tales East Lansing, Michigan
Yoshio Turner
>In article <DOCONNOR.93...@potato.sedona.intel.com>,
>doco...@sedona.intel.com (Dennis O'Connor) wrote:
>>
>> The following is indeed purely MY opinion, and I haven't even
>> checked the official Intel PowerPC* position to write it.
>>
>> Okay : Pentium(tm) processor versus PowerPC* 601 processor :
>>
>[much comparison's deleted]
>Your comparisons were based on assumptions that the PPC 601 was a 3V chip.
>My impression was that the first PPC chip the 601 would be a 5V chip and
>the second the 603 would be the low power 3V chip. Can anyone who has the
The 601 requires a 3.6-volt power supply. Interface levels
are TTL or CMOS compatible. So says Microprocessor Report.
Yoshio Turner
yos...@cs.ucla.edu
Tom Barrett
>The following is indeed purely MY opinion, and I haven't even
>checked the official Intel PowerPC* position to write it.
>Okay : Pentium(tm) processor versus PowerPC* 601 processor :
Thanks for the comparison Dennis... it was well presented.
--
|Tom Barrett, Sr. Engr., SysArch Research | 1-800-538-8450, ext. 56856 |
|AMD PCD MS-522|5900 E. Ben White|Austin, TX 78741|tom.b...@amd.com|
|AMD's VESA Local Bus and InfraRed Data Association Guru... __|
|My views are my own and may not be the same as the company of origin\/|
Tony Hamilton
|> Software prices are usually quoted Qty. 1, regardless of whether
|> the software is for Pentium(tm) or PowerPC(tm) chips. However, the
|> only prices in the public domain for Pentium chips is the Qty. 1K
|> price, and the only price in the public for PC601 is the Qty. 20K
|> price. Therefor, comparisons of these two chips prices, based on the
|> public data, have to be fudged to compensate for the high-volume
|> discount built-in to the PPC price but absent from the Pentium chip price.
I wonder if it is also relevant to think about _who_ is buying these chips
in 20K volumes - Apple and IBM, right? Well, for heaven's sake, they're _partners_
with Motorola on this deal. Is _anyone_ else placing orders for PPC in those
volumes? Is Ford yet (would be interesting to know, althought I imagine a
subcontractor may do this for Ford - or do they do it themselves?)
In addition, leaving aside all the shots in the dark at comparing actual cost
to manufacture PPCs vs. Pentiums, the prices _surely_ reflect a discount intended
as an incentive to get the PPC into the market. You can't exactly be charged
for being anti-competitive either, because in this case you're trying to
_break_ an apparent monopoly, not create one.
There's always a point at which these factors level out.
|> Also, The average user buys his system from a manufacturer who buys
|> processors in large volume. Therefor, the user cost of the processor is
|> more affected by the high-volume price than the Qty. 1K.
This is very true, Dennis. Folks keep forgetting that the Pentium is out there
now because of demand. Current pricing can't possibly reflect a situation where
supply and demand are relatively in sync.
|> ] You're also ignoring the fact that Pentium owners will have to
|> ] upgrade their software as well to take full advantage of their machines,
|> ] otherwise you have to give the performance advantage to a recompiled PPC
|> ] app over a non-recompiled '486 app running on a Pentium.
|>
|> Fact1: Pentium chips runs regular 486 programs faster than even a DX2-66.
|> Fact2: Code recompiled with the Pentium chip inspired optimizations runs
|> faster on a 486 than code from older compilers does
|> Fact3: Programs compile for Pentium chips will run on 486s.
|> Fact4: Because of Facts 2&3, many ISV (Independant Software Vendors) have
|> announced that all future code will use the new optimizations.
|>
|> This is a different scenario from PPC 601, which does not run 486 code
|> faster than a same-price 486, and whose code can not be run on a 486.
Bravo, Dennis. Good point. Corporate buyers want _standardization_. Buying
PPC's means offering and supporting new software (assuming recompiled software
here), or suffering in performance. The loop simply isn't as tight as it is
with the Pentium-486 situation Dennis describes.
|> ] Finally, you're only considering differences in the CPU itself. A brand
|> ] new Mac designed around the PPC is likely to take better advantage of
|> ] the PPC processor than a '486 with a Pentium dropped in.
|>
|> Brand new "PCs" are being designed for the Pentium chip, too.
|> Plus, motherboard replacements are very common today. I don't
|> see where _either_ chip has an advantage in this arena.
Except that many of the basic architectural considerations in a Pentium system
are still similar to existing Intel architecture systems. On the other hand, does
a PPC system architecture resemble existing RS6000 machines? If so, Dennis is
right. No advantage either way.
On the other hand, sheer volumes would tell you that chipsets and other supporting
hardware for Pentium systems will be cheaper, and more competitive. In some cases,
certain components will be standard, like a couple chips on the PCI bus, but
any hardware that directly interfaces with the CPU has to be specific. From
what Intel, Apple and IBM seem to be claiming, it looks like Pentium systems
may outnumber PPC systems by 3-to-1 or more in 1993. Big difference.
Tony
tham...@pcocd2.intel.com
(I speak only for myself, not Intel)
Mel Martinez
Johnson) wrote:
>
>
> Which leaves you with "all" meaning Windows, WindowsNT, and OS/2. I
> had heard about NT, and OS/2 doesn't suprise me, but will Windows
> 3.1/4.0/Chicago be ported to PPC?
>
'Dunno about 4.0/Chicago, but Win3.1 definitely will be. Insignia (the
SoftPC) will be doing this port (I believe they have a source license now
so it should be pretty fast). More indirectly (and much slower) one could
run Win3.1 under the MacOS on the PPC using SoftPC for Windows.
> >
> >Probably right about that. PowerOpen, though, should allow one to run
> >multiple OS's simultaneously as processes within X that can be simply
> >switched between with a mouse-click. I don't believe you will see
> >PowerOpen ported to Pentium.
>
> The recent rumor mill says that this project died with Apple's recent
> lay-offs.
I had heard that Apple had down-sized their participation in PowerOpen, but
that it was not terminated by any means. Remember, PowerOpen is not just
an Apple project. How could they(Apple) kill it? I would expect that at
this point PowerOpen needs to be near completion (at least near beta) if it
is going to be running on IBM PPC machines. Also, the real key for Apple
(IMHO) is MAS (Macintosh Application Services) development. I had not
heard anything about this being affected at all by the layoffs, and this
week's MacWeek article about Apple shopping it's OS around to potential
clonemakers implies that selling the mac OS and thus MAS must be an
important part of their strategy.
But then when dealing with rumours, one really can't trust much...
Mel Martinez
doco...@sedona.intel.com (Dennis O'Connor) wrote:
>
>
> m...@pha.jhu.edu (Mel Martinez) writes:
> ] doco...@sedona.intel.com (Dennis O'Connor) wrote:
>
> ] Your comments about futurelow-power Pentium versions (the P54C) catching up
> ] may or may not be relevant. Since both it and the even lower power PPC603
> ] are being developed on similar schedules, I don't see at the moment that
> ] Intel will be catching up with it (in regards to the low=power consumption
> ] issue), but you may be right.
>
> Well, I haven't seen anyone say PPC603 would be shipping in volume
> in 1Q94, when P54C is supposed to be. And as I said : PPC603 can't
> exploit the power saving of the 5V-to-3V conversion, which P54C will.
> We'll see what happens.
>
All the rumours I hear point to PPC603 portables appearing in mid/late '94.
That implies volume shipments of chips would have to occur by the spring,
or late spring/early summer, wouldn't it? Sounds only a matter of a month
or two behind what you said the P54C schedule is.
I disagree a little with your presumption about the power savings. From
what you described, the P54C will only catch up with the PPC601 in power
consumption. The PPC603 is significantly less power-hungry than the 601.
Reduced power use comes not just due to lower voltage, but also smaller
transister gaps, the number of transisters and, of course, the layout of
the chip's actual components. I.E.- to minimize powerconsumption/waste
heat(Joule heating) you want to move a given amount of charge over as short
a distance (impedance) and as few junctions as possible. A chip design
should take into consideration the topological flow of charge given the
most frequent operations. Thus there are many avenues besides operational
voltage reduction that can be taken to reduce power consumption. I am
sure, of course that both Intel and Motorola are racing to exploit all of
them.
> Yep. And the intense competition makes things fun (I mean FUN) for
> chip designers, so I've got no complaints !
Actually, I think it will be most fun for the consumer! Powerful machines
at cheaper and cheaper costs! :-)
Steve Kanefsky
Dennis O'Connor <doco...@sedona.intel.com> wrote:
>
>kane...@halcyon.com (Steve Kanefsky) writes:
>] In article <DOCONNOR.93...@potato.sedona.intel.com>,
>] Dennis O'Connor <doco...@sedona.intel.com> wrote:
>] >
>] >Also consider that the price difference between a PPC and a Pentium
>] >isn't enough to pay for the re-purchase of even two or three major
>] >software packages. That your existing software will run at 486DX33
>] >speeds on a PPC601 isn't a persusive reason to buy a whole new system.
>]
>] You seem to be aware that quantity can be a big factor in pricing, since
>] you negate a lot of the price advantage of the PPC over the Pentium
>] because the PPC price is quantity-20K and the Pentium price is
>] quantity-1K. But guess what: there's a heck of a lot more difference
>] between quantity-1 pricing and quantity-1K pricing than between 1K and
>] 20K. How can you compare quantity-1 software prices with quantity-1K or
>] 20K differences in chip prices?
>
>Software prices are usually quoted Qty. 1, regardless of whether
>the software is for Pentium(tm) or PowerPC(tm) chips. However, the
>only prices in the public domain for Pentium chips is the Qty. 1K
>price, and the only price in the public for PC601 is the Qty. 20K
>price. Therefor, comparisons of these two chips prices, based on the
>public data, have to be fudged to compensate for the high-volume
>discount built-in to the PPC price but absent from the Pentium chip price.
Why don't you take the Qty. 1 price for a '486 chip in Computer Shopper,
divide it by the Qty. 1000 price, then multiply the difference in
quantity prices for PowerPC versus Pentium. That should give a hint at
the difference in price to the consumer. Maybe instead of taking the
Qty. 1 price for the CPU chip alone, you could take the difference
between otherwise equivalent 486DX 33MHz and 486DX2 66MHz systems, and
compare that to the difference in quantity price for the chip alone.
>Also, The average user buys his system from a manufacturer who buys
>processors in large volume. Therefor, the user cost of the processor is
>more affected by the high-volume price than the Qty. 1K.
I buy my software from a company that buys software in large volume too.
You simply can't compare full retail price of a software package to
the difference in price between CPUs in huge quantities.
>] Even worse, you seem to think that you'll have to purchase a brand new
>] copy of the software, rather than paying a small upgrade fee (which you
>] probably would have paid anyway to upgrade to the next version
>
>The "upgrade fee" for, say, AutoCAD, is pretty stiff. Also, has any
>software maker announced this cross-platform upgrade program ?
That's because there's a lot of work (probably on the order of several
man-years) involved in upgrading a complex software package like that.
Recompiling for the PPC, on the other hand, is more like a few
man-WEEKS.
>Discount upgrade programs exist to generate extra revenue : most
>users of V5.1 just plain _will_ _not_ pay full price for 5.2,
>so software makers sell it to them at only, say, 2X cost, instead
>of 10X, and add some to the bottom line. This is not the situation
>someone who moves from host to another is in : they really need
>a new version of the software (unless they are willing to tolerate
>the performance decrease of emulation, in which case, why switch
>hardware?), so they will pay full price.
Ah, but you don't really need the upgrade, and the vast majority of Mac
owners will be getting a performance boost even switching from native
680x0 code to emulated code on a PPC. Only Quadra 800/840av/950 owners will
actually see a drop in speed when running under emulation, and
Quadra 700 and Centris 650/660av owners should see about the same speed.
But the operating system will be running much faster, including the
Finder. Any new software will cost the same and run at native speeds,
and upgrades should not be expensive at all, except when they also
include major new features. For those applications that are simply
recompiled, I wouldn't be suprised to see free upgrades. I wouldn't
even be suprised to see updater applications available via anonymous ftp
for those software packages that are already updated that way.
Remember, it's in a company's best interest for their software to run
fast, and if it's really easy to do a recompile and triple the speed,
they're going to do it. It makes them look really good.
>IMHO, The big software vendors like having One Huge Market for their
>software, as opposed to lots os smaller markets, anyway. They
>aren't going to sacrifice _their_ profits to encourage diversity
>in the installed hardware base. For an example, look at Microsoft :
>they aren't willing to port WindowsNT, they are letting Motorola
>pay for that.
One huge PPC market is fine with me :-)
The 680x0 and PPC Mac market will effectively be one big market, because
one binary can run optimally on both types of machine by containing both
types of code resources. You can optionally install just one type of code
resource to save disk space, or include both for servers that are accessed
by both types of machine.
>] You're also ignoring the fact that Pentium owners will have to
>] upgrade their software as well to take full advantage of their machines,
>] otherwise you have to give the performance advantage to a recompiled PPC
>] app over a non-recompiled '486 app running on a Pentium.
>
>Fact1: Pentium chips runs regular 486 programs faster than even a DX2-66.
And considerably slower than a PPC 601 at the same clock speed.
>Fact2: Code recompiled with the Pentium chip inspired optimizations runs
> faster on a 486 than code from older compilers does
>Fact3: Programs compile for Pentium chips will run on 486s.
>Fact4: Because of Facts 2&3, many ISV (Independant Software Vendors) have
> announced that all future code will use the new optimizations.
This will at least mean that you don't have to maintain two distinct
sets of code, but it does mean that all code needs to be recompiled or
suffer a performance disadvantage. I've also heard that the tricks
needed to take advantage of the Pentium are proprietary, so there won't
be any compilers with available source like gcc that can take full advantage
of the Pentium. Is that correct?
>This is a different scenario from PPC 601, which does not run 486 code
>faster than a same-price 486, and whose code can not be run on a 486.
Yes, that's the only thing that makes the Pentium even worth talking
about. I predict it'll be about as successful as Philips' idea that
backwards compatability with the compact cassette was crucial to any new
recordable digital format. It appears that people are MUCH more willing
to throw away their old cassettes, or just listen to them on their old
cassette players than Philips though. They'd much rather have the
snazzy optical-disc technology and they're happy to dispense with
backwards compatability in the process.
>] Finally, you're only considering differences in the CPU itself. A brand
>] new Mac designed around the PPC is likely to take better advantage of
>] the PPC processor than a '486 with a Pentium dropped in.
>
>Brand new "PCs" are being designed for the Pentium chip, too.
>Plus, motherboard replacements are very common today. I don't
>see where _either_ chip has an advantage in this arena.
My point is that you have to add the cost difference of the CPU plus any
other cost differences on the motherboard, and that PCs for the most
part need much more upgrading on the motherboard than Macs do. Macs
already have fast SCSI, NuBus, ethernet, etc. on the motherboard. In
other words, PPC Macs will be *cheaper* than today's fast '040 Macs,
whereas a decently-designed Pentium PC will be much more expensive
than today's '486 PC. That can only translate into much higher market
share.
--
Steve Kanefsky
Tony Hamilton
|> In article <DOCONNOR.93...@potato.sedona.intel.com>,
|> doco...@sedona.intel.com (Dennis O'Connor) wrote:
|> > Well, I haven't seen anyone say PPC603 would be shipping in volume
|> > in 1Q94, when P54C is supposed to be. And as I said : PPC603 can't
|> > exploit the power saving of the 5V-to-3V conversion, which P54C will.
|> > We'll see what happens.
|> >
|>
|> All the rumours I hear point to PPC603 portables appearing in mid/late '94.
|> That implies volume shipments of chips would have to occur by the spring,
|> or late spring/early summer, wouldn't it? Sounds only a matter of a month
|> or two behind what you said the P54C schedule is.
Maybe you should rethink what you just said. Dennis is saying (is that what
the press is saying, Dennis?) that P54C would be shipping in volume in spring
of 1994. If PPC603 comes out in late 1994, that means it will be out in volume
in spring of _1995_. So, you have a year's difference here. But, are any of
these guesses right? Probably not. Who knows...
|> > Yep. And the intense competition makes things fun (I mean FUN) for
|> > chip designers, so I've got no complaints !
|>
|> Actually, I think it will be most fun for the consumer! Powerful machines
|> at cheaper and cheaper costs! :-)
Yes, but the consumers can also get the short end of the stick. WIth all the
competition, both sides will be pushing consumers in all kinds of ways, and
they're bound to get confused and buy things for all the wrong reasons. For
instance, _what if_ they buy PPC's, only to find the emulation severely lacking,
and software compatibility in the dumps? Or, _what if_ they buy a Pentium, only
to find their PPC-using friends running much faster, and getting along just find
with native software? Ultimately, the marketing folks will have their way one
way or another, not necessarily to the consumer's benefit...
Tony
Edward Galarza
doco...@sedona.intel.com (Dennis O'Connor) says:
>] Your comparisons were based on assumptions that the PPC 601 was a 3V chip.
>] My impression was that the first PPC chip the 601 would be a 5V chip and
>] the second the 603 would be the low power 3V chip. Can anyone who has the
>] MOTO spec sheets on the PPC confirm this? If the 601 is indeed a 5V chip
>] then I think the power dissipation numbers would definately favor the PPC.
>
>It would be nice if someone could independantly confirm, but my
>source for the 3V assumption is the Motorola PPC601 literature.
>Specifically, page 2 of the April 93 version of
>
> "Technical Summary, PowerPC(tm) 601 RISC Microprocessor"
> Copyright 1993 Motorola Inc. Order code MPC601/D
I have a copy of that paper. It says, "The MPC601 uses an advanced,
3.6-V CMOS process technology..."
>--
>Dennis O'Connor doco...@sedona.intel.com
>Intel i960(R) Microprocessor Division Solely responsible for what I do.
Edward Galarza/2 2.1
Brooklyn College
City University of New York
Le...@cunyvm.cuny.edu
Le...@cunyvm.bitnet
Graham K. Johnson
>Dennis O'Connor <doco...@sedona.intel.com> wrote:
>>someone who moves from host to another is in : they really need
>>a new version of the software (unless they are willing to tolerate
>>the performance decrease of emulation, in which case, why switch
>>hardware?), so they will pay full price.
>
>Ah, but you don't really need the upgrade, and the vast majority of Mac
>owners will be getting a performance boost even switching from native
>680x0 code to emulated code on a PPC. Only Quadra 800/840av/950 owners will
>actually see a drop in speed when running under emulation, and
>Quadra 700 and Centris 650/660av owners should see about the same speed.
By that logic, if they don't "need" the upgrade, they don't "need" the
PPC; they can do fine with the system they have or get a faster 680x0
Mac. If someone is going to buy a PPC Mac they are going to want some
benefit over their old Mac (it should at least be faster than a 680x0
Mac they could get for the same price).
>>] You're also ignoring the fact that Pentium owners will have to
>>] upgrade their software as well to take full advantage of their machines,
>>] otherwise you have to give the performance advantage to a recompiled PPC
>>] app over a non-recompiled '486 app running on a Pentium.
Quite irrelevant...
>>Fact1: Pentium chips runs regular 486 programs faster than even a DX2-66.
(an understatement, I believe)
>And considerably slower than a PPC 601 at the same clock speed.
I hadn't heard that the PPC 601 could run 486 programs at all, let
alone considerably faster than a 486DX2-66. (Honestly, I can't
imagine what you could have meant by this).
>>Fact2: Code recompiled with the Pentium chip inspired optimizations runs
>> faster on a 486 than code from older compilers does
>>Fact3: Programs compile for Pentium chips will run on 486s.
>>Fact4: Because of Facts 2&3, many ISV (Independant Software Vendors) have
>> announced that all future code will use the new optimizations.
>
>This will at least mean that you don't have to maintain two distinct
>sets of code, but it does mean that all code needs to be recompiled or
>suffer a performance disadvantage.
A performance disadvantage over what? The "Fact4" above means that
all code _will_ be recompiled; in the meantime (i.e. until you get an
upgrade because of new _features_ in the software, and it happens to
be optimized for Pentium) all programs will still run MUCH faster on
Pentium than on a 486. Hardly a disadvantage; in fact, it's the
Pentium's greatest asset.
>My point is that you have to add the cost difference of the CPU plus any
>other cost differences on the motherboard, and that PCs for the most
>part need much more upgrading on the motherboard than Macs do. Macs
>already have fast SCSI, NuBus, ethernet, etc. on the motherboard. In
So you will just be able to drop in a new PPC chip on you Mac
motherboard and everything else you can keep? That's not what I have
heard (i.e. that the PPC "upgrade" is an entirely new motherboard)
>other words, PPC Macs will be *cheaper* than today's fast '040 Macs,
Sure, the Centris-replacement Macs will be cheaper than the Quadra is
now. Why? For the same reason the Quadra is so much more expensive
than the Centris is now. Better CPU, yes, but also better everything
else.
Where, though, did you hear that a PPC Quadra-replacement Mac will be
less expensive than it's 680x0 counterpart?
Dennis O'Connor
kane...@halcyon.com (Steve Kanefsky) writes:
] >Discount upgrade programs exist to generate extra revenue : most
] >users of V5.1 just plain _will_ _not_ pay full price for 5.2,
] >so software makers sell it to them at only, say, 2X cost, instead
] >of 10X, and add some to the bottom line. This is not the situation
] >someone who moves from host to another is in : they really need
] >a new version of the software (unless they are willing to tolerate
] >the performance decrease of emulation, in which case, why switch
] >hardware?), so they will pay full price.
]
] Ah, but you don't really need the upgrade, and the vast majority of Mac
] owners will be getting a performance boost even switching from native
] 680x0 code to emulated code on a PPC.
I don't disagree with your statement aboput MAC users. But the (unwritten)
context of my statement was 486-software users : these people will, I think,
see a drop in the performance of emulated applications. One reason may be
that, from what I know of them, the 68K architecture is a lot easier to
emulate than the x86 is.
] >] You're also ignoring the fact that Pentium owners will have to
] >] upgrade their software as well to take full advantage of their machines,
] >] otherwise you have to give the performance advantage to a recompiled PPC
] >] app over a non-recompiled '486 app running on a Pentium.
] >
] >Fact1: Pentium chips runs regular 486 programs faster than even a DX2-66.
]
] And considerably slower than a PPC 601 at the same clock speed.
In terms of available chips, Intel's has faster integer SPECmarks
and IBM's has faster FP (about 50% faster). This is according to
the Motorola ad in EETimes. And that's for naticve code. A PPC601
emulating a 486 is going to be much slower than a Pentium chip.
] >Fact2: Code recompiled with the Pentium chip inspired optimizations runs
] > faster on a 486 than code from older compilers does
]
] >Fact3: Programs compile for Pentium chips will run on 486s.
] >Fact4: Because of Facts 2&3, many ISV (Independant Software Vendors) have
] > announced that all future code will use the new optimizations.
]
] This will at least mean that you don't have to maintain two distinct
] sets of code, but it does mean that all code needs to be recompiled or
] suffer a performance disadvantage. I've also heard that the tricks
] needed to take advantage of the Pentium are proprietary, so there won't
] be any compilers with available source like gcc that can take full advantage
] of the Pentium. Is that correct?
I think not. The "tricks" have been licensed to nearly every commercial
x86 compiler vendor, and most or all have announced they will include them
in their next releases. That's what I've heard in the press, anyway.
Because GNU ships source code, the NDA nature of the disclosure
prohibits GNU from using them.
] >This is a different scenario from PPC 601, which does not run 486 code
] >faster than a same-price 486, and whose code can not be run on a 486.
]
] Yes, that's the only thing that makes the Pentium even worth talking about.
It's the only thing that made it worth building, IMHO.
] I predict it'll be about as successful as Philips' idea that
] backwards compatability with the compact cassette was crucial to any new
] recordable digital format.
Very different situations. We'll just have to see.
Edward Galarza
doco...@sedona.intel.com (Dennis O'Connor) says:
>
>] In article <DOCONNOR.93...@potato.sedona.intel.com>,
>] Dennis O'Connor <doco...@sedona.intel.com> wrote:
>] >
>Software prices are usually quoted Qty. 1, regardless of whether
>the software is for Pentium(tm) or PowerPC(tm) chips. However, the
>only prices in the public domain for Pentium chips is the Qty. 1K
>price, and the only price in the public for PC601 is the Qty. 20K
>price. Therefor, comparisons of these two chips prices, based on the
>public data, have to be fudged to compensate for the high-volume
>discount built-in to the PPC price but absent from the Pentium chip price.
>
>Also, The average user buys his system from a manufacturer who buys
>processors in large volume. Therefor, the user cost of the processor is
>more affected by the high-volume price than the Qty. 1K.
>
>] Even worse, you seem to think that you'll have to purchase a brand new
>] copy of the software, rather than paying a small upgrade fee (which you
>] probably would have paid anyway to upgrade to the next version
>
>The "upgrade fee" for, say, AutoCAD, is pretty stiff. Also, has any
>software maker announced this cross-platform upgrade program ?
I believe WordPerfect will allow you to copy someone elses wordperfect
from another platform as long as you have a current license for another
platform *same version*. A DOS WordPerfect user could simply copy
someone elses Macintosh copy as long as they are the same version. Or
WordPerfect will offer the Macintosh copy and manuals for a fee to cover
media cost and hardcopy manuals. Nice policy BTW, the way it should be
IMHO.
>
>Discount upgrade programs exist to generate extra revenue : most
>users of V5.1 just plain _will_ _not_ pay full price for 5.2,
>so software makers sell it to them at only, say, 2X cost, instead
>of 10X, and add some to the bottom line. This is not the situation
>someone who moves from host to another is in : they really need
>a new version of the software (unless they are willing to tolerate
It makes more sense to think a software company will try to maintain
their existing customers by providing cross-upgrades to another platform.
Particularly if the competition provides them. They might not have
a liberal policy like copying software as WordPerfect does, but why
would they want to lose their existing customer base to the competition?
As you said, those customers will buy upgrades.
WordPerfect provides the cross upgrades, that DOS user now a Mac user will
upgrade his Mac WordPerfect he copied (legally) to the next greatest version
Mac version for the normal upgrade fee. WordPerfect maintains their customers
and upgrade revenue. If they do not offer an upgrade, they will lose that
customer upgrade to the competition.
Simple no? Provide cross-upgrade options, keep your loyal customers and
upgrade revenue. Not to mention giving your competition hell because
you now have a large loyal customer base on two or more platforms <g>
>the performance decrease of emulation, in which case, why switch
>hardware?), so they will pay full price.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Nope, naw, no-way, que no, hell no, huh-huh-huh-he-he-huh bogus dude :-)
>
>IMHO, The big software vendors like having One Huge Market for their
>software, as opposed to lots os smaller markets, anyway. They
>aren't going to sacrifice _their_ profits to encourage diversity
>in the installed hardware base. For an example, look at Microsoft :
>they aren't willing to port WindowsNT, they are letting Motorola
>pay for that.
[...]
If thats the case, why did Microsoft design NT on top of a microkernel?
Whats the deal with IBM and its portable WorkPlaceOS then?
Whats the deal with Sun and Solaris on 3 platforms?
>--
>Dennis O'Connor doco...@sedona.intel.com
>Intel i960(R) Microprocessor Division Solely responsible for what I do.
Edward Galarza/2 2.1
Edward Galarza
>
>In article <mem-1908...@durin.pha.jhu.edu> m...@pha.jhu.edu (Mel Martinez)
>writes:
>>In article <DOCONNOR.93...@potato.sedona.intel.com>,
>>doco...@sedona.intel.com (Dennis O'Connor) wrote:
>[blah blah blah power consumption blah blah deleted]
>>> OS:
>>> Right now, the Pentium processor can run either of three vendor's "DOS"s
>>> (MS-DOS*, PC-DOS*, DR-DOS*), any of four vendor-proprietary windowing
>>> OSs ( Windows, WindowsNT*, OS/2*, and NextStep486* ) and any of a large
>>> number of UNIX versions ( Interactive, ESIX, SCO Unix*, Solaris*, ... ).
>>
>>With the possible exception of NeXTStep and some of the miscelaneous unix
>>versions, all these will be running on the PPC as well.
>
>Which leaves you with "all" meaning Windows, WindowsNT, and OS/2. I
>had heard about NT, and OS/2 doesn't suprise me, but will Windows
>3.1/4.0/Chicago be ported to PPC?
IBM will have a Windows3.1 personality for the WorkplaceOS. Sun
will have WABI. Besides, who wants to run real windows on the PPC
if you can run a Windows personality or WABI as part of an even
more powerful operating system? BTW, IBM is also working on a DOS
personality for WpOS too. oh oh, almost forgot!, IBM is also
working on Win32 for OS/2 2.2. Seems natural the OS/2 personality
for WorkplaceOS would have the same functionally as OS/2 2.2, no?
>
>>> It's said that PPC will also run a bunch of OSs. I don't doubt it,
>>> but more than the Intel architecture does now ? I _do_ doubt that.
>>
>>Probably right about that. PowerOpen, though, should allow one to run
>>multiple OS's simultaneously as processes within X that can be simply
>>switched between with a mouse-click. I don't believe you will see
>>PowerOpen ported to Pentium.
>
>The recent rumor mill says that this project died with Apple's recent
>lay-offs.
>--
> ___
> /__ ___ "Uh, you're not gonna joh...@cps.msu.edu
><__/raham / eat that, are you?" Michigan State University
> |<. <_/ohnson --Fish Tales East Lansing, Michigan
Edward Galarza/2 2.1
Edward Galarza
Hamilton) says:
[...]
>with Motorola on this deal. Is _anyone_ else placing orders for PPC in those
>volumes?
Maybe Intel will! give or take a year or two...
(Sorry, couldn't resist hehehehe)
<+:-) (pope smiley)
[...]
cbs...@ccnga.waterloo.edu
>In article <mem-1908...@durin.pha.jhu.edu>, m...@pha.jhu.edu (Mel Martinez) writes:
>|> In article <DOCONNOR.93...@potato.sedona.intel.com>,
>|> doco...@sedona.intel.com (Dennis O'Connor) wrote:
>|> > Well, I haven't seen anyone say PPC603 would be shipping in volume
>|> > in 1Q94, when P54C is supposed to be. And as I said : PPC603 can't
>|> > exploit the power saving of the 5V-to-3V conversion, which P54C will.
>|> > We'll see what happens.
>|> >
>|>
>|> All the rumours I hear point to PPC603 portables appearing in mid/late '94.
>|> That implies volume shipments of chips would have to occur by the spring,
>|> or late spring/early summer, wouldn't it? Sounds only a matter of a month
>|> or two behind what you said the P54C schedule is.
>
>Maybe you should rethink what you just said. Dennis is saying (is that what
>the press is saying, Dennis?) that P54C would be shipping in volume in spring
>of 1994. If PPC603 comes out in late 1994, that means it will be out in volume
>in spring of _1995_. So, you have a year's difference here. But, are any of
>these guesses right? Probably not. Who knows...
Ok. Maybe Mel wasn't clear enough. Apple is planning to release the
PPC603 laptops either before or during the summer. The chips will have
been available before then. In fact, I imagine they will be available in
desktops before then (it takes quite a bit of time to create a
completely *new* motherboard for a laptop, whereas the desktop bus for
the PPC603 will basically be the same as the 601's). Based on Apple's
current shipping schedule, I'd say this means the PPC603's will be
available to 3rd parties by January '94. That could be wrong, but it
would suggest it.
I was talking to a TwinHead representative the other day, and he said
that TwinHead wasn't expecting any Pentium based laptops for at least
another 2 years. This probably translates to a reality of 1 year before
Intel introduces a chips that can do it, and a couple months later for
developers to get machines going that can use it. (I'm assuming that he
was being overly negative.)
>|> > Yep. And the intense competition makes things fun (I mean FUN) for
>|> > chip designers, so I've got no complaints !
>|>
>|> Actually, I think it will be most fun for the consumer! Powerful machines
>|> at cheaper and cheaper costs! :-)
>
>Yes, but the consumers can also get the short end of the stick. WIth all the
>competition, both sides will be pushing consumers in all kinds of ways, and
>they're bound to get confused and buy things for all the wrong reasons. For
>instance, _what if_ they buy PPC's, only to find the emulation severely lacking,
>and software compatibility in the dumps? Or, _what if_ they buy a Pentium, only
>to find their PPC-using friends running much faster, and getting along just find
>with native software? Ultimately, the marketing folks will have their way one
>way or another, not necessarily to the consumer's benefit...
>
>Tony
The computer industry is quite confusing for Joe Consumer, but with it's
increase size you can bet that people are going to find ways to make
them happy. I don't think anyone is going to get left out in the cold
anymore.
--Chris
--
Christopher Smith
cbs...@boomer.uwaterloo.ca
cbs...@undergrad.math.uwaterloo.ca
DC 3527
(519) 885-1211x3581
cbs...@ccnga.waterloo.edu
>In article <mem-1908...@durin.pha.jhu.edu> m...@pha.jhu.edu (Mel Martinez) writes:
>>In article <DOCONNOR.93...@potato.sedona.intel.com>,
>>doco...@sedona.intel.com (Dennis O'Connor) wrote:
>[blah blah blah power consumption blah blah deleted]
>>> OS:
>>> Right now, the Pentium processor can run either of three vendor's "DOS"s
>>> (MS-DOS*, PC-DOS*, DR-DOS*), any of four vendor-proprietary windowing
>>> OSs ( Windows, WindowsNT*, OS/2*, and NextStep486* ) and any of a large
>>> number of UNIX versions ( Interactive, ESIX, SCO Unix*, Solaris*, ... ).
>>
>>With the possible exception of NeXTStep and some of the miscelaneous unix
>>versions, all these will be running on the PPC as well.
>
>Which leaves you with "all" meaning Windows, WindowsNT, and OS/2. I
>had heard about NT, and OS/2 doesn't suprise me, but will Windows
>3.1/4.0/Chicago be ported to PPC?
Um... you forgot Solaris & AIX. I was under the impression that the
Insignia contract would be kept open through multiple revisions of MS
Windows, because they are the company contracted to do this with
WindowsNT. I doubt you'll see MS give up on NT anytime soon. It's their
solution to escaping the Intel bind.
>>> It's said that PPC will also run a bunch of OSs. I don't doubt it,
>>> but more than the Intel architecture does now ? I _do_ doubt that.
>>
>>Probably right about that. PowerOpen, though, should allow one to run
>>multiple OS's simultaneously as processes within X that can be simply
>>switched between with a mouse-click. I don't believe you will see
>>PowerOpen ported to Pentium.
>
>The recent rumor mill says that this project died with Apple's recent
>lay-offs.
>--
The project "died"? Gimmie a break. They would have announced it.
PowerOpen is a standard more than it is an OS. Apple's only
responsibility for it was providing support for MacOS on it.
cbs...@ccnga.waterloo.edu
>In article <DOCONNOR.93...@potato.sedona.intel.com>, doco...@sedona.intel.com (Dennis O'Connor) writes:
>
>|> Software prices are usually quoted Qty. 1, regardless of whether
>|> the software is for Pentium(tm) or PowerPC(tm) chips. However, the
>|> only prices in the public domain for Pentium chips is the Qty. 1K
>|> price, and the only price in the public for PC601 is the Qty. 20K
>|> price. Therefor, comparisons of these two chips prices, based on the
>|> public data, have to be fudged to compensate for the high-volume
>|> discount built-in to the PPC price but absent from the Pentium chip price.
>
>I wonder if it is also relevant to think about _who_ is buying these chips
>in 20K volumes - Apple and IBM, right? Well, for heaven's sake, they're _partners_
>with Motorola on this deal. Is _anyone_ else placing orders for PPC in those
>volumes? Is Ford yet (would be interesting to know, althought I imagine a
>subcontractor may do this for Ford - or do they do it themselves?)
>
>In addition, leaving aside all the shots in the dark at comparing actual cost
>to manufacture PPCs vs. Pentiums, the prices _surely_ reflect a discount intended
>as an incentive to get the PPC into the market. You can't exactly be charged
>for being anti-competitive either, because in this case you're trying to
>_break_ an apparent monopoly, not create one.
Well, you may consider them "incentives", but the PPC 603 chip is going
to drop the per-chip-price by almost 1/2 from what I here. Regardless of
whether they are incentives or not, the fact remains that the
"incentives" aren't going to go away anytime soon, so you can expect the
machines to be cheaper.
>There's always a point at which these factors level out.
>
>|> Also, The average user buys his system from a manufacturer who buys
>|> processors in large volume. Therefor, the user cost of the processor is
>|> more affected by the high-volume price than the Qty. 1K.
>
>This is very true, Dennis. Folks keep forgetting that the Pentium is out there
>now because of demand. Current pricing can't possibly reflect a situation where
>supply and demand are relatively in sync.
Wait, just a moment ago you said that nobody would be buying chips in
20K volumes. Now you are saying everyone buys them in 20K volumes.
"Current pricing" is providing Intel will a smaller per-chip profit than
486's at their rock bottom prices. Considering the 603's should be
getting down to the prices of 486's, this should be a concern.
>|> ] You're also ignoring the fact that Pentium owners will have to
>|> ] upgrade their software as well to take full advantage of their machines,
>|> ] otherwise you have to give the performance advantage to a recompiled PPC
>|> ] app over a non-recompiled '486 app running on a Pentium.
>|>
>|> Fact1: Pentium chips runs regular 486 programs faster than even a DX2-66.
>|> Fact2: Code recompiled with the Pentium chip inspired optimizations runs
>|> faster on a 486 than code from older compilers does
>|> Fact3: Programs compile for Pentium chips will run on 486s.
>|> Fact4: Because of Facts 2&3, many ISV (Independant Software Vendors) have
>|> announced that all future code will use the new optimizations.
>|>
>|> This is a different scenario from PPC 601, which does not run 486 code
>|> faster than a same-price 486, and whose code can not be run on a 486.
>
>Bravo, Dennis. Good point. Corporate buyers want _standardization_. Buying
>PPC's means offering and supporting new software (assuming recompiled software
>here), or suffering in performance. The loop simply isn't as tight as it is
>with the Pentium-486 situation Dennis describes.
The PowerPC means standardization. It means that you are never going to
wonder "can I run this software on this machine"? The answer is always
going to be yes, whether it's the $1500 system that you grabbed with
from a 3rd party dealer, or a $12,000 620 based fileserver from IBM.
Being able to have one chip technology across all platforms is nice.
Having the ability to use AIX software from your RS/6000's is nice too.
Being able to do all of this without reconfiguring your machine each
time is *really* nice. Standardizing is much more implemented with the
PPC, where all the OSes will have a common ground.
Everyone who bought a 386 purchased it because they wanted to use new
software which took advantage of the new features of the 386 (like
Windows software). They happily spent tons of money on software. The 386
wasn't even guarunteed to be fully compatible with their old software
and they still did it. What new features will be implemented in new
versions of applications? Things that are already out there but will
really work with a high powered chip: voice recognition,
video-on-demand, intelligent agents, 3D video, etc.
If people made the jump for Windows, why wouldn't they make the jump for
PPC? Historical evidence to suggest you are wrong.
>|> ] Finally, you're only considering differences in the CPU itself. A brand
>|> ] new Mac designed around the PPC is likely to take better advantage of
>|> ] the PPC processor than a '486 with a Pentium dropped in.
>|>
>|> Brand new "PCs" are being designed for the Pentium chip, too.
>|> Plus, motherboard replacements are very common today. I don't
>|> see where _either_ chip has an advantage in this arena.
>
>Except that many of the basic architectural considerations in a Pentium system
>are still similar to existing Intel architecture systems. On the other hand, does
>a PPC system architecture resemble existing RS6000 machines? If so, Dennis is
>right. No advantage either way.
It does resemble it. The "Power" architecture is something that you
already find in RS/6000's.
>On the other hand, sheer volumes would tell you that chipsets and other supporting
>hardware for Pentium systems will be cheaper, and more competitive. In some cases,
>certain components will be standard, like a couple chips on the PCI bus, but
>any hardware that directly interfaces with the CPU has to be specific. From
>what Intel, Apple and IBM seem to be claiming, it looks like Pentium systems
>may outnumber PPC systems by 3-to-1 or more in 1993. Big difference.
What will interface directly with the CPU? We have PCI which removes the
need for a local bus so now nothing *has* to work directly with the CPU.
What chips do Intel based machines share anyway? Aside from a couple of
chips for their bus controllers (and there are even a few different
flavours of these) most of their hardware varies from dealer to dealer,
and that which doesn't, is frequently just as easily used in PPC's.
Tim Olson
|
| qm...@casbah.acns.nwu.edu (Charlie Kuehmann) writes:
| ] Your comparisons were based on assumptions that the PPC 601 was a 3V chip.
| ] My impression was that the first PPC chip the 601 would be a 5V chip and
| ] the second the 603 would be the low power 3V chip. Can anyone who has the
| ] MOTO spec sheets on the PPC confirm this? If the 601 is indeed a 5V chip
| ] then I think the power dissipation numbers would definately favor the PPC.
|
| It would be nice if someone could independantly confirm, but my
| source for the 3V assumption is the Motorola PPC601 literature.
| Specifically, page 2 of the April 93 version of
|
| "Technical Summary, PowerPC(tm) 601 RISC Microprocessor"
| Copyright 1993 Motorola Inc. Order code MPC601/D
The 601 is a 3.6V part that can be used in an otherwise 5V system (i.e.
tolerates 5V I/O).
--
-- Tim Olson
Apple Computer Inc. / Somerset
Mark J. Lilback
>
>On the other hand, sheer volumes would tell you that chipsets and other supporting
>hardware for Pentium systems will be cheaper, and more competitive. In some cases,
>certain components will be standard, like a couple chips on the PCI bus, but
>any hardware that directly interfaces with the CPU has to be specific. From
>what Intel, Apple and IBM seem to be claiming, it looks like Pentium systems
>may outnumber PPC systems by 3-to-1 or more in 1993. Big difference.
>
chips can be produced in higher quantity than the Pentium. I've heard
Apple is planning on selling over 1 million PPC's next year, and I
imagine IBM has pretty high expectations. Is Intel going to be producing
6 million or more Pentium's next year?
Mark
mlil...@seas.gwu.edu
Tony Hamilton
>"Current pricing" is providing Intel will a smaller per-chip profit than
>486's at their rock bottom prices. Considering the 603's should be
>getting down to the prices of 486's, this should be a concern.
First, no-one really knows what the margins are. Second, process improvements
ramp up very quickly with these things. Even as Pentium prices go down, I'd
bet the margins will increase.
>If people made the jump for Windows, why wouldn't they make the jump for
>PPC? Historical evidence to suggest you are wrong.
Because Windows offered a major shift in interfacing with the computer. It
made the PC "Mac-like", easier to use. There is nothing of PPC that provides
any such advantage over other PC's. And, Windows is _cheap_. We're talking
software with WIndows, complete system replacement with PPC. How can you
compare those?
>What will interface directly with the CPU? We have PCI which removes the
>need for a local bus so now nothing *has* to work directly with the CPU.
Well, I'm no PCI expert, but I know there are at least 4 chips in a PCI
chipset, at least one of which interfaces with the CPU. At least, that is
how it was explained to me once.
Steve Kanefsky
Graham K. Johnson <joh...@cps.msu.edu> wrote:
>In article <250o1j$e...@nwfocus.wa.com> kane...@halcyon.com (Steve Kanefsky) writes:
>>Dennis O'Connor <doco...@sedona.intel.com> wrote:
>>>...
>>>someone who moves from host to another is in : they really need
>>>a new version of the software (unless they are willing to tolerate
>>>the performance decrease of emulation, in which case, why switch
>>>hardware?), so they will pay full price.
>>
>>Ah, but you don't really need the upgrade, and the vast majority of Mac
>>owners will be getting a performance boost even switching from native
>>680x0 code to emulated code on a PPC. Only Quadra 800/840av/950 owners will
>>actually see a drop in speed when running under emulation, and
>>Quadra 700 and Centris 650/660av owners should see about the same speed.
>
>By that logic, if they don't "need" the upgrade, they don't "need" the
>PPC; they can do fine with the system they have or get a faster 680x0
>Mac. If someone is going to buy a PPC Mac they are going to want some
>benefit over their old Mac (it should at least be faster than a 680x0
>Mac they could get for the same price).
They probably don't need the PPC to run their current software. What
they want is to run new software with new features that requires more
power than their current machines can provide. If the versions of the
software they're running didn't run adequately on the machines they had,
they probably never would have bought that machine or that software in
the first place. Since they want to upgrade their software and buy new
software anyway, getting the PPC version in the process comes for free.
The PPC machines will also come with a faster version of the OS and
better video, SCSI, networking, etc. than their current machines.
People spend a fair amount of time just using the OS, and the faster
supporting hardware will be of benefit even in emulated apps.
>>>] You're also ignoring the fact that Pentium owners will have to
>>>] upgrade their software as well to take full advantage of their machines,
>>>] otherwise you have to give the performance advantage to a recompiled PPC
>>>] app over a non-recompiled '486 app running on a Pentium.
>
>Quite irrelevant...
No. very relevant. There will probably be a better selection of native
PPC apps than recompiled Pentium apps. Thus you'll be able to buy a PPC
system for less than a Pentium system that runs a larger software base
faster than the Pentium.
>>>Fact1: Pentium chips runs regular 486 programs faster than even a DX2-66.
>
>(an understatement, I believe)
Have you ever seen Dennis make an understatement? I don't think it is
much of an understatement. '486 apps benefit significantly from
recompilation for the Pentium.
>>And considerably slower than a PPC 601 at the same clock speed.
>
>I hadn't heard that the PPC 601 could run 486 programs at all, let
Word, Excel, PageMaker, PhotoShop, FreeHand, Illustrator, etc. etc.
Just about all the major applications are available for the Mac and will
be available in native PPC form at the PPC's introduction. It's likely
that many of these apps won't be recompiled for the Pentium at that
time, so they will perform significantly faster on a relatively cheap
PPC machine than they do on a relatively expensive Pentium machine.
>>This will at least mean that you don't have to maintain two distinct
>>sets of code, but it does mean that all code needs to be recompiled or
>>suffer a performance disadvantage.
>
>A performance disadvantage over what? The "Fact4" above means that
>all code _will_ be recompiled; in the meantime (i.e. until you get an
>upgrade because of new _features_ in the software, and it happens to
>be optimized for Pentium) all programs will still run MUCH faster on
>Pentium than on a 486. Hardly a disadvantage; in fact, it's the
>Pentium's greatest asset.
You just restated what I said. All code will have to be recompiled,
so all '486 users will have to pay the same upgrade fees that
PPC users will have to pay to get native versions of their apps.
If they don't, their software won't perform as well as the native PPC
version does on a Mac.
>>My point is that you have to add the cost difference of the CPU plus any
>>other cost differences on the motherboard, and that PCs for the most
>>part need much more upgrading on the motherboard than Macs do. Macs
>>already have fast SCSI, NuBus, ethernet, etc. on the motherboard. In
>
>So you will just be able to drop in a new PPC chip on you Mac
>motherboard and everything else you can keep? That's not what I have
>heard (i.e. that the PPC "upgrade" is an entirely new motherboard)
No, that's not what I said. What I said is that if you try and predict
what the prices will be like, Apple doesn't have to add that much cost
their motherboard designs to support the PPC, because the current
motherboards already have high-performance components. A '486 system
will have to include many more expensive components to take full
advantage of the Pentium, so a decent Pentium system will have a bigger
cost differential compared to a '486 than a PPC Mac will have compared
to today's Macs.
>>other words, PPC Macs will be *cheaper* than today's fast '040 Macs,
>
>Sure, the Centris-replacement Macs will be cheaper than the Quadra is
>now. Why? For the same reason the Quadra is so much more expensive
>than the Centris is now. Better CPU, yes, but also better everything
>else.
>Where, though, did you hear that a PPC Quadra-replacement Mac will be
>less expensive than it's 680x0 counterpart?
Because there's nothing to make it more expensive. What is there in a
PPC Quadra that should make it cost more than today's Quadra? There's
clearly a lot of stuff in a decent Pentium system that costs more than a
'486 system -- the CPU for one thing. Of course, Apple might choose to
keep costs of high-end machines similar but to include lots of other
neat features that still make it a much better value than today's
Quadra, but that doesn't change the fact that Pentium systems will have
to be much more expensive than '486 systems just to keep up with the
CPU, and that there will be other lower-end Macs that are much cheaper
than today's Quadra and have as good or better performance than more
expensive Pentium systems.
--
Steve Kanefsky
Tom Barrett
>The 601 is a 3.6V part that can be used in an otherwise 5V system (i.e.
>tolerates 5V I/O).
That's a strange voltage... do you mean 3.3+/-0.3 or 3.6+/-0.x? Can
it tolerate typical CMOS over/undershoot (which for 5v can go as high
as 8v in system with a really bad hair day)? This is probably in the
databook... I ordered one when they first came out, but they sent me
the programmer's ref instead :-)
--
|Tom Barrett, Sr. Engr., SysArch Research | 1-800-538-8450, ext. 56856 |
|AMD PCD MS-522|5900 E. Ben White|Austin, TX 78741|tom.b...@amd.com|
|AMD's VESA Local Bus and InfraRed Data Association Guru... __|
|My views are my own and may not be the same as the company of origin\/|
Tom Barrett
They might _want_ standardization (as in CPUs) because that's all
they think could be available, but what they _need_ is just
interoperability between platforms. That means that the Microsoft
Word Document from my UNIX station can be used on my PC and can be
used on my tech writer's MAC. The same file format and a seemless
method of file transfer... it's just a better way of looking at
standardization without focusing on specific parts of the computer
system. We have this somewhat today (with Word), but the furthering
of that concept is what Microsoft NT, ...at the office, and ...at home
is all about (so saith PC Week, EETimes, etc.).
Tom
Sean McDowell
If I remember correctly, Apple put out a press release (I read it
on AppleLink) explicitly denying this rumour and stating that the
project was running on time.
Sean
Mel Martinez
Hamilton) wrote:
> In article <mem-1908...@durin.pha.jhu.edu>, m...@pha.jhu.edu (Mel Martinez) writes:
> |> In article <DOCONNOR.93...@potato.sedona.intel.com>,
> |> doco...@sedona.intel.com (Dennis O'Connor) wrote:
> |> > Well, I haven't seen anyone say PPC603 would be shipping in volume
> |> > in 1Q94, when P54C is supposed to be. And as I said : PPC603 can't
> |> > exploit the power saving of the 5V-to-3V conversion, which P54C will.
> |> > We'll see what happens.
> |>
> |> All the rumours I hear point to PPC603 portables appearing in mid/late '94.
> |> That implies volume shipments of chips would have to occur by the spring,
> |> or late spring/early summer, wouldn't it? Sounds only a matter of a month
> |> or two behind what you said the P54C schedule is.
> Maybe you should rethink what you just said. Dennis is saying (is that what
> the press is saying, Dennis?) that P54C would be shipping in volume in spring
> of 1994. If PPC603 comes out in late 1994, that means it will be out in volume
> in spring of _1995_. So, you have a year's difference here. But, are any of
> these guesses right? Probably not. Who knows...
>
Err...maybe you should re-read what I just wrote: "PPC603 *portables*
appearing in mid/late '94." That is exactly what the rumours on the nets
and in the Mac magazines keep saying. And as I said, that implies a chip
delivery schedule MUCH earlier than what you just said. Do you have info
to indicate that the PPC603 *chip* will not arrive before late '94? It is
certainly not what I said.
Steve Kanefsky
Tony Hamilton <tham...@fnugget.intel.com> wrote:
>cbs...@ccnga.waterloo.edu writes:
>
>>If people made the jump for Windows, why wouldn't they make the jump for
>>PPC? Historical evidence to suggest you are wrong.
>
>Because Windows offered a major shift in interfacing with the computer. It
>made the PC "Mac-like", easier to use. There is nothing of PPC that provides
>any such advantage over other PC's. And, Windows is _cheap_. We're talking
>software with WIndows, complete system replacement with PPC. How can you
>compare those?
Windows makes the PC harder to use, IMHO. I actually prefer DOS when
I'm using a PC for many things. Windows may be cheap, but Word and
Excel aren't. A handful of new software for Windows costs more than a
new 486DX2-66 system.
The PPC *will* allow a major shift in interfacing with the computer,
much as the DSP chip in the new av macs does. We now have video in/out
for videoconferencing and speech recognition. Do you think we'd have
today's software if we were all still running on 8086 CPUs?
--
Steve Kanefsky
Mel Martinez
Hamilton) wrote:
>
> cbs...@ccnga.waterloo.edu writes:
> >If people made the jump for Windows, why wouldn't they make the jump for
> >PPC? Historical evidence to suggest you are wrong.
>
> Because Windows offered a major shift in interfacing with the computer. It
> made the PC "Mac-like", easier to use. There is nothing of PPC that provides
> any such advantage over other PC's. And, Windows is _cheap_. We're talking
> software with WIndows, complete system replacement with PPC. How can you
> compare those?
>
WAITAMINUTE!!!!! You are saying that the change to Windows use from DOS
use was CHEAP???? I am sorry, it was cheaper than moving from a DOS
system to a Mac system but it was not CHEAP! In the last 2-3 years nearly
EVERYONE who moved from principly DOS use to principly Windows use had to
get a NEW machine!!!!
In fact this parallels the change to PPC systems a lot in that moving to
Windows system meant:
1) buying/accelerating to a new system (from 286/386 to 386/486)
2) that system would still run your existing software(DOS) but
3) that old software would generally run kinda pokey (when run in a DOS
shell under Windows).
Sounds like the same phenomena one will encounter when moving to PPC
systems with the only difference being that you could always quit Windows
entirely and get a speed boost on DOS apps.
Don't try and say that the move to Windows was only the cost of the Windows
shell. It just doesn't wash.
Whether people will make a similar jump to PPC systems, I don't know. We
will see.
Dennis O'Connor
cbs...@ccnga.waterloo.edu writes:
] >Maybe you should rethink what you just said. Dennis is saying (is that what
] >the press is saying, Dennis?) that P54C would be shipping in volume in spring
] >of 1994. If PPC603 comes out in late 1994, that means it will be out in
] >volume in spring of _1995_. So, you have a year's difference here. But,
] >are any of these guesses right? Probably not. Who knows...
]
] Ok. Maybe Mel wasn't clear enough. Apple is planning to release the
] PPC603 laptops either before or during the summer. The chips will have
] been available before then.
Apple only needs engineering samples early. Given engineering samples
( a few hundred chips : lab-fab quantities ) you can design, test and
start to manufacture your system, warehouse them until the processors
are available , then just pop in the processor and ship. Now, this is
_not_ the smartest way to do things, but it shows that volume production
of chips doesn't have to preceed volume production of systems by more
than a low number of weeks.
] In fact, I imagine they will be available in
] desktops before then (it takes quite a bit of time to create a
] completely *new* motherboard for a laptop, whereas the desktop bus for
] the PPC603 will basically be the same as the 601's).
But if 601 and 603 have similar busses, I can design a prototype
laptop using the 601 and an AC adapter, right ? So there's no
reason introduction of a laptop should lag the desktop.
] I was talking to a TwinHead representative the other day, and he said
] that TwinHead wasn't expecting any Pentium based laptops for at least
] another 2 years. [ ... speculation of why ... ]
There are several issues here : technical issues of getting the
power down, and market issues. Is there really a strong demand for
that much power in laptops ? Can the system maker make more money
using the Pentium chips somewhere else ? Remember that right now,
Pentium processor based systems are being marketted as servers
and hi-end workstations. How much power do you need to edit a
file (or play solitaire) while on a plane ? So it's hard
to say when and why Pentium chip based laptops will appear.
Dennis O'Connor
kane...@halcyon.com (Steve Kanefsky) writes:
] >>>Fact1: Pentium chips runs regular 486 programs faster than even a DX2-66.
] >
] >(an understatement, I believe)
]
] Have you ever seen Dennis make an understatement? I don't think it is
] much of an understatement. '486 apps benefit significantly from
] recompilation for the Pentium.
That was Fact2. Pentium chips are _still_ significantly faster than
even Intel 486DX2-66/33 chips : what with the higher-bandwidth bus
(4 times as fast), larger caches, and superscalar execution, how
could it not be ?
] >>And considerably slower than a PPC 601 at the same clock speed.
] >
] >I hadn't heard that the PPC 601 could run 486 programs at all, let
]
] Word, Excel, PageMaker, PhotoShop, FreeHand, Illustrator, etc. etc.
] Just about all the major applications are available for the Mac and will
] be available in native PPC form at the PPC's introduction.
You're missing the point. By "486 software" we mean existing already-bought
software compiled for the Intel Architecture. The stuff people already
have thousands of dollars worth of already on their current computer.
] >A performance disadvantage over what? The "Fact4" above means that
] >all code _will_ be recompiled; in the meantime (i.e. until you get an
] >upgrade because of new _features_ in the software, and it happens to
] >be optimized for Pentium) all programs will still run MUCH faster on
] >Pentium than on a 486. Hardly a disadvantage; in fact, it's the
] >Pentium's greatest asset.
]
] You just restated what I said. All code will have to be recompiled,
^^^^
Adding this one word makes a _big_ difference. If Word and Excel are
already fast enough or even nearly so, buying a Pentium processor
based system will make it even faster, and I don't need to upgrade.
So you are _not_ saying the same thing.
Tony Hamilton
|> In article <thamilto.745828481@fnugget>,
|> Tony Hamilton <tham...@fnugget.intel.com> wrote:
|> >cbs...@ccnga.waterloo.edu writes:
|> >
|> >>If people made the jump for Windows, why wouldn't they make the jump for
|> >>PPC? Historical evidence to suggest you are wrong.
|> >
|> >Because Windows offered a major shift in interfacing with the computer. It
|> >made the PC "Mac-like", easier to use. There is nothing of PPC that provides
|> >any such advantage over other PC's. And, Windows is _cheap_. We're talking
|> >software with WIndows, complete system replacement with PPC. How can you
|> >compare those?
|>
|> Windows makes the PC harder to use, IMHO. I actually prefer DOS when
|> I'm using a PC for many things. Windows may be cheap, but Word and
|> Excel aren't. A handful of new software for Windows costs more than a
|> new 486DX2-66 system.
Steve, if you haven't noticed, those same software titles cost no more than
the DOS versions. PC software has become extremely complex and expensive, but
it is _not_ because of Windows.
|> The PPC *will* allow a major shift in interfacing with the computer,
|> much as the DSP chip in the new av macs does. We now have video in/out
|> for videoconferencing and speech recognition. Do you think we'd have
|> today's software if we were all still running on 8086 CPUs?
Nope, but by the time PPC is out in any kind of volumes, all us 486 users will
have all that and more. In fact, I'll wager that more people are using their
486's for those purposes than people on other platforms _before_ PPC makes
any kind of impact. There will simply be _no_ innovations in PPC. It is just
a _different_ architecture that _currently_ appears to be a better performer.
LIke I keep saying, let's look at the whole thing again this time next year.
We'll know by then what's going to happen.
Tony Hamilton
|> In article <250n6b$8...@ornews.intel.com> tham...@fnugget.intel.com (Tony Hamilton) writes:
|> >Corporate buyers want _standardization_.
|>
|> They might _want_ standardization (as in CPUs) because that's all
|> they think could be available, but what they _need_ is just
|> interoperability between platforms. That means that the Microsoft
|> Word Document from my UNIX station can be used on my PC and can be
|> used on my tech writer's MAC. The same file format and a seemless
|> method of file transfer... it's just a better way of looking at
|> standardization without focusing on specific parts of the computer
|> system. We have this somewhat today (with Word), but the furthering
|> of that concept is what Microsoft NT, ...at the office, and ...at home
|> is all about (so saith PC Week, EETimes, etc.).
I am sorry, but I'm going to have to disagree with you on this one, Tom.
Platform standardization, not just interoperability, is very important. Sure,
the latter is probably the first prority, but it doesn't solve everthing. Set
aside standard OS for a moment (like NT), and assume you have compatible
applications. You still have to support different OS's now. Okay, now maybe its
late 1994 or 1995, and _some_ OS is now supported well enough to use on all
your hardware platforms. That is much better, but you _still_ have to support
the different types of hardware. Now, whether your support for systems and
software is internal, or you contract out, more platforms costs more money.
So, the CIS/IT managers out there are not _only_ looking as OS standardization,
(I think this is where they are starting to look. Most of us have some degree
of interoperability already), but they are concerned about weeding out all the
different hardware platforms to support. It just makes good business sense.
The only caveat I see is that, if the PCI bus becomes a major standard (and
I think, Tom, we both hope it does :-), then it might alleviate much of this
problem, as we'd be able to swap peripherals between PC's and PPC's. If that
happens, and _if_ NT becomes _the_ standard (I have some doubts about this,
although I'd like to see it happen), and a few other things happen, then I
think the PPC has a great future. I just think there are a lot of if's. That is
the problem with all other efforts at breaking into the market. There was
always something missing. Perhaps not this time. Hard to tell yet.
Dennis O'Connor
tdb...@dvorak.amd.com (Tom Barrett) writes:
] In article <> tham...@fnugget.intel.com (Tony Hamilton) writes:
] >Corporate buyers want _standardization_.
]
] They might _want_ standardization (as in CPUs) because that's all
] they think could be available, but what they _need_ is just
] interoperability between platforms. That means that the Microsoft
] Word Document from my UNIX station can be used on my PC and can be
] used on my tech writer's MAC.
Tom, if you talk to most people who administer a large corporate network,
you'll find that what _they_ want is cookie-cutter standardization :
no more than maybe three different configurations (mem size, peripherals,
etc.) and all of them running the exact same operationg system on
the exact same processor architecture.
Perhaps it's just a symptom of the immaturity of the market, but
apparently getting bugs fixed and upgrades to the OS installed
is hassle enough without having to do it three or four times
because you've got three or four processor architectures.
Steve Kanefsky
>] In article <> tham...@fnugget.intel.com (Tony Hamilton) writes:
>] >Corporate buyers want _standardization_.
>]
>] They might _want_ standardization (as in CPUs) because that's all
>] they think could be available, but what they _need_ is just
>] interoperability between platforms. That means that the Microsoft
>] Word Document from my UNIX station can be used on my PC and can be
>] used on my tech writer's MAC.
>
>Tom, if you talk to most people who administer a large corporate network,
>you'll find that what _they_ want is cookie-cutter standardization :
>no more than maybe three different configurations (mem size, peripherals,
>etc.) and all of them running the exact same operationg system on
>the exact same processor architecture.
That's only true when different machines are viewed as "equivalent" in
some sense. It's certainly true that a large coproration would prefer
to standardize on PS/2s instead of a mish-mash of Dells, Gateways,and
no-name clones, although it's really the operating system and software
that matter the most. Standardizing on PS/2's running a mish-mash of
DOS, Windows, NT, and NeXTstep would be worse than choosing entirely
different hardware to run the same operating system.
It's true to a much smaller extent with Macs and PCs, because Macs are
often viewed as a separate kind of machine. It's even less true with
PCs versus workstations or mainframes. The company I'm working at right
now has entirely separate budgets for UNIX workstations and PCs. The
Macs and PCs come from the same budget, and the company does try to
standardize on PCs once in a while, but the employees simply demand Macs
in many cases so it never works, and there are literally thousands of
Macs around here in addition to the thousands of PCs.
--
Steve Kanefsky
Tony Hamilton
|> WAITAMINUTE!!!!! You are saying that the change to Windows use from DOS
|> use was CHEAP???? I am sorry, it was cheaper than moving from a DOS
|> system to a Mac system but it was not CHEAP! In the last 2-3 years nearly
|> EVERYONE who moved from principly DOS use to principly Windows use had to
|> get a NEW machine!!!!
This is similar to another reply I just made. Again, as I stated there, this
would have happened anyway. It is not Windows' fault...
|> 1) buying/accelerating to a new system (from 286/386 to 386/486)
Again, they would'a done this anyway. I personally never liked WIndows, because
the interface was slow. But now that the systems can run it at light speed,
I am glad the industry made the transition. We would have never gotten this
far if it wasn't for the first steps taken.
|> 2) that system would still run your existing software(DOS) but
|> 3) that old software would generally run kinda pokey (when run in a DOS
|> shell under Windows).
Not true at all. I find most 286 applications run faster in a 386 DOS window
than under DOS on a 286. Besides, all this movement happened _because_ we
needed to run new, more powerful software. I certainly don't run any old,
pokey software on my system.
|> Don't try and say that the move to Windows was only the cost of the Windows
|> shell. It just doesn't wash.
Sorry, but it washes well with me. Windows is just an interface to run software
under. _All_ hardware purchases either are, or should be, motivated by the
software you need to run. The DOS world _needed_ better software. If you think
we were getting along fine running one app at a time on 286's, you're mistaken.
For most people, PC's weren't anything more than really good word processors.
So, more powerful software was needed. That software would have required better
systems and upgrades regardless of Windows.
Tony Hamilton
David Hafken
>> Which leaves you with "all" meaning Windows, WindowsNT, and OS/2. I
>> had heard about NT, and OS/2 doesn't suprise me, but will Windows
>> 3.1/4.0/Chicago be ported to PPC?
>>
>SoftPC) will be doing this port (I believe they have a source license now
>so it should be pretty fast). More indirectly (and much slower) one could
>run Win3.1 under the MacOS on the PPC using SoftPC for Windows.
This is not correct. Windows 3.1 will be *emulated* not *ported* to the
PowerPC by Insignia, in much the same way Windows 3.1 is *emulated* on current
Macs by SoftPC. Currently this emulation is of a 286. On the PowerPC, the
rumor is performance should be at "near 486 speeds," which may or may not
emulate the actual 486 chip (let's hope it still isn't 286 emulation!)
The performance gain is partly from, as you say, the source license that
Microsoft gave Insignia (which means no more reverse engineering), and also
due to the speed of PPC.
Insignia may very well do an emulation of Windows 4.0, but Microsoft will never
"port" Windows 3.1 and Windows 4.0 (Chicago) to the PowerPC.
Windows NT, however, *is* being ported by Motorola (and in a separate effort,
by IBM -- nice indication on how the alliance is going, eh?), which in itself
will be able to run Windows 3.1 and DOS apps.
-Dave
Gary Snow
>
>qm...@casbah.acns.nwu.edu (Charlie Kuehmann) writes:
>] > The following is indeed purely MY opinion, and I haven't even
>] > checked the official Intel PowerPC* position to write it.
>] >
>] > Okay : Pentium(tm) processor versus PowerPC* 601 processor :
>] >
>] [much comparison's deleted]
>]
>] Your comparisons were based on assumptions that the PPC 601 was a 3V chip.
>] My impression was that the first PPC chip the 601 would be a 5V chip and
>] the second the 603 would be the low power 3V chip. Can anyone who has the
>] MOTO spec sheets on the PPC confirm this? If the 601 is indeed a 5V chip
>] then I think the power dissipation numbers would definately favor the PPC.
>
>It would be nice if someone could independantly confirm, but my
>source for the 3V assumption is the Motorola PPC601 literature.
>Specifically, page 2 of the April 93 version of
Your 3 volt (3.3 or 3.6 actually, can't remember which) is correct.
Gary
--
-----
Gary Snow
uunet!clark!gsnow or gs...@clark.edu
Brian Coomes
tham...@fnugget.intel.com writes:
> Because Windows offered a major shift in interfacing with the computer.
It
> made the PC "Mac-like", easier to use. There is nothing of PPC that
provides
> any such advantage over other PC's. And, Windows is _cheap_. We're
talking
> software with WIndows, complete system replacement with PPC. How can you
> compare those?
Microsoft and the PC industrial complex (with some help from Apple
perhaps) seem
to have many believing two of the things you say:
1. "[Windows] made the PC 'Mac-like'...", and
2. "...Windows is _cheap_" followed by the implication that going from
DOS to
Windows is only a matter of buying a software package.
There is no doubt that Windows made the PC easier to use, but IMHO (and in
Microsoft's legal opinion, wink, wink) far from "Mac-like."
I have a Mac II, a Sun IPC, and an IBM PS/2 55SX (with Windows 3.1
installed) in
my office and that is the order in which the machines are used. As one
might
guess, I no longer use Windows on my DOS box. I cannot tolerate how
poorly it
performs and I would need new hardware if I wanted to run Windows.
In the context of having a machine with at least a 25MHz 486SX CPU and
some
graphics acceleration, Windows is merely a matter of software, but in my
situation it is not.
By the way, MacOS system 7.0.1 runs quite well on my ancient Mac II.
B. Coomes, Dept. of Math. and Comp. Sci., Univ. of Miami
Mark Eaton
Hafken) wrote:
>
> Insignia may very well do an emulation of Windows 4.0, but Microsoft will never
> "port" Windows 3.1 and Windows 4.0 (Chicago) to the PowerPC.
> Windows NT, however, *is* being ported by Motorola (and in a separate effort,
> by IBM -- nice indication on how the alliance is going, eh?), which in itself
> will be able to run Windows 3.1 and DOS apps.
I hadn't heard about any IBM "port" of Win NT. IBM doesn't even have NT
code, and Micrsoft stated they'll never share NT code with IBM. Therefore,
how could IBM port it?
>
>
> -Dave
''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''
Mark Eaton '
uc52...@mizzou1.missouri.edu ' They don't allow opinions at Mizzou
'
John R. Mashey
|> I hadn't heard about any IBM "port" of Win NT. IBM doesn't even have NT
|> code, and Micrsoft stated they'll never share NT code with IBM. Therefore,
|> how could IBM port it?
1) One piece of IBM has NT source and is porting to PPC.
Rumor has it that the memory management is "exciting", given PPC's
use of a fundamentally different sort of MMU than {X86, MIPS, Alpha}.
2) In some of these discussions, people seem to assume that IBM is one
unified entity, whereas this is clearly not so. Among other things, there is:
a) Boca Raton - buys X86s, makes its own X86s, builds PCs, sells them
to run DOS, Windows, OS/2, AIX, (and NT or other UNIXes I guess).
b) Austin - designs RS/6000s, codesigns PPCs with Motorola, will
buy PPCs from Moto (and can make them; current 601s are actually
manufactured in IBM Burlington, VT fab).
OS: AIX (with whatever influence from Apple)
(If you haven't noticed, these groups compete with each other,
even saying so publicly...)
3) "IBM is porting NT to PPC" is interpreted by some people to mean that
this is a major direction for IBM. Maybe it is. On the other hand, anyone
who has followed the computer industry the last few years might think twice.
Some world-class good people are doing the port .... but let's think
what it takes in the PC business:
Chips, systems, BIOS/HAL, operating system (and maintenance), ISV support,
to get applications, distribution channels:
Chips OS ISVsupp distribution when buy
X86 Intel & others Microsoft many many now
MIPS NEC, Toshiba,etc Microsoft some various now
Alpha DEC DEC DEC DEC soon
PPC IBM/Moto IBM ?? ?? 1995?
'94 seems hard
SPARC TI? Intergraph IG IG 1995?
And of course, if you expect anything like the existing PC market, you've
got to have multiple suppliers and cost structures of the PC-market ilk.
DEC is clearly betting a big chunk, and investing serious $$, although
it will be interesting to see how the realistic cost structures work out.
They at least also have a PC business that hasn't spent huge $$$ on OS/.2.
The MIPS-based machines have the disadvantage of not having a big
U.S. produce making them [although some other parts of the world are OK],
but have the advantages of meshing well into PC-like environments and
cost structures (chipsets that convert MIPS -> X86 interface so std
PC chipsets usable, multiple vendors for CPus, boards, HALs, etc, etc),
and of course has the unique-for-a-RISC advantage of being one of the
2 archiectures Microsfot develops NT on and ships....
Very unclear what will go on with the recently-announced Intergraph
effort, given that it needs a SPARC that happens to be a high-end
chip, not expected out in systems until 1995.
To actually get product to market, with all that is needed, one can either
use an existing organization, or one can create a new one. Which existing
IBM organization seems likely to build PPC-based products that fit well into
the PC world (the way the MIPS boxes do, i.e., using typical PC support
chips, peripherals, etc), that will take on porting each new NT release, and
supporting those changes, selling thru their channels, etc....
a) (Boca Raton) seems rather unlikely....
b) (Austin) seems more likely, although there are still some serious
conflicts, aren't there?
As it happens, a curious fact is that the PPC NT port is being done by
a research group within IBM, so it will have to get transferred later to
somebody set up to do releases, continuing support, etc, etc, which
is why I'd expect it to be 1995 before much could get shipped.
of course, it may be that this is really being done with the hope that
some external PC companies will pick it up and run with it.
Clearly, Motorola would like to have some more volume customers for PPC.
Anyway: there *is* a PPC port being done in IBM, and by good people;
anyone who thinks this is *sufficient* to make PPC NT a big force
should be ready to tell us which
organization is going to build such systems, support them, support the OS port, get ISVs, etc.
-john mashey DISCLAIMER: <generic disclaimer, I speak for me only, etc>
UUCP: ma...@sgi.com
DDD: 415-390-3090 FAX: 415-967-8496
USPS: Silicon Graphics 6L-005, 2011 N. Shoreline Blvd, Mountain View, CA 94039-7311
Dennis O'Connor
ma...@mash.wpd.sgi.com (John R. Mashey) writes:
] [...] but let's think what it takes in the PC business:
which really should include Macs, I think. Questions/corrections/additions:
]
] Chips, systems, BIOS/HAL, operating system (and maintenance), ISV support,
] to get applications, distribution channels:
]
] Chips OS ISVsupp distribution when buy
-----------------------------------------------------------------------
] X86 Intel & others Microsoft many many now
^^^
and IBM, Novell, etc.
] MIPS NEC,Toshiba,etc Microsoft some various now
] Alpha DEC DEC DEC DEC soon
^^^^
I thought this was "now"?
] PPC IBM/Moto IBM ?? ?? 1995?
] '94 seems hard
^^^ ^^^^^^^
and Apple even for 601-based "Macs"?
Tim Olson
| In article <251ddv$4...@apple.com> t...@apple.com writes:
| >The 601 is a 3.6V part that can be used in an otherwise 5V system (i.e.
| >tolerates 5V I/O).
|
| That's a strange voltage... do you mean 3.3+/-0.3 or 3.6+/-0.x? Can
| it tolerate typical CMOS over/undershoot (which for 5v can go as high
| as 8v in system with a really bad hair day)? This is probably in the
| databook... I ordered one when they first came out, but they sent me
| the programmer's ref instead :-)
3.6 +/- 5%. Yes, it is CMOS level compatible as well.
John R. Mashey
|>
|> ma...@mash.wpd.sgi.com (John R. Mashey) writes:
|> ] [...] but let's think what it takes in the PC business:
|>
|> which really should include Macs, I think. Questions/corrections/additions:
part of this, which I thought was clear from the context, but obviously
wasn't.
|> ] Chips OS ISVsupp distribution when buy
|> -----------------------------------------------------------------------
|> ] Alpha DEC DEC DEC DEC soon
|> ^^^^
|> I thought this was "now"?
I thought it was still "soon", but if somebody from DEC who knows posts that the equivalent of what M/S is shipping is shipping for Alpha, then OK.
|> ] PPC IBM/Moto IBM ?? ?? 1995?
|> ] '94 seems hard
|> ^^^ ^^^^^^^
|> and Apple even for 601-based "Macs"?
Again, this was addressing NT; it is clear that Apple *must* ship 601-based
Macs in 1994. That does lead to an interesting set of questions:
Maybe somebody *knows* the answers to the following, all of which I
think is likely to be public info or at least for which exist
reasonable analyst estimates (but info I don't know offhand).
a) What were Apple's unit volumes in 1992 and 1H93?
b) Especially in 2Q93, what was split between 68030 & 68040 products?
c) What do 68030s & 68040s cost in high volumes? (Apple must pay less).
The obvious reason for asking is to figure out whether the oft-quoted
1M PPCs in 1994 is: a) Plausible, or b) Marketing overhang.
Suppose one guesses that Apple would sell 5M units in 1994,
and that 1M PPCs would thus be 20% of their volume.
One knows that Apple volumes are very price-sensitive,
that margins aren't what they used to be, and that microprocessors
must be very cheap to be used in their highest volume products.
There is no way PPC 601s can be manufactured for anything
close to 68030 costs, hence 68030s will stay for a long time.
(Without knowing the exact numbers, one has to guess that
the cost difference to Apple, 1H94, between 68030 and PPC601,
must surely be $200-300, at least, and one would figure
end user list price differences of $500 or so.
68040s are less clear; they also will get cheaper; maybe the 68LC040's
get cheap enough to displace some of the 68030s, but 68040 die
size is pretty big, and it can't get too cheap.
Anyway, the
real interesting question will be to see how Apple positions PPC
machines, with 2 obvious extreme choices:
a) Keep them mostly in high-end, getting some nice margin back,
and would be the "natural" line (i.e., put newest thing
in high end, then work down over time),
but no way getting 1M units.
OR
b) Push them fairly far down, wiping out most 68040 products,
and pricing some models close enough to top 68030 models
to get some margin, and encourage people to pay a bit more for
a bunch more performance. Maybe this way gets 1M units, at
the cost of some margin hit for a while.
Anyway, it will be interesting to see what they do.
BTW: if they do get 1M PPCs in 1994, some people think that would put
them #1 in RISCs, which is not likely ... since Intel i960s got 2M in 1992,
and I'd guess both AMD 29Ks and MIPS Rxxxs would be >1M in 1993,
and more in 1994.
Interesting times...
Tom Barrett
>
>Yes, it is CMOS level compatible as well.
Actually, that wasn't exactly what I asked. My question was whether
or not the 5v compatible inputs can handle a maximum overshoot of 8v
total? The 7-8v maximum input is a common problem with 3.3v logic.
Tom
sur...@vax2.winona.msus.edu
> By the way, MacOS system 7.0.1 runs quite well on my ancient Mac II.
>
> B. Coomes, Dept. of Math. and Comp. Sci., Univ. of Miami
Good point. I sincerely hope that in this great shift toward new
technology that we (those of us with older machines) are still able
to use new versions of software even if we won't be able to reap
all the benifits of the new processor. Bonus goes to apple for making
sure my mac IIx still runs the latest sys and apps. Lets hope that
when the PPC and Pentium products hit the market in volume that we
aren't _forced_ to buy new hardware to support *most* of the inter-
face improvements.
seth
Don North
>
>qm...@casbah.acns.nwu.edu (Charlie Kuehmann) writes:
>] doco...@sedona.intel.com (Dennis O'Connor) wrote:
>] Your comparisons were based on assumptions that the PPC 601 was a 3V chip.
>] My impression was that the first PPC chip the 601 would be a 5V chip and
>] the second the 603 would be the low power 3V chip. Can anyone who has the
>] MOTO spec sheets on the PPC confirm this? If the 601 is indeed a 5V chip
>] then I think the power dissipation numbers would definately favor the PPC.
>
>It would be nice if someone could independantly confirm, but my
>source for the 3V assumption is the Motorola PPC601 literature.
>Specifically, page 2 of the April 93 version of
The PPC601 is actually a 3.6V part.
--
Don North ---- Apple Computer, Inc. ---- Advanced Technology Group
...!apple!north no...@apple.com NORTH KD6JTT etc,etc,etc
{ Facts are facts, but any opinions expressed are my own, and *do not* }
{ represent any viewpoint, official or otherwise, of Apple Computer, Inc }
Phil Brownfield
Dennis O'Connor <doco...@sedona.intel.com> wrote:
>slightly in favor of the Pentium processor, and the PPC601 enjoys
>about a ~50% advantage in floating point. PPC601 pulls (according
>to the Motorola ads) 8.5W worst-case at 3V, while the latest stepping
>of the Pentium processor pulls I think less than 14W worst-case at 5V.
>Pentium chips cost $965 (I think) in Qty.1K, PPC601 costs ~$450 in Qty.20K.
Motorola's announced quantity 20,000 MPC601 prices are $280 at 50MHz
and $374 at 66 MHz. The $450 price is the ballpark price at 66 MHz
in quantity 1000, as advertised in the Wall Street Journal and elsewhere.
Sorry for replying late - your posting only today came to my
attention.
--
Phil Brownfield
Motorola Inc. RISC Microprocessor Division +1 512 891 3064
p...@grinch.sps.mot.com or phil_br...@email.sps.mot.com
Not a spokesman for my employer
cbs...@ccnga.waterloo.edu
>cbs...@ccnga.waterloo.edu writes:
>>"Current pricing" is providing Intel will a smaller per-chip profit than
>>486's at their rock bottom prices. Considering the 603's should be
>>getting down to the prices of 486's, this should be a concern.
>
>First, no-one really knows what the margins are. Second, process improvements
>ramp up very quickly with these things. Even as Pentium prices go down, I'd
>bet the margins will increase.
Almost certainly this will be the case. However, I'd be very surprised
if Intel is going to be having an increadible breakthrough in process
improvements to match where Motorola is. The 601 is currently being
produced at a *lower* volume than the Pentium at a *lower* per-chip
cost. The 603 isn't much more than a few months from production at an
even lower per-chip cost. We're talking about per-chip pricing that's
here *now* or *very soon*. Pentium is for the moment, behind the eight
ball, and it doesn't look like it's going to catch up anytime soon.
>>If people made the jump for Windows, why wouldn't they make the jump for
>>PPC? Historical evidence to suggest you are wrong.
>
>Because Windows offered a major shift in interfacing with the computer. It
>made the PC "Mac-like", easier to use. There is nothing of PPC that provides
>any such advantage over other PC's. And, Windows is _cheap_. We're talking
>software with WIndows, complete system replacement with PPC. How can you
>compare those?
"Windows offered a major shift in interfacing with the computer." I'll
grant you that. PPC's are going to offer similar oppurtunities w/Voice
recognition, multiple platform support, multimedia, etc.
"And, Windows is _cheap_." Gee... isn't that supposed to be the big
difference between a PPC and a Pentium? The price? So quickly you have
forgotten what all was involved with the jump to Windows. To get
computers that could decently run windows, all the old machines had to
be turfed and brand new '386's with VGA controllers & monitors had to be
purchased. Bigger harddrives & more RAM were needed. You call this
cheap?
>>What will interface directly with the CPU? We have PCI which removes the
>>need for a local bus so now nothing *has* to work directly with the CPU.
>
>Well, I'm no PCI expert, but I know there are at least 4 chips in a PCI
>chipset, at least one of which interfaces with the CPU. At least, that is
>how it was explained to me once.
>
Ok, but wouldn't that be the cost of installing the PCI bus in the
motherboard? I was under the impression the cards themselves were pretty
much removed from the CPU compeltely.
--Chris
--
Christopher Smith
cbs...@boomer.uwaterloo.ca
cbs...@undergrad.math.uwaterloo.ca
DC 3527
(519) 885-1211x3581
Adam Nash
Tony Hamilton
>>Because Windows offered a major shift in interfacing with the computer. It
>>made the PC "Mac-like", easier to use. There is nothing of PPC that provides
>>any such advantage over other PC's. And, Windows is _cheap_. We're talking
>>software with WIndows, complete system replacement with PPC. How can you
>>compare those?
>"Windows offered a major shift in interfacing with the computer." I'll
>grant you that. PPC's are going to offer similar oppurtunities w/Voice
>recognition, multiple platform support, multimedia, etc.
Really, what are these opportunities? And PC's _don't_ do voice recognition,
multiple platform support, multimedia, etc. !? Where have you been?
Tony Hamilton
tham...@pcocd2.intel.com
Terje Rydland
Hamilton) wrote:
No, PC's don't do speaker-independet, no training, voice recognition that
works in existing apps without updates.
Yes, you can run different system software on PC's, but the only good
one is not used much (NeXT Step). On the Mac the good one is used
by everybody!
Multimedia on a PC is a joke compared to what you can do on most Macs.
There is not one PC out there with the performance and capabilities of
the AV Macs. Software that uses the DSP (like PhotoShop filters etc.)
is as fast on the 840 av as it will be to run PhotoShop native on PPC.
And, yes, we've been here all the time, but our perspective is slightly
different than the ordinary PC user!
------------------------------------------------------------------------
Terje Rydland Tel.:07/591845 (+47-7-591845)
Dept. of Informatics Fax.:07/591733 (+47-7-591733)
UNIT/AVH E-Mail:terje....@ifi.unit.no
N-7055 Dragvoll AppleLink: NOR0103
Norway
-------------------------------------------------------------------------
Steve Kanefsky
Tony Hamilton <tham...@fnugget.intel.com> wrote:
>
>Really, what are these opportunities? And PC's _don't_ do voice recognition,
>multiple platform support, multimedia, etc. !? Where have you been?
If you reduce it to a grocery list of features, you're right (and
unfortunately, that's how a lot of people buy computers). Now go
actually use one and see what the difference is.
--
Steve Kanefsky
John West
Hamilton) wrote:
The more appropriate question is "Where have *you* been?" Obviously, you
weren't out learning to read.
cbsmith said "PPC's are going..." not PC's. It is apparent that Intel
employees are so fixated on their company's upcoming demise that they are
flaming everything they can; justifiable or not.
Back to your comment "Tony," you're right PC's don't do any of those
things. Macs do speech recognition and multimedia, though. And once moved
to the PowerPC, the Mac will have to have multiplatform support, won't it?
It *will* be on at least two platforms at that time.
Good luck,
-john
--
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
My views are not my own, and certainly not those of Visix Software Inc.
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
John West (jo...@visix.com) :: "Conciousness is not a plaything."
Visix Software Inc. :: -R.D. Laing
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
Timothy Jehl
In article <johnw-250...@twinpeaks.visix.com>, jo...@visix.com (John West) writes:
>
> It is apparent that Intel
> employees are so fixated on their company's upcoming demise
I keep trying to fixate on my company's upcoming demise, but my concentration
is continuously broken by my stock broker telling me how much money I'm making
in Intel stock! 8-{), for the humor impared.
A large number of people, obviously the uniformed rabble, are out there buying
Intel stock like hotcakes (with maple syrup and a side of butter). The really
good news is that this same group of unwashed masses is the most likely group
to spend money on purchasing a personal computer.
I know you guys think that Intel should be condemned to the fiery pit for
deigning to design what consumers want, rather than what they _should_ want,
but I feel that the rumors of Intel's impending doom may be a touch exagerrated.
And, as always, it's not paranoia when they really are out to get you.
TJ Jehl
--
Steve Birnbaum
Hamilton) wrote:
> >"Windows offered a major shift in interfacing with the computer." I'll
> >grant you that. PPC's are going to offer similar oppurtunities w/Voice
> >recognition, multiple platform support, multimedia, etc.
>
> Really, what are these opportunities? And PC's _don't_ do voice recognition,
> multiple platform support, multimedia, etc. !? Where have you been?
This is hilarious! The PPC will most likely have DSPs in them so voice
recognition will top anything for the current PC. Also, I wouldn't doubt
that the DSP will be faster than the one in the new av Macs. Add to that
the advanced speed, the fact that it supports multiple OS via PowerOpen
with pre-emptive tasking and the fact that it will blow any PC out of the
water speed-wise, how can you question it?
Dennis O'Connor
aa...@freenet.carleton.ca (Steve Birnbaum) writes:
] In article <>, tham...@fnugget.intel.com (Tony Hamilton) wrote:
]
] > >"Windows offered a major shift in interfacing with the computer." I'll
] > >grant you that. PPC's are going to offer similar oppurtunities w/Voice
] > >recognition, multiple platform support, multimedia, etc.
] >
] > Really, what are these opportunities? And PC's _don't_ do voice recognition,
] > multiple platform support, multimedia, etc. !? Where have you been?
]
] This is hilarious! The PPC will most likely have DSPs in them so voice
] recognition will top anything for the current PC.
Well, one has to wonder why ? Why would PPC system makers add
DSPs to their products, and PC vendors not ?
There's a few possible reasons :
1. The market isn't ready _yet_.
This was true years ago, and may not be true anymore.
Market readiness depends on price, performance and
the users ability to exploit (i.e. applications).
The apps that can exploit a DSP may now have matured
enough to be useable by Everyman, and DSPs are certainly
cheap enough these days. We'll see.
2. The market _is_ ready, but there are no standards
PPC-Macs won't have this problem : Apple sets the standards.
PPC systems running WindowsNT _will_ have this problem unless
Microsoft sets or has set a standard software DSP interface,
much like the Win32 and (future) Win-Telephony standards.
(This would also allow DSP-less systems to emulate the DSP,
solving the installed-base-compatability problem, some,
which helps the chicken-egg software-hardware problem.)
3. The market _is_ ready, but the CPU has enough horsepower
to drive the app, so why bother putting a DSP in ?
After all, most users don't multi-task, and don't fully
exploit the CPU as a result (e.g. editting : take a
character from the kbd, busy loop, ad infinitum).
Why would they pay even the $10 extra for a DSP ?
(Good marketting by DSP-proponents might overcome this.)
Point 2 is the only thing I can see that differentiates Macs from
Intel-, AMD-, IBM- and Cyrix-based systems. However, _if_ WindowsNT
on PPC does have a standard DSP interface, then WindowsNT on Intel
Architecture will too. And likely regular-old Windows will as well.
So we may see DSPs on "PC" motherboards. Stranger things have happened.
Point 3, if it's true, is the most beneficial to the CPU makers.
There's a lot more margin in a PPC601 or Pentium(tm) chip than
a DSP, after all. If it's true, then maybe neither PPC-WindowsNT nor
Intel Architecture systems may have DSPs as standard equipment.
PPC-Macs DSPs: who knows ? The Mac market is like the Euro-luxo-
sports-car market compared to the Windows and WindowsNT domestic-
mid-size-sedan-market. Totally different market forces.
As usual, the reality is to weird to allow reliable prediction.
But at least, there's no technical "stopper" to PCs having fast DSPs.
They don't _now_, but do we know that current PPC machines do, either ?
Steve Chang
:I keep trying to fixate on my company's upcoming demise, but my concentration
:is continuously broken by my stock broker telling me how much money I'm making
:in Intel stock! 8-{), for the humor impared.
:A large number of people, obviously the uniformed rabble, are out there buying
:Intel stock like hotcakes (with maple syrup and a side of butter). The really
Actually, IMHO Intel stock seems undervalued right now. Its P/E ratio
is one of the lower ones I've seen. If I had a few spare $1000's then
I would probably buy some. I would if I hadn't ordered that C660av.
(I'm not flaming Intel :). )
Steve
Tony Hamilton
>>
>> >"Windows offered a major shift in interfacing with the computer." I'll
>> >grant you that. PPC's are going to offer similar oppurtunities w/Voice
>> >recognition, multiple platform support, multimedia, etc.
>>
>> Really, what are these opportunities? And PC's _don't_ do voice recognition,
>> multiple platform support, multimedia, etc. !? Where have you been?
>>
>> Tony Hamilton
>> tham...@pcocd2.intel.com
>The more appropriate question is "Where have *you* been?" Obviously, you
>weren't out learning to read.
>cbsmith said "PPC's are going..." not PC's. It is apparent that Intel
>employees are so fixated on their company's upcoming demise that they are
>flaming everything they can; justifiable or not.
*sigh*. Look at my message again. Yes, cbsmith said PPC, and I said PC, and
both are used in the right places. If that makes no sense to you, then stop
posting and making yourself look like a fool. Sorry, but you had it coming.
You're accusing me of flaming unjustly.
>:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
>My views are not my own, and certainly not those of Visix Software Inc.
>:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
Well, I can hardly argue with someone who doesn't even speak for themselves.
Tony Hamilton
tham...@pcocd2.intel.com
Charles Chung
>Actually, IMHO Intel stock seems undervalued right now. Its P/E ratio
>is one of the lower ones I've seen. If I had a few spare $1000's then
>I would probably buy some. I would if I hadn't ordered that C660av.
>
>Steve
Intel's P/E ratio may be low simply because its a computer stock. Given
the extreme volatility and cutthroat competition in the computer industry,
buying stock in something like Coca-Cola may be a better decision.
At least's Coke's flagship product won't be obsolete in 24 months.
-Chuck
--
*******************************************************************
Chuck Chung (919) 660-2539 (O)
Duke University Dept. of Physics (919) 684-1517 (H)
Timothy Jehl
In article <19...@news.duke.edu>, cch...@phy.duke.edu (Charles Chung) writes:
>
> Intel's P/E ratio may be low simply because its a computer stock. Given
> the extreme volatility and cutthroat competition in the computer industry,
> buying stock in something like Coca-Cola may be a better decision.
>
> At least's Coke's flagship product won't be obsolete in 24 months.
>
You've got to be kidding? Haven't you hearn about New Coke? It's this
great new product that tastes even better than the original. All of the
experts agree that, any day now, the entire Coke drinking population is going
to stop buy Classic Coke and switch to Coke II.
Yeah, right ...
TJ
--
Erik. A Speckman
>
>Intel's P/E ratio may be low simply because its a computer stock. Given
>the extreme volatility and cutthroat competition in the computer industry,
>buying stock in something like Coca-Cola may be a better decision.
>
>At least's Coke's flagship product won't be obsolete in 24 months.
Aren't you forgetting New Coke?
--
_____________________________________________________________________________
Erik Speckman Good Brain
espe...@romulus.reed.edu Doesn't Suck
Steve Chang
: Intel's P/E ratio may be low simply because its a computer stock. Given
: the extreme volatility and cutthroat competition in the computer industry,
: buying stock in something like Coca-Cola may be a better decision.
Its P/E was low even for a hi-tech firm a few weeks ago, but looking at it
now it seems improved.
Steve
Steve Birnbaum
doco...@sedona.intel.com (Dennis O'Connor) wrote:
> This was true years ago, and may not be true anymore.
> Market readiness depends on price, performance and
^^^^^
Yes, let's talk price. The PPC will sell for about $2000 and the Pentium
for what, oh, about $5000? The PPC is faster, runs cooler, uses less power
and runs more varied operating systems than the Pentium. The PPC will be
able to run Windoze as well as System 7. This makes life easier for lot of
people. Even for present Intel users who have no intention of ever loading
System 7, you'd think that since they've always complained that Macs were
too expensive and they were going for the lowest price, they'd keep with
that and buy a PowerPC.
> 3. The market _is_ ready, but the CPU has enough horsepower
> to drive the app, so why bother putting a DSP in ?
I've heard two difference stories on this. One, they will include the DSP
and the other that the CPU will handle it all, no problem. Guess we'll
have to wait and see.
Graham K. Johnson
>Yes, let's talk price. The PPC will sell for about $2000 and the Pentium
>for what, oh, about $5000? The PPC is faster, runs cooler, uses less power
Wrong. Basic Pentium systems _now_ are less (much less?) than $5000;
when you can actually buy a PPC system for less than that (what, 7
months from now?), Pentium will undoubtedly cost less (not to mention
new, faster, cooler, uses-less-power Pentium designs that will
probably be introduced by then).
Let's talk price when you can actually _buy_ a PPC system.
BTW: PC/Computing Sept. 93 reports that the 5-6 Pentium systems they
tested actually ran _cooler_ than a DX2/66.
>and runs more varied operating systems than the Pentium.
Excuse me? "Runs" more varied operating systems? Maybe by, say, 1995
the number of OS's for PPC will be comparable to the x86's selection,
but not before then. When Apple introduces its PPC Macs there will be
what, _2_ OS's for PPC? AIX and System 7?
--
___
/__ ___ "Uh, you're not gonna joh...@cps.msu.edu
<__/raham / eat that, are you?" Michigan State University
|<. <_/ohnson --Fish Tales East Lansing, Michigan
Steve Kanefsky
>] In article <DOCONNOR.93...@potato.sedona.intel.com>,
>] doco...@sedona.intel.com (Dennis O'Connor) wrote:
>]
>] > This was true years ago, and may not be true anymore.
>] > Market readiness depends on price, performance and
>] ^^^^^
>]
>] for what ....
>
>In the discussion given, which was about wether DSPs would be incorporated
>intop PCs and PPCs, it's the price of the *DSP* and it's support circuits
>that matter. If, as you assume, Pentium(tm) chips drive the cost of
>systems using them way up (an assumption I disagree with), it actually
>_favors_ the probability that Pentium chip based systems would have DSPs.
You may not agree with the prices that Steve offered up for PPC and
Pentium systems, but I consider it highly unethical to modify what he
said and then quote it without mentioning that you changed it. He said
that the PPC would sell for about $2000 and a Pentium system for about
$5000. He may not be right, but don't misquote him.
Since you think the extra cost of a Pentium wouldn't drive the cost of a
system way up, I would think you'd agree that a system using two PPC
chips or a PPC chip and a DSP would still be no more than the price of a
Pentium system without a DSP, or even much more than a single PPC system,
since you apparently think that 2-3 times the price for the one of the most
expensive components in the system won't have a large impact on the price
of the system. There may be extra support circuitry required for two
processors versus one, but you've got plenty of money to spare after
buying two PPC chips (or a PPC and a DSP) instead of a Pentium :-)
>] and runs more varied operating systems than the Pentium. The PPC will be
>] able to run Windoze as well as System 7.
>
>Pentium _right_ _now_ runs MS-DOS, PC-DOS, DR-DOS, Windows, WindowsNT,
>OS/2, NeXTStep, Solaris, and more variants of Unix than I have fingers.
>I can't imagine a more varied set of operating systems.
I can. MS-DOS, PC-DOS, and DR-DOS aren't exactly very varied. The PPC
will run almost all of those plus PowerOpen, Taligent, and System 7.
Nows *that's* varied.
--
Steve Kanefsky
Dennis O'Connor
aa...@freenet.carleton.ca (Steve Birnbaum) writes:
] In article <DOCONNOR.93...@potato.sedona.intel.com>,
] doco...@sedona.intel.com (Dennis O'Connor) wrote:
]
] > This was true years ago, and may not be true anymore.
] > Market readiness depends on price, performance and
] ^^^^^
]
] for what ....
In the discussion given, which was about wether DSPs would be incorporated
intop PCs and PPCs, it's the price of the *DSP* and it's support circuits
that matter. If, as you assume, Pentium(tm) chips drive the cost of
systems using them way up (an assumption I disagree with), it actually
_favors_ the probability that Pentium chip based systems would have DSPs.
] and runs more varied operating systems than the Pentium. The PPC will be
] able to run Windoze as well as System 7.
Pentium _right_ _now_ runs MS-DOS, PC-DOS, DR-DOS, Windows, WindowsNT,
OS/2, NeXTStep, Solaris, and more variants of Unix than I have fingers.
I can't imagine a more varied set of operating systems.
However, be that as it may, what does this have to do with whether PCs
will use DSPs ? Is it a positive influence ? Maybe not. It depends
on whether a cross-OS standard interface to the DSP (as part of an ABI,
I imagine) is available, as my original post says.
] > 3. The market _is_ ready, but the CPU has enough horsepower
] > to drive the app, so why bother putting a DSP in ?
]
] I've heard two difference stories on this. One, they will include the DSP
] and the other that the CPU will handle it all, no problem. Guess we'll
] have to wait and see.
Yep.
Garrett Drysdale~
>ug241...@ccnga.waterloo.edu> <thamilto.746246817@fnugget>
>Date: Wed, 25 Aug 1993 17:33:21 GMT
>Lines: 16
Gee, I thought the PPC would be soooo fast as to eliminate the need for
a DSP. :) for the humor impaired. I mean, wow, maybe there really is a
use for that FPU after all. You know, if you put enough processors
on the PPC, it will probably outperform a CRAY. Oh wait, I'm way off
track, the Cray doesn't do multimedia, does it? :))
Enjoy your A&PCPPC (architecturally & politically correct PPC) :). Intel
will enjoy selling millions of 486's and Pentiums to those who really
don't care (or know in many cases) whether there PC has an
architecturally correct RISC or a obviously dead CISC. :))
Note the generous smileys. Take it as intended.
Garrett
--
=============================================================================
Garrett Drysdale Intel Corp.
gdry...@sedona.intel.com
Disclaimer: The views of the poster are his own and not necessaryily those
of Intel.
Mark M Lee
Steve Kanefsky writes.....
>>Pentium _right_ _now_ runs MS-DOS, PC-DOS, DR-DOS, Windows, WindowsNT,
>>OS/2, NeXTStep, Solaris, and more variants of Unix than I have fingers.
>>I can't imagine a more varied set of operating systems.
>
>I can. MS-DOS, PC-DOS, and DR-DOS aren't exactly very varied. The PPC
>will run almost all of those plus PowerOpen, Taligent, and System 7.
>Nows *that's* varied.
Really? I see you use "will." You mean 1995 and beyond?
I think the person is emphasizing "_right_ _now_"
-mark
mm...@athena.mit.edu
Dennis O'Connor
] You may not agree with the prices that Steve offered up for PPC and
] Pentium systems, but I consider it highly unethical to modify what he
] said and then quote it without mentioning that you changed it. He said
] $????. He may not be right, but don't misquote him.
I took it out because I considered it irrelevant to the particular
discussion, regardless of it's correctness. I explicitely stated
my opinion that it was irrelevant. It was pretty obvious editting.
Editting out elements of a discussion that you don't wish to address
is pretty standard practice on the net.
So, you are wrong. There was nothing unethical in what I did.
I did not misrepresent the content of anything anyone said.
Bill Schneller
to make the PENTIUM systems cost 2x as much as PPC systems. The PPC is
$450 the PENTIUM is $995 thats $445 more. It is fairly safe to assume that
the rest of the PC parts cost about the same for either. So this makes
PENTIUM machine $445 more than PPC machines. This $2000 PPC machine that
people keep quoting will probably be a 50Mz PPC w/ 4Meg ram and 80Meg HD
no montior and no keyboard. Much like the PENTIUM for $2495 with 8meg ram
no video no HD no monitor and 60Mz processor. People seem to forget 66Mz
motherboards are still expensive regardless of the processor put on them.
Bill
Russ Taylor
: I have seen gobs and gobs of posts about how the price of the cpu is going
Bill. Guess what. Parts are a SMALL part of the cost of a computer. For
the final shock to the consumer, double or triple the CPU price difference.
Take some basic economics and learn how raising the raw material price by
$50 translates into a greater price increase down the line.
Or are PC fans just not capable of understanding pricing?
Chung-Hsiung Yang
For the competitive market that PCs are in the cost to the consumer
is minimal because there are so much competition. The Pentium is an exception
at this moment of course, it is limited in quanty from Intel and only
major PC makers makes the Pentium machines. Its prices will be inflated
for now.
Secondly, the Pentium based machines, unlike what Apple plans for
the first PPCs, are not built to be cheap machines. A Pentium based
machine now would be *much* more expandable than the first PPCs, because
many the initial P5 based machines are designed to be server class
machines. The buses are the faster, there are more expansion slots,
and there are often multiple standard supports (ISA, VESA, EISA, maybe
PCI).
Also, the pricing between the Pentium based machines and
PPC 601 is a bit difficult to compare one to one. Pentium based
machine cost about $4000-$5000, but it comes up a huge*disk
the ones I seen have 500MEG+ to 1Gig (<12ms access time). The monitors
are often 17" (there are some 15" monitors that are also being packaged).
About 16 MEGs of RAM. And Keyboard, mouse. If you add all this up,
Pentium shouldn't be a cheap computer.
One the otherhand, the $2000 price for the PPC is just the
box maybe with 8 MEG RAM and 230MEG of harddisk. Everything else
extra. (maybe you get the mouse free.)
So it is kind of unfair to say the a Pentium based machines
costs about twice the PPC without considering the package it comes with.
Granted PPC is supposed to be faster, which I agree with largely.
But to think one can have PPC up and running by only spending $2000, I
think it is a bit too naive. I am looking at $3500-$4000 realistically,
because I want a PPC with about 400MEG of HD, 16MEG RAM or even 32MEGs.
17" monitor, and keyboard + mouse. And a decent Operating System,
which ever it is. Chances are it will cost more $$$$.
- Chung Yang
Bill Schneller
Clone makers selling direct with margins in the teens don't triple the
price of the cpu. Look at the adds that have a system then the prices
for the same system with different cpus. The difference is only slightly
more than the difference in the cpu prices. The MAC world is just accustomed
to being raped on price. This is not so true any more with more 50% price
slashes of mac hardware. If APPLE is going to sell PPC based machines for
$2000 and compete with clone makers they are going to have to sell their
machines at very low margins like the clone makers do. So the price difference
between machines will be the cost difference of the hardware plus margins.
Not too long ago IBM and APPLE did seel their machines with 300% and 400%
margins and yes selling this way and $500 increase in cpu price would translate
to a $1500 to $2000 system price increase. Times have changed and margins have
dropped. Clone makers are already paying more for DX2-66 than apple pays
68040s and the PCs are still cheaper. Kind of strange isn't it.
Bill
Roy Eassa
The chip prices you quoted are wholesale, no? Don't forget markups.
Also, I believe both prices are now lower: something like $385 vs. $845 (I
forget the exact numbers) in the most-commonly-purchased quantities.
Anyway, if you assume that the standard markup is 2x (I have no basis for
picking this particular value, BTW), then the difference in price of the
final system is a lot greater than your post suggests.
Steve Kanefsky
Mark M Lee <mm...@athena.mit.edu> wrote:
>
>Steve Kanefsky writes.....
>>>Pentium _right_ _now_ runs MS-DOS, PC-DOS, DR-DOS, Windows, WindowsNT,
>>>OS/2, NeXTStep, Solaris, and more variants of Unix than I have fingers.
>>>I can't imagine a more varied set of operating systems.
>>
>>I can. MS-DOS, PC-DOS, and DR-DOS aren't exactly very varied. The PPC
>>will run almost all of those plus PowerOpen, Taligent, and System 7.
>>Nows *that's* varied.
>
>
>I think the person is emphasizing "_right_ _now_"
I was merely responding to the remark that the list of eight operating
systems he gave, (three of which were DOS and two of which were Windows,
at least interface-wise), was not the most varied set of operating system
he could imagine.
As far as when they're available, if you can't wait, by all means buy a
PC now if you *have* to run one of those operating systems. As soon as
they're available for the PPC (as little as 6 months for some of them),
avoid the 486/Pentium platform like the plague. And if you need the only
decent graphical OS with a large native software base right now, get a Mac.
--
Steve Kanefsky
Dan Hildebrand
Steve Birnbaum <aa...@freenet.carleton.ca> wrote:
>
>Yes, let's talk price. The PPC will sell for about $2000 and the Pentium
>for what, oh, about $5000?
Pentium machines already sell for $2495.
>The PPC is faster,
The same SpecInt, but faster on SpecFP. Are you willing to bet that the
Pentium won't be faster again by the time PowerPC systems ship?
>runs cooler, uses less power
Other than the laptop market, does this rally matter? Compared to the overall
power consumption of the system, does a few extra watts on the processor make
that much difference?
>and runs more varied operating systems than the Pentium.
Have you ever counted all the OS's available for the x86 processor? How many
OS's are available for the Mac? From all this, how do you determine that the
PowerPC will have more to offer?
>The PPC will be
>able to run Windoze as well as System 7. This makes life easier for lot of
>people. Even for present Intel users who have no intention of ever loading
>System 7, you'd think that since they've always complained that Macs were
>too expensive and they were going for the lowest price, they'd keep with
>that and buy a PowerPC.
How many apps remain for which only a Mac version, and not an MS Windows
version, is available?
--
Dan Hildebrand email: da...@qnx.com
QNX Software Systems, Ltd. QUICS: danh (613) 591-0934 (data)
(613) 591-0931 x204 (voice) mail: 175 Terence Matthews
(613) 591-3579 (fax) Kanata, Ontario, Canada K2M 1W8
Gary Snow
>
>] and runs more varied operating systems than the Pentium. The PPC will be
>] able to run Windoze as well as System 7.
>
>Pentium _right_ _now_ runs MS-DOS, PC-DOS, DR-DOS, Windows, WindowsNT,
>OS/2, NeXTStep, Solaris, and more variants of Unix than I have fingers.
>I can't imagine a more varied set of operating systems.
Ah, come on, lets not try and slip that by us. You can't count three
different flavors of DOS as three different operating systems (MS-DOS,
PC-DOS, DR-DOS).
Gary
--
-----
Gary Snow
uunet!clark!gsnow or gs...@clark.edu
Dennis O'Connor
kane...@halcyon.com (Steve Kanefsky) writes:] >> [ I wrote ]
] >>>Pentium _right_ _now_ runs MS-DOS, PC-DOS, DR-DOS, Windows, WindowsNT,
] >>>I can't imagine a more varied set of operating systems.
] >>
] >>I can. MS-DOS, PC-DOS, and DR-DOS aren't exactly very varied. The PPC
] >>will run almost all of those plus PowerOpen, Taligent, and System 7.
] >>Nows *that's* varied.
] >
] >Really? I see you use "will." You mean 1995 and beyond?
] >I think the person is emphasizing "_right_ _now_"
]
] I was merely responding to the remark that the list of eight operating
] he could imagine.
Since I included NeXTStep, an advanced object-oriented windowing OSs,
I beg to differ. (Also, as a minor nit, the list is 9+ OSs, counting
"more variants of UNIX than I have fingers" as "1+".) What could be
more varied than a box that runs both not-much-more-than-a-file-handler
MS-DOS _and_ NeXTStep486 ? That's like a car that runs on anything from
banana peels and beer to plutonium and tritium (movie reference :-).
I don't know that any of the current or future Mac OSs are much more
innovative or unique than NeXTStep is. ( I'm pretty sure MAC OS's
don't go as far the other way as MS_DOS! :-) Anyone intimate with both
the Mac OSs and NeXTStep care to comment ?
Tony Hamilton
First, understand that this is more of a query than an affirmation, since I
would like to know myself. But, from what I know, those "formulas" of using
the cost of the CPU as a basis are basically based on the premise that you
"get what you pay for" in a CPU. Well, assuming of course the prices given,
and that these chips are of relative equal performance (for most tasks, this
seems to be true), I can't see how you can simplify it like that.
Now to the query part. As a RISC chip, doesn't the PPC require more off-CPU
support, such as a larger L2 cache, primary memory (more instructions required
to perform equivalent tasks) and maybe more things? I would also guess that this
extra cache and memory might need to be of a higher performance type also, given
the CPU's dependence upon it (I am assuming that a RISC chip might be more
dependent upon the system design for performance optimization than a CISC chip.
Please let me know if this assumption is baseless).
Well, anyway, the way I see it, firstly the two chips won't have the same
price difference by the time they are truly competing. Second, based on the
above, I am _thinking_ that the PPC might require more hardware, but maybe
the Intel architecture requires hardware PPC doesn't. Anyway, I simply cannot
see where the prices of systems would be so different, for equivalent performance.
Just as a $3,000 Pentium today is suspected of being an other-than-optimum
system design, I suspect a cheap PPC system might not be of the best design
either (and of course cheap Pentium systems will surely cost less in 6 months.)
I know some of these topics have been discussed in parts on this group, but
could some _knowledgeable_ person please outline the system design issues of
a Pentium system vs. a PPC system? Is anyone out there working on both, and
therefore could provide good insight into the matter?
Tony Hamilton
tham...@pcocd2.intel.com
Dennis O'Connor
gs...@clark.edu (Gary Snow) writes:
] In article <> doco...@sedona.intel.com (Dennis O'Connor) writes:
] >
] >] and runs more varied operating systems than the Pentium. The PPC will be
] >] able to run Windoze as well as System 7.
] >
] >Pentium _right_ _now_ runs MS-DOS, PC-DOS, DR-DOS, Windows, WindowsNT,
] >OS/2, NeXTStep, Solaris, and more variants of Unix than I have fingers.
] >I can't imagine a more varied set of operating systems.
]
] Ah, come on, lets not try and slip that by us. You can't count three
] different flavors of DOS as three different operating systems (MS-DOS,
] PC-DOS, DR-DOS).
I wasn't trying to address the cardinality of the set. I threw those
three in to show how many vendors there are for even the most basic OS.
This is at least slightly relevant to how "varied" the set is.
I at least had only one entry (not counting Solaris) for all the
UNIX OS variants and vendors !
But mainly by "varied set" I refer to how much difference there is
between the various members of the set : as another post of mine says,
NeXTStep486 is a way way different from MS-DOS.
Steve Kanefsky
Dan Hildebrand <da...@quantum.qnx.com> wrote:
>In article <aa229-260...@192.139.46.27>,
>Steve Birnbaum <aa...@freenet.carleton.ca> wrote:
>>
>>Yes, let's talk price. The PPC will sell for about $2000 and the Pentium
>>for what, oh, about $5000?
>
>Pentium machines already sell for $2495.
And are basically just '486 machines with a Pentium thrown in. They
don't take full advantage of the Pentium, and would be blown away by a
$2495 PPC machine.
>>The PPC is faster,
>
>The same SpecInt, but faster on SpecFP. Are you willing to bet that the
>Pentium won't be faster again by the time PowerPC systems ship?
Yes. In order to keep the heat and power down, it'll probably be a
while before they can bump up the speed. Prototype 80MHz PPC have
already been shown months ago, and 66MHz PPCs are shipping to customers
now (the customers of a CPU chip are system manufacturers like Apple,
IBM, Compaq, Gateway, etc.) I'd bet the 80MHz PPC 601 is ahead of any
80MHz or faster Pentium.
>>runs cooler, uses less power
>
>Other than the laptop market, does this rally matter? Compared to the overall
>power consumption of the system, does a few extra watts on the processor make
>that much difference?
Would 10 extra pounds resting on top of you make any difference? If it
were spread out, no. If it were the size of a needle, it might kill
you. That much extra power concentrated in a tiny space makes a big
difference. For one thing it means that most of those "Pentium ready"
systems out there couldn't use it without melting down, if the power was
not reduced.
>>and runs more varied operating systems than the Pentium.
>
>Have you ever counted all the OS's available for the x86 processor? How many
>OS's are available for the Mac? From all this, how do you determine that the
>PowerPC will have more to offer?
Because Motorola and Microsoft has announced a Windows NT port, Insignia has
announced a DOS/Windows port, Sun has announced a Solaris port, Apple
and IBM have announced PowerOpen, Apple has announced System 7, IBM has
announced AIX, Taligent has all but announced Taligent (or whatever they
decide to call it), NeXT is rumored to have been porting NeXTstep for use
on their own PowerPC hardware before they sold the hardware part of the
business, etc.
It appears that the first IBM PowerPC workstations running AIX will be
out in September (based on another article on the net I read today).
They'll be expensive but they should blow away any of the existing
Pentium systems with their high-performance hardware and operating system
that's PPC optimized (unlike any of the Pentium operating systems).
--
Steve Kanefsky
Dennis O'Connor
kane...@halcyon.com (Steve Kanefsky) writes:
] >The same SpecInt, but faster on SpecFP. Are you willing to bet that the
] >Pentium won't be faster again by the time PowerPC systems ship?
]
] Yes. In order to keep the heat and power down, it'll probably be a
Steve, the next Pentium(tm) chip, the P54C, is on a newer, smaller,
faster process ( .6 instaead of .8 micron, I _think_ ), and operates
at 3.3V instead of 5. This means big reduction in power dissipation
( halving it, or better ) and at the same time an increase in speed
and yield-per-wafer from the newer process.
All of which adds up to a faster, cheaper, more competive Pentium
chip, compared to the current incarnation.
When this chip and PPC will each be shipping in enough volume for you
or me to _really_ care about, I can't predict. P54C is rumoured by the
press to be shipping in volume 1H94. I haven't trracked the PPC rumours.
dipa...@esvx12.es.dupont.com
Come on guys,
This thread is getting very tiresome. The PPC is a great chip. I'm
sure that it is going to give users great increases in performance.
But get real, there are how many TENS of MILLIONS of 386 or better
x86 computers out there? Do you think that the current installed base
of PC users is going to just drop thier equipment and buy PPC's? Do
you really think that Joe Shmo on the street who has a 486 on his desk
is going to buy a PPC for home use with 486 systems selling for ~$1500?
Dare I say, NOT! I believe that the market for PPC based PC's will
explode when:
a) Systems are shipping in quantity.
b) Native software becomes available, and upgrades to that software
are provided at low cost to the end user.
c) PPC systems provide a functionality that is not available on x86
based systems.
Now, sure the PPC beats that pants off of the Pentium in FP. But
don't you think that in most cases the FP of the Pentium will prove
adequate? I feel that there is plenty of room in the PC pie to
accommodate Pentium/PPC/Alpha or whatever. I personally don't care
wether the chip in my PC is RISC or CISC, as long as I can get my work
done on the doggone thing. As I've stated before, I enjoy learning
the technical details of Pentium/PPC, but enough with the RELIGIOUS
WARS. Some of us have become way too uptight :*( .
Nick (Trying to get some peace and quiet) DiPaolo
Disclaimer: I do not speak for the DuPont Co.
Disclaimer: I'm not flaming Dennis either ;*)
(I just had to say that!)
Hank Oredson
Nor can we count several flavors of unix
(unix is unix, right? just like DOS is DOS)
So I get four operating systems: unix, DOS, OS/2, Windows
That's four for Pentium.
For PPC I count zero.
For Alpha I count two.
For HP-PA I count three.
Anyone want to count SPARC, MIPS, Rx000, etc. etc. ?
Let us not forget 29k, 680x0, et al.
Time for a table or spreadsheet anyone?
--
Disclaimer: No claiming, No claiming!
Hank Oredson @ Mentor Graphics
Internet : hank_o...@mentorg.com
Amateur Radio: W0...@W0RLI.OR.USA.NA
Jim Smith
wrote:
It's a minor point, but as long as you're throwing numbers around:
$995 - $450 = $545
--
Jim Smith
j...@netcom.com
smit...@tandem.com
Keith Dow
computers with Intel microprocessors. There are about
10 million Macs with Motorola microprocessors. Lastly,
there are about 1 million Unix machines with RISC
microprocessors.
We buy computers for the software, not the hardware.
Therefore, whatever chip either has a large installed
base (Intel) or provides good price/performance (Risc)
has a possibility of doing well, provided that the
chip runs the right software.
Since Big Bill has ported his Windows-NT software to
the MIPS risc architecture, the battle in the future
may be between MIPS chips, Motorola power pc chips,
and Intel chips. Probably in three years, all
popular operating systems will be running on all
popular microprocessors. Only at that point will
price/performance be a huge determinant. Till then,
Intel will do well just because of the installed
base.
The great news is that the microprocessor power curve
has improved. Every year the power of microprocessors
will double because of risc. Intel will try to keep
up by making their CISC chip as Risc like as possible.
I personally, am looking forward to buying a 620
based Mac. The machine will be 100 times faster
than my IIci. I can live with that for two years.
Take care
--
Keith Dow A silicon valley guy.
Give me a Macintosh or give me death.
"Ditto? Ditto? You provincial putz!" Harvey Corman in Blazing Saddles
Graham K. Johnson
>Nor can we count several flavors of unix
>(unix is unix, right? just like DOS is DOS)
>
>So I get four operating systems: unix, DOS, OS/2, Windows
1. If we're talking OS's, you meant "Windows NT", not just Windows
(another year, probably, before you can add Windows itself)
2. You seem to have forgotten NeXTStep. (Mach kernel emulating UNIX?)
3. UNIX is _not_ UNIX; for example, Solaris has many quite important
features like multithreading, that other UNIXes don't have.
4. Isn't Novell Netware an operating system?
>That's four for Pentium.
Five, at least.
>For PPC I count zero.
Since IBM will demonstrate AIX for the PPC in two weeks, you can
surely count at least one, anyway. Still, it's pretty silly to argue
that a platform that _isn't available yet_ doesn't _currently_ have
this or that OS.
Dan Hildebrand
Russ Taylor <rta...@cie-2.uoregon.edu> wrote:
>
>Bill. Guess what. Parts are a SMALL part of the cost of a computer. For
>the final shock to the consumer, double or triple the CPU price difference.
>Take some basic economics and learn how raising the raw material price by
>$50 translates into a greater price increase down the line.
>
>Or are PC fans just not capable of understanding pricing?
Price out what you pay for a motherboard without a CPU.
Price out what you pay for that same motherboard with a CPU installed.
Compute the mark up on the CPU. Double or triple? Nope.
Ishir Bhan
>Pentium machines already sell for $2495.
The only one I've seen at that price had something like no hard
disk/monitor and an ISA backplane (ugh). To make it into a full system
would raise the price considerably (the system I saw advertised was from ALR).
--
-----------------------------------------------------------------------------
Ishir Bhan | Apple Newton:
ib...@husc.harvard.edu | I saw it...I did it...I want it...
-----------------------------------------------------------------------------
Ishir Bhan
>Since I included NeXTStep, an advanced object-oriented windowing OSs,
>I beg to differ. (Also, as a minor nit, the list is 9+ OSs, counting
>"more variants of UNIX than I have fingers" as "1+".) What could be
>more varied than a box that runs both not-much-more-than-a-file-handler
>MS-DOS _and_ NeXTStep486 ? That's like a car that runs on anything from
>banana peels and beer to plutonium and tritium (movie reference :-).
The problem with NeXTStep is that you can't use it on just any old 486 or
Pentium. There aren't too many hardware drivers released yet. There's
also not a great deal of software out for NeXTStep. It is a nice OS,
though. I also believe that a PowerPC port is in the works.
Dan E Babcock
>In article <7k3...@quantum.qnx.com>,
>Dan Hildebrand <da...@quantum.qnx.com> wrote:
>>In article <aa229-260...@192.139.46.27>,
>>Steve Birnbaum <aa...@freenet.carleton.ca> wrote:
>>>
>>>Yes, let's talk price. The PPC will sell for about $2000 and the Pentium
>>>for what, oh, about $5000?
>>
>>Pentium machines already sell for $2495.
>
>And are basically just '486 machines with a Pentium thrown in. They
>don't take full advantage of the Pentium, and would be blown away by a
>$2495 PPC machine.
I don't know what you mean by "don't take full advantage of...".
Historically PC vendors have been happy to spend extra money on
external caches to extract an extra 3-10% of performance, whereas
Apple has not (for good reason - Apple doesn't have to compete with
100 other companies offering virtually identical machines).
I am pretty sure that the ALR Pentium systems, which are
the cheapest, are not retrofit 486 designs. The el-cheapo systems will
come later, when the Pentium is in good supply.
Dan
Miguel Cruz
>
>aa...@freenet.carleton.ca (Steve Birnbaum) writes:
>] In article <DOCONNOR.93...@potato.sedona.intel.com>,
>] doco...@sedona.intel.com (Dennis O'Connor) wrote:
>]
>] > This was true years ago, and may not be true anymore.
>] > Market readiness depends on price, performance and
>] ^^^^^
>]
>] Yes, let's talk price. The PPC will sell for about $??? and the Pentium
>] for what ....
>Dennis O'Connor doco...@sedona.intel.com
>Intel i960(R) Microprocessor Division Solely responsible for what I do.
What's this crap, Dennis? What's the sleazy excuse for changing (as
opposed to editing for brevity) someone's original text in your quote?
Let's have some ethics here.
miguel
Tim Olson
| Now to the query part. As a RISC chip, doesn't the PPC require more off-CPU
| support, such as a larger L2 cache, primary memory (more instructions
required
| to perform equivalent tasks) and maybe more things? I would also guess that
this
| extra cache and memory might need to be of a higher performance type also,
given
| the CPU's dependence upon it (I am assuming that a RISC chip might be more
| dependent upon the system design for performance optimization than a CISC
chip.
| Please let me know if this assumption is baseless).
Actually, for the PowerPC-601 and the Pentium, it's the other way around. When
these processors are used in systems without L2 caches, the 601 performs much
better than the Pentium, due to its much larger on-chip L1 cache (32KB unified
vs. 8K/8K) -- i.e. Pentium requires a larger, higher-performance L2 for the
same system performance.
--
-- Tim Olson
Apple Computer Inc. / Somerset
Bruce Grubb
wrote:
> people keep quoting will probably be a 50Mz PPC w/ 4Meg ram and 80Meg HD
> no montior and no keyboard.
PowerPC Rumors
IBM PowerPC: rumored delayed until 1Q 1994 (Digital News & Review 06/07/93
vol 10 n11); ~$3000 - MicroChannel bus, XGA video, 8 MB RAM, 200 MB hard
drive (Computer Reseller News, 3/22/93). Will run native version AIX and
Mac apps (PC Week 3/15/93) and support Windows/DOS.
Apple PowerPC: March 14, 1994; ~$2,000 - Centris 610-like box {PDM} with
MPC601/50 MHz, 8 MB RAM, 160 MB hard drive; ~$3,000 - Centris 650-like box
{Carl Sagan} MPC601/66 MHz, 8 MB RAM, 230 MB hard drive, 3 NuBus 90 slots;
~$4,000 - {Cold Fusion} MPC601/66 MHz, 8 MB RAM, 230 MB hard drive.
Options: All - CD-ROM; PDM - 230MB hard drive; Carl Sagan - 500 MB
Hard Drive; Cold Fusion - 16 MB RAM, 500 MB/1GB Hard Drive (PC Week 08/30/93)
The Centris 610/660av/650, IIvx, IIvi, and Quadra 800/840av are all planned
to have PowerPC upgrades (PC Week 5/10/93; Apple Computer). MPC603 and
MPC604 Mac planned by late 1994-early 1995 (MacWeek 08/09/93).
The "$2000 PPC machine that people keep quoting" is a MPC601/50 MHz, 8 MB RAM,
160 MB hard drive {likely without the montior keyboard}.
This is a substantial price drop from May projections which *were*
of a "50Mz PPC w/ 4Meg ram and 80Meg HD no montior and no keyboard"
By the way developers are going to get prototypes of the PowerPC Macs in
September 1993 (PC Week 08/30/93). Which meens that by March 1994
Apple will have the potential of having NATIVE CODE programs ready
for the new machine {hence the delay from the Jan 1994 target date}
Apple has also learned from past FUBARS and will use the two months
to make sure that the PowerPC macs are not in short supply {as with
other models such as the IIsi, Quadra 800, and the LC line}
A
A
`