Win95, Fast Pentium PCs - are we in trouble?
Marciano Siniscalchi
out to buy one of those cheap-but-fast Pentium PCs:
The Facts:
Yesterday evening I friend of mine who happens to be a Win95 beta tester
invited me to his place. Being an avid Mac fan and relentless PC basher, I
just could not resist and asked him to show me how the *thing* really
looked like and worked, the idea being to see it crash and burn before my
eyes.
Well, it did not happen. I hasten to add my friend has a cheap but really
fast Pentium-90 PC with 16Meg, a 64-bit video card (fast, but not
accelerated for Windows -- e.g. no block moves), a screaming fast 2 Gig HD
and a gorgeous 17" Trinitron monitor -- plus a 28.8K modem.An impressive
system overall.
Win95 performed absolutely flawlessly. It launched Word and Excel in
approx 2/3 seconds, while playing a couple of movies in 2 simultaneous
QT-like windows (audio, too) with no apparent performance degradation. I
launched MSN, opened a few windows, etc.: nothing bad happened. Netscape
also worked like a charm -- my friend also showed me a neat app that
receives audio from Web servers and plays it live (do we have anything
like that on the Mac?).
I cannot say I performed a thorough test of the OS's capabilities,
stability etc., but purely from a (relatively sophisticated) user's
perspective, it seems to work more or less as advertised.
Is it good looking? Well, that's subjective. Overall, I would say the GUI
is closer to OS/2 than to the Mac, but you can drop files and folders on
the desktop, you have long file names (anthough I don't know how this
works with Win3.1 apps). In general, it is *much* better than Win3.1 --
definitely worth the upgrade if one is committed to Windows and has an
adequate machine.
So here's my point. To the average user Win95 will "look like the Mac".
It's prettier than Windows, and intuitive enough to use (things work more
or less like they should, i.e. as on a Mac :-) ). Of course, "average
users" may not be expected to own Pentium-90 PC's with super-fast video
cards & HD's, etc., although PC prices are dropping like crazy and will
continue to do so.
What's worse, the average user will have no particular reason to buy a Mac
instead. I mean it -- especially if all one wants to run on her/his
computer is Word and Excel (I don't). The superiority of the Mac's
interface becomes questionable once Win95 ships (well...). As for speed,
well, none of the PowerMacs I've seen at the Stanford Bookstore seemed as
zippy as my friend's clone box :-( .
SO what is a passionate Mac *lover* to do? Sigh...
Marciano
cricket
> receives audio from Web servers and plays it live (do we have anything
> like that on the Mac?).
Since you're a mac-lover, you ought to know that the RealAudio is also
available for the Mac at http://www.realaudio.com/ ... it's works pretty
well over a 14.4 and excellent over a 28.8 ...
:)
cricket
+___________________________________________________________________+
| The trees all waved their giant arms ... |
| And happiness bled from every street corner ... |
| And biplanes bombed with fluffy pillows. |
| _____________________________________ |
| cri...@cybernetics.net |
+___________________________________________________________________+
Sangria
marc...@gsb-ecu.stanford.edu says...
>Well, it did not happen. I hasten to add my friend has a cheap but really
>fast Pentium-90 PC with 16Meg, a 64-bit video card (fast, but not
>accelerated for Windows -- e.g. no block moves), a screaming fast 2 Gig HD
>and a gorgeous 17" Trinitron monitor -- plus a 28.8K modem.An impressive
>system overall.
Christ!
Cheap???
17" Tiritron is what, $800 - $1000?
Ok, you're definition of cheap...
Me, I perfer "cheap workstation caliber".
>It launched Word and Excel in
>approx 2/3 seconds,
Let me guess, either an integrated PCI FastSCSI controller
or a bus master PCI FastSCSI controller yes?
>my friend also showed me a neat app that
>receives audio from Web servers and plays it live (do we have anything
>like that on the Mac?).
Umm, would you happen to know what this app is called?
Sounds like something I would like to get a copy of....
>As for speed,
>well, none of the PowerMacs I've seen at the Stanford Bookstore seemed as
>zippy as my friend's clone box :-( .
More than likely it's due to the poor disk I/O of the PMacs.
Compared to a PCI FastSCSI controller, the disk subsystem of
most Macs are pretty weak. It's only when comparing to
ISA IDE that Mac disk I/O looks good...
>SO what is a passionate Mac *lover* to do? Sigh...
Wait.
If you can still work with your PowerBook, there is no need to
trash it. If you absolutely must have something today, then
look at your software base first before going out and plunking
$3000 or so on a new machine.
Determine what software you need for you task, then the OS for
those software, and _then_ go and look for the best hardware
you can get for your needs (remember this doesn't necessarily
mean the most expensive...buying a dual P5-120 box with 64MB
to do word processing is a _slight_ overkill... :-)
Also, once you figure out the software base, look at both
OS/2 and NT before making a decision.
-- Sang.
*************************************************************
* Sang K. Choe san...@inlink.com *
* http://www.inlink.com/~sangria/index.html *
*************************************************************
Marciano Siniscalchi
>
> Christ!
> Cheap???
> 17" Tiritron is what, $800 - $1000?
> Ok, you're definition of cheap...
> Me, I perfer "cheap workstation caliber".
I completely agree. However, I doubt the whole thing cost him more than a
7100/80, say (including the monitor, of course). PC boxes can be really
cheap if you know where to buy them (monitors are different of course).
>
> >It launched Word and Excel in
> >approx 2/3 seconds,
>
> Let me guess, either an integrated PCI FastSCSI controller
> or a bus master PCI FastSCSI controller yes?
Yes, I think he has a FastSCSI controller. I wonder why the Mac's I/O
sucks so badly?
>
> >my friend also showed me a neat app that
> >receives audio from Web servers and plays it live (do we have anything
>
> Umm, would you happen to know what this app is called?
> Sounds like something I would like to get a copy of....
>
It's called RealAudio, and I just learned that they have a Mac version,
too!!! :-)
Thanks for your comments.
Marciano
Stephen Carter Morgan
>Well, it did not happen. I hasten to add my friend has a cheap but really
>fast Pentium-90 PC with 16Meg, a 64-bit video card (fast, but not
>accelerated for Windows -- e.g. no block moves), a screaming fast 2 Gig HD
>and a gorgeous 17" Trinitron monitor -- plus a 28.8K modem.An impressive
>system overall.
I believe all 64-bit graphics cards are accelerated for Windows. Some just
accelerate different primitives. Get your friend to run MSD or equivalent or
just ask him what graphics chipset is on his card. There are several 64-bit
64-bit graphics accelerator chipsets out there. The common ones (in approx.
order of performance) are Cirrus Logic 5434, S3 Trio64, S3 864, Ati Mach 64,
and S3 964. The Mach 64 can use either DRAM or VRAM. DRAM cards perform
similarly to S3 864, and the VRAM cards perform similarly to S3 964. The
new S3 868 and S3 968 add digital video playback acceleration to the S3 864 and
S3 964, respectively.
Cheap? Well...
Mid-tower case and 230W power supply = $75
P-90 motherboard with PCI+ISA, PCI EIDE, Floppy controller, IO = $775
3.5" 1.44MB floppy drive = $35
PCI Adaptec 2940 SCSI-2 controller = $230 (some boards can also use the $90
NCR PCI SCSI-2 controller for booting.)
PCI 64-bit graphics card (assume 2MB, since that is needed for maximum
performance on 64-bit cards) - $150 for CL 5434, $180 for Trio64, $220 for
S3-864, $285 for S3-964, ? for Ati cards. Prices quotes are for name
brand cards. Generics are slightly cheaper. I'll assume $285 for the 964.
16 MB RAM (2 8MB SIMMs) = $590
Nice PC keyboard = $30. (That's right - a nice one. The cheap ones are $15)
I believe that's $1920. (My calculator and pencil and paper are not currently
available.) I'm not quite up on prices for 2MB SCSI-2 drives. Is $900
reasonable?
$40 for Microsoft Mouse
$90 for SB16
$200 for Toshiba 4X EIDE CD-ROM with 195 ms access.
$110 for Intel Ethernet card.
Total: $3260 or so. These are retail prices for the parts. A System price
can be considerably cheaper. For example, Micron sells P-100s with 256K cache,
16MB EDO RAM, 850MB HD, 64-bit graphics accelerator, SB16, 4X CD-ROM keyboard,
mouse, speakers, 15" 1280x1024 monitor with .28 dot pitch, DOS, Windows, and
Microsoft Office Professional for $2999. (BTW, Micron is one of the top-rated
PC manufacturers and has won several Editor's Choice awards from PC Magazine and
other magazines due to their systems' reliability, top-notch performance,
very reasonable prices, and good technical support. In a reader satisfaction
survey, they were the top-rated PC vendor, beating IBM, Compaq, Digital
Equipment Corportation, Gateway, and other vendors.
A PowerMac 7100/80 with 8MB,730MB HD, and 2X CD-ROM is $3050 at mass-market
retail stores like CompUSA and Computer City. No keyboard, no monitor.
--
Stephen Carter Morgan
Georgia Institute of Technology, Atlanta Georgia, 30332
uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!gt4148b
Internet: gt4...@prism.gatech.edu
Shimpei Yamashita
Stephen Carter Morgan <gt4...@prism.gatech.edu> wrote:
:marc...@gsb-ecu.stanford.edu (Marciano Siniscalchi) writes:
:
:>Well, it did not happen. I hasten to add my friend has a cheap but really
:>fast Pentium-90 PC with 16Meg, a 64-bit video card (fast, but not
:>accelerated for Windows -- e.g. no block moves), a screaming fast 2 Gig HD
:>and a gorgeous 17" Trinitron monitor -- plus a 28.8K modem.An impressive
:>system overall.
:Mid-tower case and 230W power supply = $75
:P-90 motherboard with PCI+ISA, PCI EIDE, Floppy controller, IO = $775
:3.5" 1.44MB floppy drive = $35
:PCI Adaptec 2940 SCSI-2 controller = $230 (some boards can also use the $90
:NCR PCI SCSI-2 controller for booting.)
:PCI 64-bit graphics card (assume 2MB, since that is needed for maximum
:performance on 64-bit cards) - $150 for CL 5434, $180 for Trio64, $220 for
:S3-864, $285 for S3-964, ? for Ati cards. Prices quotes are for name
:brand cards. Generics are slightly cheaper. I'll assume $285 for the 964.
:16 MB RAM (2 8MB SIMMs) = $590
:Nice PC keyboard = $30. (That's right - a nice one. The cheap ones are $15)
:I believe that's $1920. (My calculator and pencil and paper are not currently
:available.) I'm not quite up on prices for 2MB SCSI-2 drives. Is $900
:reasonable?
:$40 for Microsoft Mouse
:$90 for SB16
:$200 for Toshiba 4X EIDE CD-ROM with 195 ms access.
:$110 for Intel Ethernet card.
:
:Total: $3260 or so. These are retail prices for the parts.
I might point out that the above doesn't include a monitor....
--
Shimpei Yamashita, Stanford University shi...@leland.stanford.edu
Stephen Carter Morgan
I wrote:
>:Total: $3260 or so. These are retail prices for the parts.
Space saver: That's my estimate for building a P90 with 256K cache, 16MB RAM,
Adaptec PCI SCSI-2 controller, 2GB SCSI-2 HD, PCI Stealth 64 VRAM with 2MB,
16-bit sound card, Toshiba 5302 4X IDE CD-ROM with 195ms access, keyboard,
mouse, and ethernet from parts, buying retail. Incidentally, I just took a
look at the new price lists. There is now $100 difference between a similarly
configured P-90 and a P-100. There's also only $100 difference between the
P-75 and P-90 versions.
>I might point out that the above doesn't include a monitor....
Neither does the ~$3050 PowerMac 7100/80 with 8MB, 730MB HD, and 2X CD-ROM in
Computer City and CompUSA. As a matter of fact, that price doesn't include a
keyboard, either.
Robert C. Barris
>
>>:Total: $3260 or so. These are retail prices for the parts.
>Space saver: That's my estimate for building a P90 with 256K cache, 16MB RAM,
>Adaptec PCI SCSI-2 controller, 2GB SCSI-2 HD, PCI Stealth 64 VRAM with 2MB,
>16-bit sound card, Toshiba 5302 4X IDE CD-ROM with 195ms access, keyboard,
>mouse, and ethernet from parts, buying retail. Incidentally, I just took a
>look at the new price lists. There is now $100 difference between a similarly
>configured P-90 and a P-100. There's also only $100 difference between the
>P-75 and P-90 versions.
>
>>I might point out that the above doesn't include a monitor....
>
>Neither does the ~$3050 PowerMac 7100/80 with 8MB, 730MB HD, and 2X CD-ROM in
>Computer City and CompUSA. As a matter of fact, that price doesn't include a
>keyboard, either.
You know, you're right. Maybe I should look for a cheaper Mac. Hey, hang on
a sec...
$3049 can also buy you a Power Computing Power-100.
100-MHz PowerPC 601, 256K cache
16MB DRAM
1GB HD
twin SCSI-2
twin video display (DRAM video, plus 64-bit local bus 2MB VRAM card)
Ethernet (does not include $35 transceiver for 10BaseT or coax.)
dual DMA serial/GeoPort
16bit audio in/out
keyboard, mouse, baby AT case.
software bundle (ClarisWorks, Quicken, HD/CDROM toolkit, big font collection)
3 free slots!
1yr warranty on entire system.
I will note that a CDROM is not included in this config. But, I do
believe it goes to show that you can indeed buy a Mac compatible for
substantially lower than Apple pricing. One dollar less than your CompUSA
number (which might just not be the best price in the world, but that's
*.advocacy for ya), but with more CPU, double the RAM, more hard drive,
keyboard, and a decent software bundle.
[This is the machine I ordered about three weeks ago, it finally shipped
this last Friday, yay! I'll try to post a first impressions next week. ]
I think the real price crunch will happen this summer as PCI takes hold
in the Mac-compatible market. That Stealth 64 you mentioned, I believe
that's a Diamond? (one of the board makers having announced Mac-PCI
support)
Rob Barris
Quicksilver Software Inc.
rba...@quicksilver.com
M. Wade Heninger
(Stephen Carter Morgan) wrote:
> 16MB EDO RAM, 850MB HD, 64-bit graphics accelerator, SB16, 4X CD-ROM keyboard,
> mouse, speakers, 15" 1280x1024 monitor with .28 dot pitch, DOS, Windows, and
> Microsoft Office Professional for $2999. (BTW, Micron is one of the top-rated
> PC manufacturers and has won several Editor's Choice awards from PC
Magazine and
> other magazines due to their systems' reliability, top-notch performance,
> very reasonable prices, and good technical support. In a reader satisfaction
> survey, they were the top-rated PC vendor, beating IBM, Compaq, Digital
> Equipment Corportation, Gateway, and other vendors.
>
> A PowerMac 7100/80 with 8MB,730MB HD, and 2X CD-ROM is $3050 at mass-market
> retail stores like CompUSA and Computer City. No keyboard, no monitor.
>
Of course you can get a similarly configured Power 100 for around the 3000
level...and you get two benefits: it is faster than the Pentium 90 and it
comes with a massive load of useful software like ClarisWorks (which BTW
is what MS office should be).
____________________________________________________________________
M. Wade Heninger |
801.375.7018 | A 100% Microsoft Free Macintosh
benn...@yvax.byu.edu |
____________________________________________________________________
Macintosh: "The gateway, not the Gates way, to Plug & Play Internet"
M. Wade Heninger
> 16MB EDO RAM, 850MB HD, 64-bit graphics accelerator, SB16, 4X CD-ROM keyboard,
> mouse, speakers, 15" 1280x1024 monitor with .28 dot pitch, DOS, Windows, and
> Microsoft Office Professional for $2999. (BTW, Micron is one of the top-rated
> PC manufacturers and has won several Editor's Choice awards from PC
Magazine and
> other magazines due to their systems' reliability, top-notch performance,
> very reasonable prices, and good technical support. In a reader satisfaction
> survey, they were the top-rated PC vendor, beating IBM, Compaq, Digital
> Equipment Corportation, Gateway, and other vendors.
>
> A PowerMac 7100/80 with 8MB,730MB HD, and 2X CD-ROM is $3050 at mass-market
> retail stores like CompUSA and Computer City. No keyboard, no monitor.
>
Of course you can get a similarly configured Power 100 for around the 3200
Stephen Carter Morgan
>In article <3q9dcu$i...@acmex.gatech.edu>, gt4...@prism.gatech.edu
>(Stephen Carter Morgan) wrote:
>> 16MB EDO RAM, 850MB HD, 64-bit graphics accelerator, SB16, 4X CD-ROM keyboard,
>> mouse, speakers, 15" 1280x1024 monitor with .28 dot pitch, DOS, Windows, and
>> Microsoft Office Professional for $2999. (BTW, Micron is one of the top-rated
>Of course you can get a similarly configured Power 100 for around the 3000
>level...and you get two benefits: it is faster than the Pentium 90 and it
>comes with a massive load of useful software like ClarisWorks (which BTW
>is what MS office should be).
Actually, the Micron system is a P-100. I'm not entirely convinced the Power
100 would be all that much faster. All the Pentium vs. PowerMac studies
quoted by Joe Ragosta and others were done on (1) early Pentium mainboards that
had slow memory access (actually slower than some 486 boards), (2) early PCI
chipsets, (3) Pentium systems that were not optimal for the CPU (The Gateway
P-90 which was the ONLY P-90 used in one survey is actually beat in Winstone
benchmarks by a Zeos Pantera P-66), and (4) systems with ordinary DRAM and cache
rather than EDO RAM and burst-mode cache (together, EDO RAM and burst-mode
cache add 10-15% performance.) Right now, the Micron P-100 is either the
fastest P-100 system sold by a major vendor or almost tied for it. It IS
faster than the machines sold by IBM and Hewlett-Packard....
As for MS Office Professional vs. ClarisWorks, one sells for $500 retail, and
another is $90. I'm sure someone would be willing to buy MS Office Pro from
you for $175 or so so you can get ClarisWorks. For a while, Micro Center was
throwing in ClarisWorks for Windows for $25 with a system purchase....
If I went with a local dealer and got a system with same specs but ordinary
DRAM and MS Works (closer to ClarisWorks) and MS Encarta as the software along
with DOS and Windows I could do a lot better. HiQ's Family PC P-100 with 8MB
RAM, 850MB HD, PCI graphics with 2MB, PCI Enhanced IDE, Soundblaster 16,
14.4K Fax/Modem with voice mail, 4X CD-ROM, 1.44MB floppy, speakers, keyboard,
MS Mouse, and 15" .28 dot pitch non-interlaced monitor with digital controls
is now $2299. Add $285 to upgrade it to 16MB RAM and $150 to bring the
graphics to the Stealth64 VRAM with 2MB found in the Micron system to bring the
cost to $2739. And the HiQ system has a 14.4K Fax/Modem. They are viewed as
being a very reliable local dealer and have a 2 years labor, 1 year parts
warranty and toll-free tech support (and probably have someone on the other
end who actually knows what they're doing, possibly even the person who
built your system.)
Robert C. Barris
>100 would be all that much faster. All the Pentium vs. PowerMac studies
You sure have a surprisingly detailed list of deficiencies in previously
tested systems at your fingertips though.
>If I went with a local dealer and got a system with same specs but ordinary
>DRAM and MS Works (closer to ClarisWorks) and MS Encarta as the software along
>with DOS and Windows I could do a lot better. HiQ's Family PC P-100 with 8MB
>RAM, 850MB HD, PCI graphics with 2MB, PCI Enhanced IDE, Soundblaster 16,
>14.4K Fax/Modem with voice mail, 4X CD-ROM, 1.44MB floppy, speakers, keyboard,
>MS Mouse, and 15" .28 dot pitch non-interlaced monitor with digital controls
>is now $2299. Add $285 to upgrade it to 16MB RAM and $150 to bring the
>graphics to the Stealth64 VRAM with 2MB found in the Micron system to bring the
>cost to $2739. And the HiQ system has a 14.4K Fax/Modem. They are viewed as
I don't understand how IBM, Gateway, Dell, Compaq, AST and Packard Bell
can possibly stay in business with guys like HiQ around. Could it
possibly indicate that there's a range of consumers to sell to, with
different requirements and different budgets and tastes?
A few months ago: "Why buy a PowerMac, a Dell 100-MHz is twice as fast
for many hundreds less."
[Power Computing ships Mac clones, cost 100's less [got ours today!] ]
Now: "Well, I know a guy that can make a Pentium system pretty cheap."
[an admitted exaggeration to make a point]
It sounds like progress to me.
M. Wade Heninger
(Stephen Carter Morgan) wrote:
> benn...@yvax.byu.edu (M. Wade Heninger) writes:
>
> >In article <3q9dcu$i...@acmex.gatech.edu>, gt4...@prism.gatech.edu
> >(Stephen Carter Morgan) wrote:
>
> >> 16MB EDO RAM, 850MB HD, 64-bit graphics accelerator, SB16, 4X CD-ROM
keyboard,
> >> mouse, speakers, 15" 1280x1024 monitor with .28 dot pitch, DOS,
Windows, and
> >> Microsoft Office Professional for $2999. (BTW, Micron is one of the
top-rated
>
> >Of course you can get a similarly configured Power 100 for around the 3000
> >level...and you get two benefits: it is faster than the Pentium 90 and it
> >comes with a massive load of useful software like ClarisWorks (which BTW
> >is what MS office should be).
>
> Actually, the Micron system is a P-100. I'm not entirely convinced the Power
> 100 would be all that much faster. All the Pentium vs. PowerMac studies
> quoted by Joe Ragosta and others were done on (1) early Pentium
mainboards that
> had slow memory access (actually slower than some 486 boards), (2) early PCI
> chipsets, (3) Pentium systems that were not optimal for the CPU
And of course, those PPC machines also had an early designs that had
similar problems...so your point is moot. I believe it is generally
acknowledged that mhz for mhz, a PPC 601 will beat a pentuim of the same
clock speed.
Then cometh the 604 PCI macs (READ: in 15 days or less) and there will be
even a greater speed difference. Again apple will claim "the fastest
personal computer" sticker on their forehead...at least for a few months.
Damir Smitlener
[...snip...]
> If I went with a local dealer and got a system with same specs but ordinary
> DRAM and MS Works (closer to ClarisWorks) and MS Encarta as the software along
> with DOS and Windows I could do a lot better. HiQ's Family PC P-100 with 8MB
> RAM, 850MB HD, PCI graphics with 2MB, PCI Enhanced IDE, Soundblaster 16,
> 14.4K Fax/Modem with voice mail, 4X CD-ROM, 1.44MB floppy, speakers, keyboard,
> MS Mouse, and 15" .28 dot pitch non-interlaced monitor with digital controls
> is now $2299. Add $285 to upgrade it to 16MB RAM and $150 to bring the
> graphics to the Stealth64 VRAM with 2MB found in the Micron system to
bring the
> cost to $2739. And the HiQ system has a 14.4K Fax/Modem. They are viewed as
> being a very reliable local dealer and have a 2 years labor, 1 year parts
> warranty and toll-free tech support (and probably have someone on the other
> end who actually knows what they're doing, possibly even the person who
> built your system.)
HiQ is actually a nationwide outfit, with outlets in numerous (well,
several) cities. They've even won some PC Mag Editor's choice awards.
--
damir smitlener |
da...@mindspring.com |
smi...@optica.mirc.gatech.edu |
Stephen Carter Morgan
>You sure have a surprisingly detailed list of deficiencies in previously
Most of the studies quoted here (like Ingram) were done at least a year ago.
Early Pentium mainboards did have slow memory transfer rates. And Triton
chipset Pentium systems with EDO RAM and Burst-mode cache have only been
available for 2 months or less. A Triton with EDO RAM and burst-mode cache is
already 10-15% faster than the Neptune (which Triton replaces) with ordinary
DRAM and cache. Triton chipset Pentium mainboards now have bus-mastering IDE
(!) supporting up to mode 4 for an additional increase in performance of systems
with Enhanced IDE hard disks. In the latest PC Magazine review, a system with
fast/wide Adaptec SCSI-2 had only AVERAGE disk data transfer rates. The
fastest rates were with the Triton chipset boards using the on-board IDE
interface and EIDE hard disks supporting mode 4 PIO, like the 1.6 GB WD drive.
>I don't understand how IBM, Gateway, Dell, Compaq, AST and Packard Bell
>can possibly stay in business with guys like HiQ around.
HiQ only has branches in 4 or 5 cities. Atlanta is also a VERY competitive
market for systems from local dealers (There are about 30 in the area, and all
advertise in the magazine Atlanta Computer Currents, which is widely available
for free. Besides, HiQ's manufacturing capacity is limited to the 3 or 4
people working in the back room of the shop. You fill out the sheet as to what
you want, and then pick up your system 2 days later.... These kinds of stores
have very slim margins for system sales, getting most of their margin from
parts prices (although they do have better prices than most mail-order outfits,
once you thow in the shipping costs.) The situation might be different in
other cities, and smaller cities might not even have any local computer
manufacturers.
Besides, Packard Bell systems are slightly cheaper, at least until the first
repair bill. I've heard the sales guy say point-blank to a customer that he
can't match the price on a Packard Bell system, but his systems are much more
upgradeable and would be much cheaper to repair if something did go wrong. Of
course, the same thing could be said about the current IBM and Compaq systems,
but they have a name-brand status that enables them to sell their systems at a
higher price.
Could it
>possibly indicate that there's a range of consumers to sell to, with
>different requirements and different budgets and tastes?
Yes. Some people want their computer to come from a large company. And Joe
Average just walks into Computer City, CompUSA, or Best Buy to see what they
have... (although Computer City is one of the places that distributes Atlanta
Computer Currents, even though it seem to me that that is counter to their own
interests.)
-
Sangria
>I don't understand how IBM, Gateway, Dell, Compaq, AST and Packard Bell
>can possibly stay in business with guys like HiQ around. Could it
>possibly indicate that there's a range of consumers to sell to, with
>different requirements and different budgets and tastes?
No it's called corporate sales.
Right now our company buys from Compaq and IBM. Why? Because
both are large enough company that there is a direct and well
defined line of accountability. If something goes wrong, we know
who we can point our fingers at--which makes getting the problem
fixed real easy.
Smaller vendors don't have the same sales numbers as Compaq and
IBM simply because they can't get their foot in to doors of
major corporations. However, this does not mean those machines
are substandard. There is less accountability and maybe less
support, but the quality of the machines are pretty leveled.
We often see DOAs of Compaqs, IBMs, Apples, HPs etc...
>A few months ago: "Why buy a PowerMac, a Dell 100-MHz is twice as fast
>for many hundreds less."
It's still true.
The price of a 100Mhz PowerMac clones are at the same level as a
P5-120 these days. Using the new generation chip set + the new
EDO DRAMs, these puppies are incredibly fast.
Also the prices of these machines also dropped another big chunk.
This week, Intel annouced another major price drop in their P5
line up. The 100Mhz dropped something like >30% and the 120Mhz
dropped another 21%.
The Micron Millenia Plus 120, is a system completely decked out.
(And I mean decked as in you don't need to change a damn thing
to make it into a workstation class machine). And it costs
around $4000. And this includes everything you can want except
an ethernet card. It even has a 6x CD-ROM!
Drop down to a P5-100 Millenia, and you can get a system for
about $2500...not as decked out as the 120, but close.
Stephen Carter Morgan
>In article <3qgieg$m...@acmey.gatech.edu>,
>Stephen Carter Morgan <gt4...@prism.gatech.edu> wrote:
>>Actually, the Micron system is a P-100. I'm not entirely convinced the Power
>>100 would be all that much faster. All the Pentium vs. PowerMac studies
>These are very competitive systems. If you want to split hairs, that is OK.
>You sure have a surprisingly detailed list of deficiencies in previously
>tested systems at your fingertips though.
These were mentioned in the February _MacWorld_ issue on p. 43 when they did
a comparison between a Micron PowerStation P-90 and an Apple PowerMac 8100/110.
They were given as reasons why the PowerStation held its own in that test,
while MacWorld's earlier test of a P-90 system (June 1994) found it to run only
at the speed of a PowerMac 7100/66. The P-90 was running DOS 6.22/Windows for
Workgroups 3.11 and the PowerMac 8100/110 was running System 7.5.
The PowerStation was/is a system with Neptune chipset, ordinary DRAM, and
ordinary cache, making it slower than a Millennia system at the same clock
speed. For $300, you can go from the P-100 system I gave the price for ($2999)
to a P-120. (These have 16MB EDO RAM, 256K cache, 850MB HD, 64-bit graphics
accelerator with 2MB VRAM, 16-bit soundcard, 4X CD-ROM, keyboard, mouse, 15"
monitor, DOS, Windows 3.11, and MS Office.)
The numbers: (PowerStation P-90 time (s), PowerMac 8100/110 time (s)))
Adobe PageMaker 5.0
Open file 4.8, 9.6
Change font 8.9, 7.6
Adobe PhotoShop 3.0
Open file 43.1, 30.7
Unsharp mask 2.8, 3.4
Convert to CMYK 3.3, 1.8
Fractal Design Painter 2.0
Adjust dye concentration 1.6, 4.9
Apply surface texture 7.0, 10.8
MicroSoft Excel 5.0
Scientific recalculation 2.6, 4.1
Integer recalculation 0.9, 1.0
Scroll 15.8, 14.5
QuarkXPress 3.31
Scroll 6.6, 6.7
Save pages as EPS 9.9, 6.1
Print to file 46.8, 84.1
Wolfram Mathematica 2.2.2
Workbook calculation 172.7, 102.5
PhotoShop uses its own disk I/O routines to get around the emulated ones in
System 7.5. The only MAJOR win for the 8100/110 vs. P-90 was in Mathematica,
but I suppose almost everyone will agree that Mathematica is heavily optimized
for a Mac.
Of course, we all know what the results would have looked like in MS Word
6.0.....
Sangria
>Then cometh the 604 PCI macs (READ: in 15 days or less) and there will be
>even a greater speed difference. Again apple will claim "the fastest
>personal computer" sticker on their forehead...at least for a few months.
With the "second most brain-damaged" OS available on the market sticker
sitting right next to it...
David Kurtz
>In article <bennetk1-310...@128.187.250.8>, M. Wade Heninger,
benn...@yvax.byu.edu says...
>
>>Then cometh the 604 PCI macs (READ: in 15 days or less) and there will be
>>even a greater speed difference. Again apple will claim "the fastest
>>personal computer" sticker on their forehead...at least for a few months.
>
>With the "second most brain-damaged" OS available on the market sticker
>sitting right next to it...
Not to mention the "Greatest User Satisfaction Level OS," "Most productive
personal computing OS" and "Least expensive to own OS & Hardware" stickers
on the same box.
Sure, MacOS doesn't have the greatest memory-management routines... but no
other OS runs MacWrite Pro or CodeWarrior... or Bolo. It all depends on
what you want to do. Sangria obviously is in the server-OS market. He
would be foolish to buy a Mac; almost as foolish as a home, business or
edu user who wanted to integrate excellent graphics, word-processing,
communications, game playing and multimedia into one usable (key word
there) package and then went and bought OS/2 or NT, or Linux, or Windows,
or an Amiga. Not to say that any of the aforementioned don't have
individual strengths that outshine the Mac... I just have yet to see any
of them perform well in the same number of different types of genres as I
have seen the Mac.
--
David A. Kurtz | Portions (c)1995 David A. Kurtz
da...@ucla.edu | The Microsoft Network is prohibited
dku...@lightside.com | from redistributing this work in any http://lightside.com/~dkurtz | form whatsoever.
Bill G. Riel
: I've heard "bolo" mentioned int this group a number of times as being a
: rather cool game. What's so coll about this game in particular that it
: is mentioned in .advocacy?
: Shawn
Bolo is a cool networkable tank game available _only_ for Macs. It's
been around for a while, but for the longest time many people considered
it the best network game available for any platform (not sure if that
opinion still holds with stuff like Marathon out there). I believe it's
freeware (willing to be corrected here).
Check on one of the mac.games groups for more info.
--Bill
---------------------------------------------------------------------
Bill G. Riel Canadian Forest Service
br...@pfc.forestry.ca Victoria, BC
---------------------------------------------------------------------
Bob Noel
> Shawn V. Hernan (vale...@pitt.edu) wrote:
> : I've heard "bolo" mentioned int this group a number of times as being a
> : rather cool game. What's so coll about this game in particular that it
> : is mentioned in .advocacy?
>
>
> Bolo is a cool networkable tank game available _only_ for Macs. It's
> been around for a while, but for the longest time many people considered
> it the best network game available for any platform (not sure if that
> opinion still holds with stuff like Marathon out there). I believe it's
> freeware (willing to be corrected here).
Shareware, not freeware.
>
> Check on one of the mac.games groups for more info.
>
check rec.games.bolo. A Bolo FAQ is posted regularly
--
Bob Noel aka Kobyashi Maru
my views are my own not MITRE's
(why use a disclaimer when people are
too --------- to understand it?)
M. Wade Heninger
> In article <bennetk1-310...@128.187.250.8>, M. Wade Heninger,
benn...@yvax.byu.edu says...
>
> >Then cometh the 604 PCI macs (READ: in 15 days or less) and there will be
> >even a greater speed difference. Again apple will claim "the fastest
> >personal computer" sticker on their forehead...at least for a few months.
>
> With the "second most brain-damaged" OS available on the market sticker
> sitting right next to it...
As a point of curiosity, does yours have the gold sticker for first place
on your machine? :-)
Wade
____________________________________________________________________
M. Wade Heninger |
801.375.7018 |"They can drive you crazy. They really can."
benn...@yvax.byu.edu |
____________________________________________________________________
Shawn V. Hernan
I've heard "bolo" mentioned int this group a number of times as being a
rather cool game. What's so coll about this game in particular that it
is mentioned in .advocacy?
Shawn
Shawn Valentine Hernan |
The University of Pittsburgh | Dump the RICO Laws
vale...@pitt.edu |
412-624-6425 |
Ken Sherman
are in concerning the P6 and its floating point performance. Are you ready?
The P6 is about 30% faster than the P5 at the same clock speed... Intel claims
its twice as fast as a P5, but that includes the speed up due to faster
clocking speeds.
Let's let this sink in a moment, shall we. The current P5-90
is around 80 SPECfp92, is it not? The 100 MHZ 601 yields 125 SPECfp92.
The 100 MHz 604 yields 165 SPECfp92. The new Apple machines are coming out
at 120 and 133 MHZ. Can you say 604 kicks the P6s ass? Do a little math.
The 601 already outperforms the P5 and the 604 is about 1.6 X faster. The P6 is only about 1.3 X faster than the P5.
By the way the P6 is currently running at about 25 - 30 watts. The 604?
13 watts! One other point, at the moment the P6 is about $1,800.00!
:)
Ken Sherman
Edmond Underwood
>In article <3r79vb$d...@hearye.mlb.semi.harris.com>,
>predicted lousy FP performance on comp.sys.intel several times. By my
>calculations, the above data gives Specfp92 of 116 for the P6 at 100 MHz
>compared to 165 for the 604. We'll have to see what the relative clock
>speeds look like when things settle down.
>
>Of course, this also neglects the fact that you'll be able to buy a 604
>Mac in a week or so and the P6 boxes will be available about the same time
>as WIn95. (Any guesses?)
>
>Prediction #2:
>
>We'll hear a lot of messages from Wintel fanatics about how FP doesn't
>really matter because most users don't need it (remember the Pentium FP
>fiasco?), so the P6 is still just as good as the 604.
>
>--
>Regards, Joe Ragosta -- 100% Chemical and proud of it.
>
>Microsoft Network is prohibited from redistributing this work in any form,
>in whole or in part. Copyright, Joseph Ragosta, 1995. License to distribute
>this post is available to Microsoft for $1000. Posting without permission
>constitutes an agreement to these terms. Please send notices of violation
>to doc...@interramp.com and Postm...@microsoft.com
I thought you were a scientist Mr. 100% chemical. Where is the P6 at
100 MHz, I have never heard of such a thing. What about this 30%
thing. Talk about an arbitrary number. Yet you eat it up like candy.
You are more of an evangelist than anything. Stick to science. Oops,
I just noticed that this post is being cross-posted to the mac .advocates.
No wonder.
---------------------------------------------------------------------------------
Edmond Underwood
Systems Management Group
Computing & Network Services (University of Colorado)
E-mail: unde...@Colorado.Edu
Joe Ragosta
David T. Wang
: In article <3r79vb$d...@hearye.mlb.semi.harris.com>,
: ke...@mars.mlb.semi.harris.com (Ken Sherman) wrote:
: > Well, macophiles, I come bearing good news. The first numbers
: > are in concerning the P6 and its floating point performance. Are you ready?
: > The P6 is about 30% faster than the P5 at the same clock speed... Intel claims
: > its twice as fast as a P5, but that includes the speed up due to faster
: > clocking speeds.
: > Let's let this sink in a moment, shall we. The current P5-90
: > is around 80 SPECfp92, is it not? The 100 MHZ 601 yields 125 SPECfp92.
no, go look up the correct spec marks please.
: > The 100 MHz 604 yields 165 SPECfp92. The new Apple machines are coming out
: > at 120 and 133 MHZ. Can you say 604 kicks the P6s ass? Do a little math.
: > The 601 already outperforms the P5 and the 604 is about 1.6 X faster.
: The P6 is only about 1.3 X faster than the P5.
You're funny, generate nice random numbers.
: > By the way the P6 is currently running at about 25 - 30 watts.
: The 604?
: > 13 watts! One other point, at the moment the P6 is about $1,800.00!
: > :)
: >
This is really funny. Please state your sources, I'd like to look these
up, me thinks they're in error.
: > Ken Sherman
: Thanks for the news. I personally never thought the P6 was so great and
: predicted lousy FP performance on comp.sys.intel several times. By my
: calculations, the above data gives Specfp92 of 116 for the P6 at 100 MHz
: compared to 165 for the 604. We'll have to see what the relative clock
: speeds look like when things settle down.
HA, Ha, Ha, Funniest thing I've seen so far. Joe predicts numbers for a
chip that he's never seen before, but predicts it from some rules of
thumb he created. Good one Joe.
: Of course, this also neglects the fact that you'll be able to buy a 604
: Mac in a week or so and the P6 boxes will be available about the same time
: as WIn95. (Any guesses?)
Good for Mac users, but P6's will be released shortly. There's a lot of
people working hard at it in Oregon :)
: Prediction #2:
: We'll hear a lot of messages from Wintel fanatics about how FP doesn't
: really matter because most users don't need it (remember the Pentium FP
: fiasco?), so the P6 is still just as good as the 604.
Floats do matter. P6 will have decent numbers, don't want to baited into
speculating numbers along with you, but I'll make a wager with you, it'll
be , oh, slightly highers than 116 :) Have a nice day.
: --
: Regards, Joe Ragosta -- 100% Chemical and proud of it.
: Microsoft Network is prohibited from redistributing this work in any form,
: in whole or in part. Copyright, Joseph Ragosta, 1995. License to distribute
: this post is available to Microsoft for $1000. Posting without permission
: constitutes an agreement to these terms. Please send notices of violation
: to doc...@interramp.com and Postm...@microsoft.com
--
David Wang veak...@eng.umd.edu dave...@wam.umd.edu
Grad student- EE/Computer Engineering NeXT4Intel3.2/Linux/Dos(W4W)
Summer P6 guy @ Intel-Oregon. dw...@ichips.intel.com
http://www.wam.umd.edu/~davewang/ VLSI Design Automation Lab
I no speak for no Intel nor U of Md. Verstehen Sie?
Rob Barris
(David T. Wang) wrote:
> Floats do matter. P6 will have decent numbers, don't want to baited into
> speculating numbers along with you, but I'll make a wager with you, it'll
> be , oh, slightly highers than 116 :) Have a nice day.
The chip's clock rate is supposed to be higher than 100-MHz as well, a
factor Joe did not include in his SpecFP guesstimate.
Me, I want to find out where this $1800 price came from...
Mike Schmit
doc...@interramp.com (Joe Ragosta) writes:
>
>In article <3r79vb$d...@hearye.mlb.semi.harris.com>,
>ke...@mars.mlb.semi.harris.com (Ken Sherman) wrote:
>
>> Well, macophiles, I come bearing good news. The first numbers
>> are in concerning the P6 and its floating point performance. Are you
ready?
>> The P6 is about 30% faster than the P5 at the same clock speed... Intel
claims
>> its twice as fast as a P5, but that includes the speed up due to faster
>> clocking speeds.
>> Let's let this sink in a moment, shall we. The current P5-90
>> is around 80 SPECfp92, is it not? The 100 MHZ 601 yields 125 SPECfp92.
>> at 120 and 133 MHZ. Can you say 604 kicks the P6s ass? Do a little math.
>> The 601 already outperforms the P5 and the 604 is about 1.6 X faster.
>The P6 is only about 1.3 X faster than the P5.
>The 604?
>> 13 watts! One other point, at the moment the P6 is about $1,800.00!
>> :)
>>
The P6 is 14 watts, typical, 20 watts max. This includes the L2 cache.
Where did you get the $1800 price. They aren't for sale yet. Intel
has said they will cost about the same as every other x86 they
have made at introduction -- which is about $1000.
>Thanks for the news. I personally never thought the P6 was so great and
>predicted lousy FP performance on comp.sys.intel several times. By my
>calculations, the above data gives Specfp92 of 116 for the P6 at 100 MHz
>compared to 165 for the 604. We'll have to see what the relative clock
>speeds look like when things settle down.
>
>Of course, this also neglects the fact that you'll be able to buy a 604
>Mac in a week or so and the P6 boxes will be available about the same time
>as WIn95. (Any guesses?)
>
>Prediction #2:
>
>We'll hear a lot of messages from Wintel fanatics about how FP doesn't
>really matter because most users don't need it (remember the Pentium
FP
>fiasco?), so the P6 is still just as good as the 604.
For most users FP performance does not matter. For those that FP
performance really matters, they have never used x86 PC's or
switch to something else, like a Unix workstation. Virtually
every market research study about this has shown this to be true.
FP is important to 5 - 10% of the market.
Mike Schmit
-------------------------------------------------------------------
msc...@ix.netcom.com author:
408-244-6826 Pentium Processor Programming Tools
800-765-8086 ISBN: 0-12-627230-1
-------------------------------------------------------------------
Joe Ragosta
<unde...@Colorado.Edu> wrote:
>
> I thought you were a scientist Mr. 100% chemical. Where is the P6 at
> 100 MHz, I have never heard of such a thing. What about this 30%
> thing. Talk about an arbitrary number. Yet you eat it up like candy.
> You are more of an evangelist than anything. Stick to science. Oops,
> I just noticed that this post is being cross-posted to the mac .advocates.
> No wonder.
That's OK--it was being compared to a 604 at 100 MHz which doesn't exist,
either. 100 MHz is a simple clock speed for comparison of performance. For
a GIVEN CHIP, the relative performance is approximately a linear function
of clock speed. Therefore, since the 604 is over 40% faster in Specfp ata
calculated 100 MHz, then it's reasonable to guesstimate that it will be
40% faster at 132-133 MHz where both chips are currently available.
Sangria
>As a point of curiosity, does yours have the gold sticker for first place
>on your machine? :-)
You are perhaps assuming that I run DOS/Windows?
Ahhh....nevermind.
-- Sang.
**************************************************************
* Sang K. Choe - san...@inlink.com *
* - http://www.inlink.com/~sangria/index.html *
**************************************************************
Joe Ragosta
>
> : Thanks for the news. I personally never thought the P6 was so great and
> : predicted lousy FP performance on comp.sys.intel several times. By my
> : calculations, the above data gives Specfp92 of 116 for the P6 at 100 MHz
> : compared to 165 for the 604. We'll have to see what the relative clock
> : speeds look like when things settle down.
>
> HA, Ha, Ha, Funniest thing I've seen so far. Joe predicts numbers for a
> chip that he's never seen before, but predicts it from some rules of
> thumb he created. Good one Joe.
If you'd bother learning something in whatever school you're attending,
you might be able to post something intelligent. I never claimed to
predict numbers for fp performance--all I said was that it would be poor.
That's not too hard, given the description Intel has given for the P6.
Essentially, it's nothing but a glorified P5 with a big cache. Can you
please name anyone with any experience in these things who really expected
the P6 to keep up with a RISC chip in FP performance? Why do you think
Intel released SpecInt guesstimates several months ago and has been mum on
SpecFP?
WndrLlamma
I thought you were a scientist Mr. 100% chemical. Where is the P6 at
100 MHz, I have never heard of such a thing. What about this 30%
thing. Talk about an arbitrary number. Yet you eat it up like candy.
You are more of an evangelist than anything. Stick to science. Oops,
I just noticed that this post is being cross-posted to the mac .advocates.
No wonder.
>>>>>
I think you might have missed Mr. 100% Chemical's point. Of course there
is no 100mhz P6.... it was merely used as a theoretical start point.
You must often do this when comparing chips from different families. For
example, an 80mhz 601 is faster than a 90mhz p-5, but slower than a 120mhz
p-5... i.e., an 80mhz 601 is as fast as a (theoretical) 110mhz p-5.
WndrL...@aol.com / Designer Guru Person
----there is no god....only fractals-----
UNDERWOOD EDMOND D
You are definitely an evangelist also. Because there is almost no factual
information in your comment. According to SPEC, your comparisons are false.
Trying to determine a 110 MHz pentium sounds way unscientific. What is it,
a clock doubler, or tripler? That would also determine how fast it is since
you have to get your timings correct between CPU and cache as well as cache
and memory. I don't know where your theory comes from, but it is very
suspect. Stick to SPEC ratings not evangelical Mac .advocacy ratings.
-Edmond
UNDERWOOD EDMOND D
You really are just an evangelist. The P6 is not just a glorified P5.
It is capable of decoding, renaming, and breaking down instructions
into smaller parts. How is this just a glorified P6. It's on board
cache allows it to pipeline effectively. Just read Intel's web page
and then write back to us about what Intel didn't change. Then
look at your incredibly silly statement about the FP performance
and see if you still find it correct. I don't think anyone for that
matter can speculate the P6's FP performance and certainly not you.
Oh and you and your experts explain why the P6's FPU cannot keep
close to the low end RISC workstations like PPC. I want to hear
this one.
-Edmond
UNDERWOOD EDMOND D
Wenhsu Lin
post, you actually included Joe's post. Joe's post is copyrighted by
Microsoft. What the hell????
hey guys, are we going to allow Micro$oft to do that???
Wen.
Edmond Underwood (unde...@Colorado.Edu) wrote:
: doc...@interramp.com (Joe Ragosta) wrote:
[Deleted because I DO NOT want to pay $1000 to Micro$oft]
: I thought you were a scientist Mr. 100% chemical. Where is the P6 at
: 100 MHz, I have never heard of such a thing. What about this 30%
: thing. Talk about an arbitrary number. Yet you eat it up like candy.
: You are more of an evangelist than anything. Stick to science. Oops,
: I just noticed that this post is being cross-posted to the mac .advocates.
: No wonder.
: ---------------------------------------------------------------------------------
Dan Pop
>In article <3r9p2q$8...@lace.Colorado.EDU>, Edmond Underwood
><unde...@Colorado.Edu> wrote:
>
>> I thought you were a scientist Mr. 100% chemical. Where is the P6 at
>> 100 MHz, I have never heard of such a thing. What about this 30%
>> thing. Talk about an arbitrary number. Yet you eat it up like candy.
>> You are more of an evangelist than anything. Stick to science. Oops,
>> I just noticed that this post is being cross-posted to the mac .advocates.
>> No wonder.
>
>That's OK--it was being compared to a 604 at 100 MHz which doesn't exist,
>either. 100 MHz is a simple clock speed for comparison of performance. For
>a GIVEN CHIP, the relative performance is approximately a linear function
>of clock speed. Therefore, since the 604 is over 40% faster in Specfp ata
>calculated 100 MHz, then it's reasonable to guesstimate that it will be
>40% faster at 132-133 MHz where both chips are currently available.
How about waiting until hard data, obtained on real systems, are available?
Too "unscientific" for a Mac advocate? :-)
Dan
--
Dan Pop
CERN, CN Division
Email: Dan...@mail.cern.ch
Mail: CERN - PPE, Bat. 31 R-004, CH-1211 Geneve 23, Switzerland
Dan Pop
>You must often do this when comparing chips from different families. For
>example, an 80mhz 601 is faster than a 90mhz p-5, but slower than a 120mhz
>p-5... i.e., an 80mhz 601 is as fast as a (theoretical) 110mhz p-5.
What's the point of making this kind of silly comparisons? All that
matters is the performance of real systems (or their price/performance
ratios), not the performance of some hypothetical cpu's.
Who cares about the theoretical speed of a given cpu, if it's part of a
lousy system which cripples its performance and the situation is
further aggravated by the use of cheap compilers, generating code which
is considerably slower than that generated by the compilers used by the
benchmarkers (and this is quite common in the world of cheap personal
computers)?
People use computers, not cpu's. In the workstation world you get SPEC
ratings for every model, obtained with the same compilers the average
user is using for his own codes. In the personal computers world, you
compare the SPEC ratings obtained on other systems than those used by
Joe User (and with other compilers), which is more or less a waste of
time and bandwidth.
David T. Wang
: In article <3r9pm7$e...@cville-srv.wam.umd.edu>, veak...@glue.umd.edu
: (David T. Wang) wrote:
[Joe wrote..]
: > : Thanks for the news. I personally never thought the P6 was so great and
: > : predicted lousy FP performance on comp.sys.intel several times. By my
: > : calculations, the above data gives Specfp92 of 116 for the P6 at 100 MHz
: > : compared to 165 for the 604. We'll have to see what the relative clock
: > : speeds look like when things settle down.
: >
: > HA, Ha, Ha, Funniest thing I've seen so far. Joe predicts numbers for a
: > chip that he's never seen before, but predicts it from some rules of
: > thumb he created. Good one Joe.
[Joe writes again..]
: If you'd bother learning something in whatever school you're attending,
: you might be able to post something intelligent. I never claimed to
: predict numbers for fp performance--all I said was that it would be poor.
: That's not too hard, given the description Intel has given for the P6.
: Essentially, it's nothing but a glorified P5 with a big cache. Can you
: please name anyone with any experience in these things who really expected
: the P6 to keep up with a RISC chip in FP performance? Why do you think
: Intel released SpecInt guesstimates several months ago and has been mum on
: SpecFP?
: --
: Regards, Joe Ragosta -- 100% Chemical and proud of it.
I attend the University of Maryland. I do learn some things from time to
time. One of which is learning to produce relatively logical arguements..
Let's see...... Joe, you wrote, "I predict P6 will have lousy SpecFP
performance" then you attempted to flame by reversing course.."I never
claimed to predict fp performace" Seem like a pretty obvious
contradiction to me. Secondly P6's internal architecture is drastically
different from P5's and its quite a bit more than just " a big P5 with a
large cache" You should pick up the ISSCC paper and read up on the
architecture. That's the more technical paper, but if you like the garden
variety easier to understand paper on the Internal architecture of P6,
just read it on the Web. http://www.intel.com/procs/p6/index.html
please refrain from making idiotic generalized statements until you've at
least read those papers. As to releasing SpecFP92 or SpecFP95 numbers, I
wish I could, Joe. Just so I can flame you, but alas, it is not to be,
and that joy will have to be delayed.
: Microsoft Network is prohibited from redistributing this work in any form,
: in whole or in part. Copyright, Joseph Ragosta, 1995. License to distribute
: this post is available to Microsoft for $1000. Posting without permission
: constitutes an agreement to these terms. Please send notices of violation
: to doc...@interramp.com and Postm...@microsoft.com
--
Kevin Stone
: : > Let's let this sink in a moment, shall we. The current P5-90
: : > is around 80 SPECfp92, is it not? The 100 MHZ 601 yields 125 SPECfp92.
If you think that these are wrong.. please explain why? We would
all like to know the numbers you believe to be correct. :)
: : > By the way the P6 is currently running at about 25 - 30 watts.
: : The 604?
: : > 13 watts! One other point, at the moment the P6 is about $1,800.00!
: : > :)
: This is really funny. Please state your sources, I'd like to look these
: up, me thinks they're in error.
Acctualy this is genneraly misleading. As it is true that the P6
was running at 30 watts, it was only durring prototypeing and simulations.
Now.. if they couldn't get it below 25 watts in simulations.. then how
do we expect them to give us a chip that is any lower in real life?
Anyone know what the real wattage input is now? I would assume that we
won't be seeing any P6 notebooks very soon what ever it is. ;)
: : Thanks for the news. I personally never thought the P6 was so great and
: : predicted lousy FP performance on comp.sys.intel several times. By my
: : calculations, the above data gives Specfp92 of 116 for the P6 at 100 MHz
: : compared to 165 for the 604. We'll have to see what the relative clock
: : speeds look like when things settle down.
: HA, Ha, Ha, Funniest thing I've seen so far. Joe predicts numbers for a
: chip that he's never seen before, but predicts it from some rules of
: thumb he created. Good one Joe.
Very well Laughing boy.. let's hear what YOUR calculations are..
no doubt they put the P6 ahead at a Specfp92 of 116million. :P
: : Of course, this also neglects the fact that you'll be able to buy a 604
: : Mac in a week or so and the P6 boxes will be available about the same time
: : as WIn95. (Any guesses?)
: Good for Mac users, but P6's will be released shortly. There's a lot of
: people working hard at it in Oregon :)
And Apple, Motorola, and IBM will benefit by producing a chip
that costs less than half that of the P6. Although it has been a
tradition of sorts for Intel to release high prices on thier newest, most
powerful chips, it is my oppinion that this price difference will play a
big part in the PPC markets favor.
: : We'll hear a lot of messages from Wintel fanatics about how FP doesn't
: : really matter because most users don't need it (remember the Pentium FP
: : fiasco?), so the P6 is still just as good as the 604.
: Floats do matter. P6 will have decent numbers, don't want to baited into
: speculating numbers along with you, but I'll make a wager with you, it'll
: be , oh, slightly highers than 116 :) Have a nice day.
Then.. what? Higher than the 604.. it'll have to be to compete.
The P6 has been made out by Intel to be a god chip... something that
will revolutionize the PC industry. I can't see it as that.. although
Intel and it's advocates have done a good job at trying to make it seem
as though it is. Intel's chips have never been much up on floating
point... Motorola has always had the upperhand in that department. So we
don't see Intel's new design being any different. And with the 604
blowing away every other aspect of processing in addition to Motorola's
consitent Floating point dominence, I believe the PowerPC (IBM, Apple,
and others) will gain a substantial margin on the PC market, perhaps
begining to sap some of the profits out of the big tippers like Digital
and Compaq. Well... we'll see. :)
: : Microsoft Network is prohibited from redistributing this work in any form,
: : in whole or in part. Copyright, Joseph Ragosta, 1995. License to
: : distribute
: : this post is available to Microsoft for $1000. Posting without permission
: : constitutes an agreement to these terms. Please send notices of violation
: : to doc...@interramp.com and Postm...@microsoft.com
ditto. ;)
-Kevin Stone
Inconnu
GUY NUSSBAUM <bro...@ix.netcom.com> wrote:
>n <bennetk1-090...@128.187.250.7> benn...@yvax.byu.edu (M.
>Wade Heninger) writes:
>In article <3rar00$9...@ixnews5.ix.netcom.com>, bro...@ix.netcom.com
>(GUYNUSSBAUM) wrote:
>What I don't understand is the reason why some people(ALL MAC PEOPLE,
>AS WELL AS SOME PC PEOPLE) try to put them down,
TRUE! Intel takes _way_ too much flack.
intel right now has
>the fastest chip on the market and microsoft has, well, a big company
Let's not get into another specwar, but I think we can all agree that the
604@133MHz is faster than the P5@120 (Intel's fastest chip on the market)
>So I would expect windows 95.1!!!! to be the best O/S ever! using 32bit
Just curious-- what's the .1 (is there already a revision?, it hasn't been
released yet..)
[snipped stuff about software "magic" a la MPEG and other dlls]
>uhhhhhhh..... take your time 8.0 is out in a few months as well as 95,
>give me an example of where exactly did copland go where the big M
>didn't, it took copland about 10 years to go where the big M
well, the _first_ version of Windows was announced (in the works) a full year
before the release of the Macintosh (1983), so both systems have been
competing for the same ammount of time; it's just taken Microsoft this long to
finally catch up. Copland looks _nice_ from a user point-of-view, I won't go
into details, but check out the July MacWorld for a full side-by-side of the
2 systems. MacWorld is _very_ objective.
>Is in a few short years, THE P6 and win95 is the future, and there is
>no stopping it
Actually, this is your opinion and mine is that the many platforms complement
each other. Microsoft takes the good features of the Mac, Apple takes
features from other OS's (X11); competition is _good_ if MS owned
everything there would be no innovation. All these companies are american,
and I'm proud of them.
--
-INK <kell...@isuux.isu.edu>
Idaho State University
GUY NUSSBAUM
Wade Heninger) writes:
(GUYNUSSBAUM) wrote:
--Snip--
>intel right now has the fastest chip on the market and microsoft has,
>well, a big company
>Let's not get into another specwar, but I think we can all agree that
>the
>604@133MHz is faster than the P5@120 (Intel's fastest chip on the
>market)
Is the 604 out yet?, NO i didn't think so!
>>So I would expect windows 95.1!!!! to be the best O/S ever!using
>>32bit
>Just curious-- what's the .1 (is there already a revision?, it hasn't
>been released yet..)
>[snipped stuff about software "magic" a la MPEG and other dlls]
The .1 is for the second release not a revision! it is going to be
something like win 3.1 and win 3.11, and even if it is a revision, at
least I'm not saying that it is perfect, I'm sticking to facts here
unlike some other MAC people I know that seem to relay on RANDOM SPEC
numbers (NOT you) and future products (that's you, as well as all the
mac people that try to justify their reason for buying a mac)
seriously now, since I see you are not one of these people (except for
the 604, but I'll forgive you for that) :-)
I think both the pc and the mac have something to gain from each others
platform.
>SNIP you know what I sniped (I'll forgive you for that too :-). All
>these companies are >american,and I'm proud of them.
I agree w/ you 100% here!!
except for the fact we (americans) are still buying memory made in
japan, not that I have anything against japan, but, as long as we relay
on the small people (joking) :-) for our RAM we're as dependent on them
as we are w/ electronics, we have to talk to all these companies
(microsoft, intel, apple, and IBM) and persuade them to built a ram
fabrication plant!
>--
> -INK <kell...@isuux.isu.edu>
> Idaho State University
TAKE CARE
GUY NUSSBAUM
Robert Colwell
>I think you might have missed Mr. 100% Chemical's point. Of course there
>is no 100mhz P6.... it was merely used as a theoretical start point.
>
>example, an 80mhz 601 is faster than a 90mhz p-5, but slower than a 120mhz
>p-5... i.e., an 80mhz 601 is as fast as a (theoretical) 110mhz p-5.
On the off chance that somebody really does care about this stuff, I will
remind you all of one subtlety: the P6 is superpipelined. This is *not* a
vacuous buzzword, it means P6 can wring a higher clock out of a given
implementation technology. For instance, we got 133MHz out of the same
process tech that the Pentium got 100MHz. (The higher speed Pentiums were
on different processes.)
We paid a price for that higher clock, in deeper pipelines, need for a more
capable (and larger) branch target buffer, and so on, but we felt it was a
good trade-off overall for the higher clock.
If you insist on making the clock rates the same when you attempt a
microarchitectural comparison, you are distorting the overall picture,
because you're throwing away the performance win associated with the P6's
superpipelining.
Bob Colwell bcol...@ichips.intel.com
Intel Corp. JF1-19
5200 NE Elam Young Parkway
Hillsboro, Oregon 97124
Rob Barris
bcol...@pdx145.intel.com (Robert Colwell) wrote:
> On the off chance that somebody really does care about this stuff, I will
> remind you all of one subtlety: the P6 is superpipelined. This is *not* a
> vacuous buzzword, it means P6 can wring a higher clock out of a given
> implementation technology. For instance, we got 133MHz out of the same
> process tech that the Pentium got 100MHz. (The higher speed Pentiums were
> on different processes.)
>
> We paid a price for that higher clock, in deeper pipelines, need for a more
> capable (and larger) branch target buffer, and so on, but we felt it was a
> good trade-off overall for the higher clock.
>
> If you insist on making the clock rates the same when you attempt a
> microarchitectural comparison, you are distorting the overall picture,
> because you're throwing away the performance win associated with the P6's
> superpipelining.
" *I* care! " - _Star Wars_
I hesitate to jump in and argue with Dr.P6, but:
When comparing chips that are going to be available at similar or very
close clock rates in similar time frames, there must be some validity to
it. If 604 and P6 systems are both shipping with 133-MHz clocks, then
knowing how much work either chip can get done, with those MHz, at what
price, seems OK to me.
If one forgoes that basis for comparison, what's left?
chip price
system price
compatibility
reliability
upgradeability
runs DOOM-2 really fast
<insert your favorite criteria here>
I agree with your complaint about oversimplifying, the point about being
able to wring out more raw speed from an older process (and of course even
more speed once you move to the newer processes, relative to P5) is very
appealing.
Eric H. Larson
> remind you all of one subtlety: the P6 is superpipelined. This is *not* a
> vacuous buzzword, it means P6 can wring a higher clock out of a given
> implementation technology. For instance, we got 133MHz out of the same
> process tech that the Pentium got 100MHz. (The higher speed Pentiums were
> on different processes.)
>
> We paid a price for that higher clock, in deeper pipelines, need for a more
> capable (and larger) branch target buffer, and so on, but we felt it was a
> good trade-off overall for the higher clock.
>
> If you insist on making the clock rates the same when you attempt a
> microarchitectural comparison, you are distorting the overall picture,
> because you're throwing away the performance win associated with the P6's
> superpipelining.
I'm sure that you achieved a performance boost by superpipelining, but as
you say, superpipelining isn't free. It costs in more complex logic, and
perhaps larger cache demands, too. How do we compare such costs?
SPEC/WATT? SPEC/$$$? SPEC/???. Joe's method of SPEC/MHz may not be that
far off the real economic worth of these designs, because of the added
cost of achieving a superpipelined design. Also, we shouldn't forget that
the 604 seems to also be available in a 132 MHz version.
GUY NUSSBAUM
doc...@interramp.com (Joe Ragosta) writes:
>
>In article <3r79vb$d...@hearye.mlb.semi.harris.com>,
>ke...@mars.mlb.semi.harris.com (Ken Sherman) wrote:
>
>> Well, macophiles, I come bearing good news. The first
numbers
>> are in concerning the P6 and its floating point performance. Are you
ready?
>> The P6 is about 30% faster than the P5 at the same clock speed...
Intel claims
>> its twice as fast as a P5, but that includes the speed up due to
faster
>> clocking speeds.
>> is around 80 SPECfp92, is it not? The 100 MHZ 601 yields 125
SPECfp92.
coming out
>> at 120 and 133 MHZ. Can you say 604 kicks the P6s ass? Do a little
math.
>> The 601 already outperforms the P5 and the 604 is about 1.6 X
faster.
>The P6 is only about 1.3 X faster than the P5.
watts.
>The 604?
>> 13 watts! One other point, at the moment the P6 is about $1,800.00!
>> :)
>>
>
>Thanks for the news. I personally never thought the P6 was so great
and
>predicted lousy FP performance on comp.sys.intel several times. By my
>calculations, the above data gives Specfp92 of 116 for the P6 at 100
MHz
>compared to 165 for the 604. We'll have to see what the relative clock
>speeds look like when things settle down.
>
604
>Mac in a week or so and the P6 boxes will be available about the same
time
>as WIn95. (Any guesses?)
>
>
>We'll hear a lot of messages from Wintel fanatics about how FP doesn't
>really matter because most users don't need it (remember the Pentium
FP
>fiasco?), so the P6 is still just as good as the 604.
>
>Regards, Joe Ragosta -- 100% Chemical and proud of it.
>
form,
>in whole or in part. Copyright, Joseph Ragosta, 1995. License to
distribute
>this post is available to Microsoft for $1000. Posting without
permission
>constitutes an agreement to these terms. Please send notices of
violation
>to doc...@interramp.com and Postm...@microsoft.com
And where did you get your numbers from?
Ok lets compare a mac and a pc!
PC!!!!!
1)Windows 95!
going to be 32bit O/S which w/ the use of 32bit applications promises
to be twice as fast as win 3.1 runnig the same application in 16bit
code!
2)P6!
the first version of the p6 running at 133mhz promises to be twice as
fast as the pentium 100.
I don't want to say anything about the MAC because I have no experiance
with the mac other then using one (I didn't like it).
SO, ARE WE IN TROUBLE?
GUY NUSSBAUM
---SNIP----
13 watts! One other point, at the moment the P6 is about $1,800.00!
---SNIP---
Wait just one minute here, wait, wait, ok! :-)
IF!!!!!!(which is bull)
THe first P6 will be around 1800, and as more and more chips sell the
price goes down!
Now how many chips do you expect intel to sell and how many do you
expect apple?
It is a fact that intel sells more chips then apple, so I would expect
the price of the P6 to drop way below that of the 604 and 620 (when the
p6 goes 210mhz)
So if your comparing the two chips please I'll talk to you after the
chip has been selling for two months and discuss the prices w/ you!
GUY NUSSBAUM
M. Wade Heninger
> In article <bennetk1-060...@128.187.250.19>,
benn...@yvax.byu.edu says...
>
> >As a point of curiosity, does yours have the gold sticker for first place
> >on your machine? :-)
>
> You are perhaps assuming that I run DOS/Windows?
> Ahhh....nevermind.
Yes, actually I *know* you run windoze...it may be NT but its name still
is windoze...er did M$ decide to change the name to something smancy like
NT95?
Wade
____________________________________________________________________
M. Wade Heninger |
801.375.7018 |"They can drive you crazy. They really can."
benn...@yvax.byu.edu | Holden Caulfield
____________________________________________________________________
M. Wade Heninger
> In article <bennetk1-060...@128.187.250.19>,
benn...@yvax.byu.edu says...
>
> >As a point of curiosity, does yours have the gold sticker for first place
> >on your machine? :-)
>
> You are perhaps assuming that I run DOS/Windows?
> Ahhh....nevermind.
Actually I dont have to assume. Judging from you many posts, you often
speak of using NT...which is, of course, called *WindozeNT*. So why the
denial :-)
M. Wade Heninger
>
> And where did you get your numbers from?
> Ok lets compare a mac and a pc!
> PC!!!!!
> 1)Windows 95!
> going to be 32bit O/S which w/ the use of 32bit applications *promises*
There is the key word: Promises...M$ promises alot of things that they
dont carry thru with.
> to be twice as fast as win 3.1 runnig the same application in 16bit
> code!
> 2)P6!
> the first version of the p6 running at 133mhz promises
Eeek...there it is again...
> to be twice as fast as the pentium 100.
> I don't want to say anything about the MAC because I have no experiance
> with the mac other then using one (I didn't like it).
> SO, ARE WE IN TROUBLE?
Wade
GUY NUSSBAUM
>>:And where did you get your numbers from?
>>:Ok lets compare a mac and a pc!
>>:PC!!!!!
>>:1)Windows 95!
>>:going to be 32bit O/S which w/ the use of 32bit applications
>>:*promises*
>There is the key word: Promises...M$ promises a lot of things that
>they don't carry thru with.
At least they don't generate RANDOM numbers
>>:to be twice as fast as win 3.1 running the same application in 16bit
>>:code!
>>:2)P6!
>>:the first version of the p6 running at 133mhz promises
>Eeek...there it is again...
I could write will, because in INTEL's case I never seen a promise that
was broken.not that microsoft ever did, but applications(well the
majority) are not made by microsoft!
What I don't understand is the reason why some people(ALL MAC PEOPLE,
AS WELL AS SOME PC PEOPLE) try to put them down, intel right now has
the fastest chip on the market and microsoft has, well, a big company
:-)
So I would expect windows 95.1!!!! to be the best O/S ever! using 32bit
code and all those GUI enhancements including OpenGL(probably in a form
of a dll), winG (which will let us PC people, run doom under
windows),wintoon,And may include a new technology called RL library (a
small RenderMorphics company recently acquired by the big M whose
reality lab supports high end 3-d rendering in software and hardware)
Plus it might (all probability is it will) also include a special mpeg
technology player!, this w/ the use of autoplay which will
automatically play cd's, w/ the use of True speechvoice compression, a
general MIDI standard for windows, a standard for digital joystick, and
a new multisession cd format that combines audio and data on one disk!
So w/ all these GUI as well as all the other 32bit enhancements for
speed, we will soon see all of the MAc people say, well but we have
uhhhhhhh..... take your time 8.0 is out in a few months as well as 95,
give me an example of where exactly did copland go where the big M
didn't, it took copland about 10 years to go where the big M
Is in a few short years, THE P6 and win95 is the future, and there is
no stopping it
-----------You're on the right track, on the wrong train---------------
>>to be twice as fast as the pentium 100.
>>I don't want to say anything about the MAC because I have no
>>experience
>>with the mac other then using one (I didn't like it).
>>SO, ARE WE IN TROUBLE?
>Wade
>____________________________________________________________________
>M. Wade Heninger |
>801.375.7018 |"They can drive you crazy. They really can."
>benn...@yvax.byu.edu | Holden Caulfield
>____________________________________________________________________
Thank you
GUY NUSSBAUM
Remeber don't let the train pass you BYE!!!!
Lawson English
[snipt]
: For most users FP performance does not matter. For those that FP
: switch to something else, like a Unix workstation. Virtually
: every market research study about this has shown this to be true.
: FP is important to 5 - 10% of the market.
Maybe....
Maybe not. On the PowerMac, there is no DSP. All DSP-functions are done
using "Native Signal Processing." These include sound I/O, Speech
Recognition, and possibly other things. The new QuickDraw 3D library is
such a heavy user of floating point that Apple only targets it at PowerPC
and P5/P6 systems. Fractal Design Power is a VERY cool application that
allows non-artists to create professional looking 3D poses of human
figures. My significant other is a cartoonist with dyslexia who has a
hard time with perspective. Poser makes her just as good (if not better)
than the most skilled commercial artist. Poser is a real sleeper, IMHO,
as EVERY aspiring artist will want a copy. As far as I can tell, it is
very FP intensive. While Doom-level games don't require FP, more
realistic 1POV games will almost certainly require it. 3D VR probably will.
The bottom line is that as FP becomes better, programmers will find
low-end uses for it that no-one has ever heard of before because it was
never feasible to use a $25,000 graphics workstation for
home-user-oriented tasks.
--
-------------------------------------------------------------------------------
Lawson English __ __ ____ ___ ___ ____
eng...@primenet.com /__)/__) / / / / /_ /\ / /_ /
/ / \ / / / / /__ / \/ /___ /
-------------------------------------------------------------------------------
Edmond Underwood
>I agree w/ you 100% here!!
>except for the fact we (americans) are still buying memory made in
>japan, not that I have anything against japan, but, as long as we relay
>on the small people (joking) :-) for our RAM we're as dependent on them
>as we are w/ electronics, we have to talk to all these companies
>(microsoft, intel, apple, and IBM) and persuade them to built a ram
>fabrication plant!
>>--
>> -INK <kell...@isuux.isu.edu>
>> Idaho State University
>TAKE CARE
>GUY NUSSBAUM
For your info, all my ram is made in America. There are 3 Co.'s that do
this. Micron, IBM electronics, and Texas Instruments. New ram technologies
are being made in the Colorado Area all the time. Ever heard of EDRam?
Just head out for Colorado Springs.
Sam Sarmast x4489
: In article <3rar00$9...@ixnews5.ix.netcom.com>, bro...@ix.netcom.com (GUY
: >
: > And where did you get your numbers from?
: > Ok lets compare a mac and a pc!
: > PC!!!!!
: > 1)Windows 95!
: > going to be 32bit O/S which w/ the use of 32bit applications *promises*
Well, all the arguing about wither the P6 will be faster than a 604 is
rather mute for several reasons:
1-the P6 will probably still run some sort of DOS/WIN
variant. i.e. it'll be running like a really, really
really fast 8088 with 32bit registers.
2-What software will the PowerMac run? Some sort of native
mode OS to take advantage of it's performance. I'd be
surprised if it doesn't emulate like a 68000 like the current
generation of PowerPC's do.
sam
GUY NUSSBAUM
>Systems Management Group
>Computing & Network Services (University of Colorado)
>E-mail: unde...@Colorado.Edu
>
company in japan :-)
Well I guess I was wrong!
I really didn't know the big blue, and Texas Instruments where into RAM
I'm really happy :-) good for us!
GUY NUSSBAUM
Bruce Grubb
>---SNIP---
>---SNIP----
>13 watts! One other point, at the moment the P6 is about $1,800.00!
>---SNIP---
>---SNIP---
>Wait just one minute here, wait, wait, ok! :-)
>IF!!!!!!(which is bull)
>THe first P6 will be around 1800, and as more and more chips sell the
>price goes down!
I do not even know where this number is coming from. Clearly not from
THIS planet.
>Now how many chips do you expect intel to sell and how many do you
>expect apple?
sell and how many do you expect MOTOROLA to sell?"
Macs are NOT made by JUST Apple any more; you have PowerComputing,
Radius, DayStar, and two asian companies making Macs. Besides the MPC6xx
line is a MOTOROLA chip not an Apple chip and is found in some high end
in IBMs.
Besides there are the DOS Macs with their 486dx2s. Funny I never knew
APPLE made 486dx2s but you are saying EXACTLY that. :)
This is all in the Mac-IBM-compare sheet. Reading it saves one from
having so much hoof and mouth problems.
>It is a fact that intel sells more chips then apple, so I would expect
>the price of the P6 to drop way below that of the 604 and 620 (when the
>p6 goes 210mhz)
This does not hold true presently for the P5 vs. the 601. The 601 still
is about 75% the cost of a P5 of the same MHz. Never mind that in real
world tests by Ingram the 601 is 33% faster than a P5 of the same MHz.
All the poorly optimized 386 and 486 code out there has made the 486dx4
eclipse the Pentium in sales anyhow.
>So if your comparing the two chips please I'll talk to you after the
>chip has been selling for two months and discuss the prices w/ you!
I'll be glad to talk to you about the P6 and 604 when you get you facts
right as to who make the 6xx line of chips and who uses and will be using
them.
GUY NUSSBAUM
>>Now how many chips do you expect intel to sell and how many do you
>>expect apple?
>Wrong question. Should be "Now how many chips do you expect Intel to
>sell and how many do you expect MOTOROLA to sell?"
:I'm not talking about the 60830 or whatever you call those MACs!!the
:chips I'm referring to are the power MAcs which if you want to get
:technical on me are made by motorola together w/ IBM and apple, so if
:you want me to write motorola OK, I will :-) (shaking head)
>Macs are NOT made by JUST Apple any more; you have PowerComputing,
>Radius, DayStar, and two asian companies making Macs. Besides the
>MPC6xx
>line is a MOTOROLA chip not an Apple chip and is found in some high
>end in IBMs.
:I'm not talking about the 10 people who bought those work stations
:using the power pc by IBM. I know that apple has licensed other
:manufacturers to produce apples, I had this same discussion a couple
:of months ago on this same subject about the mac and pc comparison
>Besides there are the DOS Macs with their 486dx2s. Funny I never knew
>APPLE made 486dx2s but you are saying EXACTLY that. :)
:Where exectly did I say this, Hey mac man, wake up and look at what
:your reading!as you say, Reading, it saves one from having so much
hoof :and mouth problems.
>This is all in the Mac-IBM-compare sheet. Reading it saves one from
>having so much hoof and mouth problems.
:The MAC-IBM compare sheet? and where is that?
>>It is a fact that intel sells more chips then apple, so I would
>>expect the price of the P6 to drop way below that of the 604 and 620
>>(when the p6 goes 210mhz)
>This does not hold true presently for the P5 vs. the 601. The 601
>still is about 75% the cost of a P5 of the same MHz. Never mind that
>in real
:I don't know where you got your 75% figure, but the fact a 601 is
:cheaper in per mhz is true
>Never mind that in real world tests by Ingram the 601 is 33% faster
>than a P5 of the same MHz.
:NO,NO, and another NO where do all of you mac people get your numbers?
:Stop reading that MAC world magazine, or whatever your reading and go
:pick up some good books!
:First thing I have to say is in today market the mhz/performance
:figure doesn't hold anymore!
:eventually we are going to reach the limit at which the mhz is going
:to stop and the only feature we must add will be those of super
:pipelining, and other enhancement
:And if your looking for real world figures it is a fact pc's are
:faster then macs!!!(I get my facts where you get yours)
:So what your saying (writing) is that a 601 which is now (I think, you
:see I'm not saying I know everything, I admit, where I'm not sure)
:at 100mhz, am I right? well If I'm then your saying a 100mhz 601 is as
:fast as a pentium133mhz? yes?
:well I have to say bulllllllll sh***, you know what It's not even
:worth it arguing w/ somebody like you, you have no idea what your
:talking about, next time show some figures URL's FAQ's something, Just
:forget it, use your 80mhz 8meg $ 700megHD $3000 MAc, and dream of
:those 100mhz selling for the same price w/ 16meg 1.2gb HD!!!
:And just dream of the performance (IN REAL WORLD APPLICATIONS) you
:will get using a monster machine like that!
:AND PLEASE STOP w/ all that stupid stuff!
:The less you write the smarter you'll look
>All the poorly optimized 386 and 486 code out there has made the
>486dx4 eclipse the Pentium in sales anyhow
Oppsss, OK could you please stop w/ the stupid stuff?
>>So if your comparing the two chips please I'll talk to you after the
>>chip has been selling for two months and discuss the prices w/ you!
>I'll be glad to talk to you about the P6 and 604 when you get you
>facts
:When I get my facts, you know what, forget it, it's hard to talk to
:somebody living in a dream world. OK, just go to sleep and we'll talk
:about it when you wake up form that dream you're having
>right as to who make the 6xx line of chips and who uses and will be
>using them.
:Who will use them, people like you, that think that they are fast
:until they play around w/ a pentium and realize that the mac they have
:been using is to slow to use, and that that 80mhz RISC chip is moving
:at the speed of that optimized 386sx LMAO
GUY NUSSBAUM
Loren Petrich
>Well, all the arguing about wither the P6 will be faster than a 604 is
>rather mute for several reasons:
> 1-the P6 will probably still run some sort of DOS/WIN
> variant. i.e. it'll be running like a really, really
> really fast 8088 with 32bit registers.
Complete with a CISC instruction set and a small and specialized
register set. With only 8 general-purpose registers and some specialized
ones, the chip will have to waste a lot of time moving data to and from
them.
The PowerPC chip is free from those defects, thanx to its
RISCiness --
* It has 32 general-purpose registers, any one of which can be used for
addressing. So if you wish to do operations on several arrays at once,
you can have pointers to each one of them in the register set at once,
which may be difficult with the Intel architecture.
* All of its instructions are the same size (4 bytes), and they all have
4-byte alignment. This *greatly* simplifies instruction handling, since
these features can easily be hardwired into the chip.
* The only instructions that move data to and from main memory are load
and store ones. This means that other instructions can easily be
scheduled between load and store ones, so they can take several cycles to
complete, if need be, without holding up the rest of the processing.
> 2-What software will the PowerMac run? Some sort of native
> mode OS to take advantage of it's performance. I'd be
> surprised if it doesn't emulate like a 68000 like the current
> generation of PowerPC's do.
In the immediate future, we'll still be having a largely-emulated
MacOS, but much of the time-critical parts are native, and Apple is making
more and more of it native. It's next major release, System 8 / Copland,
is supposedly to have 95% native PPC code, but that is probably
excessively optimistic. However, they may be able to get at least 50%
native in 8.0 (IMO, if they have any sense, they'd first try to make only
the new stuff native, and if they have trouble with the old stuff, then
leave that emulated for the time being).
--
Loren Petrich, the Master Blaster Happiness is a fast Macintosh
pet...@netcom.com And a fast train
Visit ftp://ftp.netcom.com/pub/pe/petrich
Or ftp to ftp.netcom.com, then go to /pub/pe/petrich
Eric H. Larson
> : For most users FP performance does not matter. For those that FP
> : performance really matters, they have never used x86 PC's
>
>
>
> Maybe not. On the PowerMac, there is no DSP. All DSP-functions are done
> using "Native Signal Processing." These include sound I/O, Speech
> Recognition, and possibly other things. The new QuickDraw 3D library is
> such a heavy user of floating point that Apple only targets it at PowerPC
> and P5/P6 systems.
It certainly seems to me that if you use a CPU for which FP is a 'penalty
box', you are not going to see much software that makes heavy use of
floating point. Comments that 'floating point is not important' arise at
least partly because there is no effective floating point on many x86
CPUs, and nobody develops software that requires fast FP. The advent of
CPU's with fast FP as a standard feature is very much likely to increase
the percentage of software that makes use of this feature, and thus
increases it's market importance.
GUY NUSSBAUM
P5-90
>: : > is around 80 SPECfp92, is it not? The 100 MHZ 601 yields 125
SPECfp92.
>
> If you think that these are wrong.. please explain why? We would
>all like to know the numbers you believe to be correct. :)
In <3rbkka$e...@krypton.hpc.sdsmt.edu> kbs...@silver.sdsmt.edu (Kevin
Stone) writes:
>
>David T. Wang (veak...@glue.umd.edu) wrote:
>>Let's let this sink in a moment, shall we. The current P5-90
>>is around 80 SPECfp92, is it not? The 100 MHZ 601 yields 125
>>SPECfp92.
>: no, go look up the correct spec marks please.
>
>If you think that these are wrong.. please explain why? We would
>all like to know the numbers you believe to be correct. :)
YOU want some SPEC numbers, enjoy! And eat your heart out MAC people!
NOw I would like to see where you got your 125?
SPEC92 Results
-1 MB L2 Cache Configuration
SPEC92 Pentium(R) Pentium Pentium
Processor Proc. Proc.
90 MHz 100 MHz 120 MHz
SPECint92 110.1 121.9 140.0
SPECfp92 84.4 93.2 103.9
-512 KB L2 Cache Configuration
SPEC92 Pentium(R) Pentium Pentium Pentium
Processor Proc. Proc. Proc.
75 MHz 90 MHz 100 MHz 120 MHz
SPECint92* 89.1 106.5 118.1 133.7
SPECfp92* 68.5 81.4 89.9 99.5
-256 KB L2 Cache Configuration
SPEC92 Pentium(R) Pentium
Processor Proc.
60 MHz 66 MHz
SPECint92* 70.4 78.0
SPECfp92* 55.1 63.6
Landmark 3.0 Pentium(R) Pentium Pentium Pentium Pentium Pentium
Processor Proc. Proc. Proc. Proc. Proc.
60 MHz 66 MHz 75 MHz 90 MHz 100MHz 120MHz
CPU Mark(OPs/Sec) 2599.7 2894.6 2970.9 3493.8 3988.1 4721.5
FPU Mark(Ops/Sec) 1069.5 1188.3 1337.1 1604.4 1782.6 2134.9
PC Bench 9.0 Pentium(R) Pentium Pentium Pentium Pentium Pentium
Processor Proc. Proc. Proc. Proc. Proc.
Processor 60 MHz 66 MHz 75 MHz 90 MHz 100 MHz 120 MHz
Harmonic 134.4 149.3 165.3 199.0 221.0 256.8
Whetstone Results
Whetstone Pentium(R) Pentium Pentium Pentium Pentium Pentium
(MWhet/Sec) Processor Proc. Proc. Proc. Proc. Proc.
60 MHz 66 MHz 75 MHz 90 MHz 100MHz 120MHz
Single Precision 61.9 68.4 80.5 96.4 106.0 128.8
Double Precision 61.2 67.1 78.2 93.6 103.0 125.1
SYSmark95 Pentium(R) Pentium Pentium Pentium Pentium Pentium
for Windows Processor Proc. Proc. Proc. Proc. Proc.
60 MHz 66 MHz 75 MHz 90 MHz 100 MHz 120 MHz
Overall 227.7 250.5 290.7 339.5 373.7 418.5
Word Processing 207.9 229.5 264.6 310.7 340.6 374.5
Spreadsheet 208.9 229.7 259.9 309.5 338.5 388.0
Database 288.9 314.6 385.6 428.5 475.4 517.9
Desktop Graphics 227.4 252.0 284.5 341.2 378.5 442.0
Desktop Presen. 205.7 226.3 262.8 311.8 342.0 378.9
Desktop Publish. 190.8 212.8 249.4 290.7 313.9 343.3
NOW here are the figures for the "power"PC
PowerPC 601
Performance: 80 MHz
SPECint92* 85
SPECfp92* 105
Performance 100 MHz
SPECint92* 105
SPECfp92* 125
*Estimated performance.
Why is this here, Estimated performance?
Why don't the pentiums have an estimated performance?
Why is it the INTEL never glorifies their chips?
I'll tell you why, here is why.
I know somebody that works in a lab and they have a couple of 6100/66
as well as a dozen p5's 60's and 90's
and he says that...well here is their exact words"- they ARE a joke :
reaction of my colleagues when we got the PowerMac:
"Why is it so slow" - and about Mac addicts "
So please stop w/ the mac talk, it's a joke that laughes in your face
every time you turn it on :-)
TAKE CARE
GUY NUSSBAUM
Edmond Underwood
>That's OK--it was being compared to a 604 at 100 MHz which doesn't exist,
>either. 100 MHz is a simple clock speed for comparison of performance. For
>a GIVEN CHIP, the relative performance is approximately a linear function
>of clock speed. Therefore, since the 604 is over 40% faster in Specfp ata
>calculated 100 MHz, then it's reasonable to guesstimate that it will be
>40% faster at 132-133 MHz where both chips are currently available.
>
>--
>Regards, Joe Ragosta -- 100% Chemical and proud of it.
>
>Microsoft Network is prohibited from redistributing this work in any form,
>in whole or in part. Copyright, Joseph Ragosta, 1995. License to distribute
>this post is available to Microsoft for $1000. Posting without permission
>constitutes an agreement to these terms. Please send notices of violation
>to doc...@interramp.com and Postm...@microsoft.com
You lost me Joe. What are you comparing now? Are you saying that the 604
is over 40% faster in SPECfp than the Pentium or the P6? If it's the P6, where the
heck are you getting this information. Please let us know.
---------------------------------------------------------------------------------
Edmond Underwood
>Edmond, I think you just owe Microsoft $1000, because when you reply the
>post, you actually included Joe's post. Joe's post is copyrighted by
>Microsoft. What the hell????
>
>
>hey guys, are we going to allow Micro$oft to do that???
>
>
>Wen.
>
>we...@cory.berkeley.edu
>
I'll try to be more careful next time. :^)
Julian A F Bradbury
writes:
> low-end uses for it that no-one has ever heard of before because it was
> never feasible to use a $25,000 graphics workstation for
> home-user-oriented tasks.
Er VR & 3D?
- I think Messrs Lockheed & Glint may make this argument partly irrelevent :)
- Indeed both PPC & PC are limited in this area by memory-bus bandwidth.
The PC (P5) has a slight advantage with 66Mhz memory-bus, asynchronous
burst-mode caches, and chipsets like the Triton. FPU isn't everything.
- Yet Tritons 100MB/s whilst impressive is a total joke, as are PPCs,
compared to SGI m/cs. Future? R10000 has memory-cpu of 1200MB/s.
The PPCs memory-bus limitations here negate its FPU advantage.
--
____________________________________________________________________________
Julian A F Bradbury Ergonomist & IT Specialist
Rob Barris
> In article: <3ree6b$3...@news.primenet.com> eng...@primenet.com (Lawson
English)
> writes:
> > Mike Schmit (msc...@ix.netcom.com) wrote:
> > [snipt]
> > The bottom line is that as FP becomes better, programmers will find
> > low-end uses for it that no-one has ever heard of before because it was
> > never feasible to use a $25,000 graphics workstation for
> > home-user-oriented tasks.
>
> Er VR & 3D?
> - I think Messrs Lockheed & Glint may make this argument partly irrelevent :)
available.
The machine that can make do with no custom hardware at all will always be
cheaper to make (assuming reasonable economies of scale)...
> - Indeed both PPC & PC are limited in this area by memory-bus bandwidth.
> The PC (P5) has a slight advantage with 66Mhz memory-bus, asynchronous
> burst-mode caches, and chipsets like the Triton. FPU isn't everything.
Your opinion is that memory bus bandwidth is one of a number of limiting
factors. How about PCI bus performance? PCI card makers have indicated
speed boosts of 50 to 250% when running their PCI cards on PCI Macs.
> - Yet Tritons 100MB/s whilst impressive is a total joke, as are PPCs,
> compared to SGI m/cs. Future? R10000 has memory-cpu of 1200MB/s.
Mike did say home-oriented. Name the cheapest usable R10000 system
available. SGI aren't magicians, they build amazing systems at amazing
prices.
> The PPCs memory-bus limitations here negate its FPU advantage.
The PPC (601, 604) can run at 1:1 core:bus speed ratio if the system
designer so chooses. Fact is though, today's systems are achieving
comparable performance to Pentium systems, but with 2:1 buses and
therefore cheaper support systems (that home market again)...
Even the 601 implements a split transaction bus as do the 604 and P6:
"PowerPC 601 Microprocessor User's Manual, page 9-1:"
"... the interface allows one level of pipelining, that is, there can be
two outstanding reads and writes at any given time."
Which memory bus limitations were you referring to? Try to be specific.
Joe Ragosta
<unde...@Colorado.Edu> wrote:
> You lost me Joe. What are you comparing now? Are you saying that the 604
> is over 40% faster in SPECfp than the Pentium or the P6? If it's the
P6, where the
> heck are you getting this information. Please let us know.
This was based on SpecFP number posted by someone on the net. While I
can't vouch for the accuracy of the figures, the numbers they posted
showed the P6 to be a relative FP dog.
What I did, was normalize to the same clock speed (since the data was at
80 MHz or something ridiculous). Since the P6 and 604 run at similar clock
speeds, this doesn't seem like too much of a stretch.
The results is that the P6 is way behind the 604 in FP performance. I
can't see why everyone is so surprised by this result (I got more flames
for this than for anything I've ever written). Intel has been bragging
about SpecInt for months now and hasn't said a word about SpecFP. Doesn't
that tell you something?
Joe Ragosta
Petrich) wrote:
>
> In the immediate future, we'll still be having a largely-emulated
> MacOS, but much of the time-critical parts are native, and Apple is making
> more and more of it native. It's next major release, System 8 / Copland,
> is supposedly to have 95% native PPC code, but that is probably
> excessively optimistic. However, they may be able to get at least 50%
> native in 8.0 (IMO, if they have any sense, they'd first try to make only
> the new stuff native, and if they have trouble with the old stuff, then
> leave that emulated for the time being).
I agree with most of this. However, I'll make an exception for the file
manager. This is a SERIOUS drag on performance and needs to be native-ized
immediately.
Joe Ragosta
NUSSBAUM) wrote:
> I know somebody that works in a lab and they have a couple of 6100/66
> as well as a dozen p5's 60's and 90's
> and he says that...well here is their exact words"- they ARE a joke :
> reaction of my colleagues when we got the PowerMac:
> "Why is it so slow" - and about Mac addicts "
> So please stop w/ the mac talk, it's a joke that laughes in your face
> every time you turn it on :-)
> TAKE CARE
> GUY NUSSBAUM
Then your friend has a problem with his PowerMac. No matter how much you
want to quibble about details, a PowerMac 6100 should at least be in the
same class as a P60. If it is noticeably slower, it hasn't been configured
properly.
In fact, almost every published study shows the PowerMac 6100/66 to
handily beat a P-60 and not to be that far behind a P90.
Get someone with a clue to look at the system.
Joe Ragosta
>
> How about waiting until hard data, obtained on real systems, are available?
> Too "unscientific" for a Mac advocate? :-)
>
> Dan
> --
> Dan Pop
> CERN, CN Division
> Email: Dan...@mail.cern.ch
> Mail: CERN - PPE, Bat. 31 R-004, CH-1211 Geneve 23, Switzerland
If you read the post, you would seed that the extrapolation was based on a
reported figure for the P6. While I can't vouch for the accuracy of the
number that was reported, I used it to extrapolate to equal clock speed
with the 604. Regardless, the number reported was a MEASURED number.
Is that too unscientific for a government employee?
Joe Ragosta
NUSSBAUM) wrote:
>
> Ok lets compare a mac and a pc!
> PC!!!!!
> 1)Windows 95!
> going to be 32bit O/S which w/ the use of 32bit applications promises
> to be twice as fast as win 3.1 runnig the same application in 16bit
> code!
So? MacOS has been 32 bit for almost a decade. I'm glad Windows is finally
catching up. As for performance....let's just wait and see.
> 2)P6!
> the first version of the p6 running at 133mhz promises to be twice as
> fast as the pentium 100.
Perhaps at integer calculations. Running applications? I doubt it VERY,
VERY much.
Joe Ragosta
> You are definitely an evangelist also. Because there is almost no factual
> information in your comment. According to SPEC, your comparisons are false.
> Trying to determine a 110 MHz pentium sounds way unscientific. What is it,
> a clock doubler, or tripler? That would also determine how fast it is since
> you have to get your timings correct between CPU and cache as well as cache
> and memory. I don't know where your theory comes from, but it is very
> suspect. Stick to SPEC ratings not evangelical Mac .advocacy ratings.
In theory, you're right. In practice, EVERYONE does it because SPEC
ratings are fairly close to linear functions of clock speed within a given
family.
I'll be happy to drop my extrapolation when you provide me with MEASURED
SPECfp values for the P6 at 133 MHz so we can compare it to the 604's
value at 132 MHz.
GUY NUSSBAUM
>can't see why everyone is so surprised by this result (I got more
>flames for this than for anything I've ever written). Intel has been
>bragging about SpecInt for months now and hasn't said a word about
>SpecFP. Doesn'tthat tell you something?
Yes, it tells me that the P6 performance is better then that of the 604
in specint92, and this is what really matters!!!
The real power of a processor is in it's specint not specFP!!!
The specint is the processors power on GENERAL applications, the Specfp
is on application that only use the FP, how many people have FP only
applications, well they are not in the higher precentage of the user's
that's for sure!
So if you are a scientist that needs a CPU that will give you a number
00000007ms faster then the P6 go for the 604 :-)
>--
>Regards, Joe Ragosta -- 100% Chemical and proud of it.
>
>Microsoft Network is prohibited from redistributing this work in any
>form,in whole or in part. Copyright, Joseph Ragosta, 1995. License to
Edmond Underwood
>In article <3rf52s$i...@natasha.rmii.com>, Edmond Underwood
><unde...@Colorado.Edu> wrote:
>
>> You lost me Joe. What are you comparing now? Are you saying that the 604
>> is over 40% faster in SPECfp than the Pentium or the P6? If it's the
>P6, where the
>> heck are you getting this information. Please let us know.
>
>This was based on SpecFP number posted by someone on the net. While I
>can't vouch for the accuracy of the figures, the numbers they posted
>showed the P6 to be a relative FP dog.
>
>What I did, was normalize to the same clock speed (since the data was at
>80 MHz or something ridiculous). Since the P6 and 604 run at similar clock
>speeds, this doesn't seem like too much of a stretch.
>
>can't see why everyone is so surprised by this result (I got more flames
>for this than for anything I've ever written). Intel has been bragging
>that tell you something?
>
>--
They didn't initially release the Pentium FP numbers either and one of its
main selling points is how much faster the FPU is than the 486. The fact
that they did not release it doesn't mean much to me. Based on its architeture,
I don't see it failing on FP instructioins and doing so well on integer
instructions. That really doesn't make a whole lot of sense. And I have
news for you, I found that SPECfp rating also and it is a hoax. Someone
made that up by interpolating the 33% increase from the P54c 100Mhz. This
is a very poor assumption. So, everything you wrote deserved the flames
it got because you trusted and advertised a bogus number.
Joe Ragosta
NUSSBAUM) wrote:
> Yes, it tells me that the P6 performance is better then that of the 604
> in specint92, and this is what really matters!!!
> The real power of a processor is in it's specint not specFP!!!
I suppose you can supply a reference for this? The P6 won't keep up with
the 604 in SPECint (although it's not too far behind).
> The specint is the processors power on GENERAL applications, the Specfp
> is on application that only use the FP, how many people have FP only
> applications, well they are not in the higher precentage of the user's
> that's for sure!
> So if you are a scientist that needs a CPU that will give you a number
> 00000007ms faster then the P6 go for the 604 :-)
Did you ever hear of graphics? Do you know why any PowerMac blows the
socks off Pentia for Photoshop?
Christopher C. Wood
|> Wade Heninger writes: [I think; the attributions may be munged...]
[ Someone's claim that Intel has the fastest processors challenged... ]
|> >Let's not get into another specwar, but I think we can all agree
|> >that the 604@133MHz is faster than the P5@120 (Intel's fastest
|> >chip on the market)
|> Is the 604 out yet?, NO i didn't think so!
Wake up, Rip Van Winkle! The PPC 604 has been in volume production
for quite some time; several months. In seven days you will be able
to buy a PowerPC 604 PCI-expansion bus Macintosh running at 132 MHz.
Chris
--
Speaking only for myself, of course.
Chris Wood chr...@lexis-nexis.com Chris...@aol.com
David T. Wang
: In article <BCOLWELL.95...@pdx145.intel.com>,
: bcol...@pdx145.intel.com (Robert Colwell) wrote:
: When comparing chips that are going to be available at similar or very
: close clock rates in similar time frames, there must be some validity to
: it. If 604 and P6 systems are both shipping with 133-MHz clocks, then
: knowing how much work either chip can get done, with those MHz, at what
: price, seems OK to me.
: If one forgoes that basis for comparison, what's left?
: chip price
: system price
: compatibility
: reliability
: upgradeability
: runs DOOM-2 really fast
: <insert your favorite criteria here>
: I agree with your complaint about oversimplifying, the point about being
: able to wring out more raw speed from an older process (and of course even
: more speed once you move to the newer processes, relative to P5) is very
: appealing.
: Rob Barris Quicksilver Software Inc. rba...@quicksilver.com
I have been told that Doom runs very very fast on the P6 :)
I will have to do a personal evaluation shortly :)
--
David Wang veak...@eng.umd.edu dave...@wam.umd.edu
Grad student- EE/Computer Engineering NeXT4Intel3.2/Linux/Dos(W4W)
Summer P6 guy @ Intel-Oregon. dw...@ichips.intel.com
http://www.wam.umd.edu/~davewang/ VLSI Design Automation Lab
I no speak for no Intel nor U of Md. Verstehen Sie?
David T. Wang
: David T. Wang (veak...@glue.umd.edu) wrote:
: : : > Let's let this sink in a moment, shall we. The current P5-90
: : : > is around 80 SPECfp92, is it not? The 100 MHZ 601 yields 125 SPECfp92.
: : no, go look up the correct spec marks please.
: If you think that these are wrong.. please explain why? We would
: all like to know the numbers you believe to be correct. :)
my apologies, I don't know what I was thinking, P5's specfp numbers are
below 1 per mhz, I was thinking of P6's
: : : > By the way the P6 is currently running at about 25 - 30 watts.
: : : The 604?
: : : > 13 watts! One other point, at the moment the P6 is about $1,800.00!
: : : > :)
: : This is really funny. Please state your sources, I'd like to look these
: : up, me thinks they're in error.
: Acctualy this is genneraly misleading. As it is true that the P6
: was running at 30 watts, it was only durring prototypeing and simulations.
: Now.. if they couldn't get it below 25 watts in simulations.. then how
: do we expect them to give us a chip that is any lower in real life?
: Anyone know what the real wattage input is now? I would assume that we
: won't be seeing any P6 notebooks very soon what ever it is. ;)
Yes they are lower, P6 notebooks should be out shortly after P6 desktops
are introduced. I'll wait for Intel to announce to you what the Wattages
are.
: : HA, Ha, Ha, Funniest thing I've seen so far. Joe predicts numbers for a
: : chip that he's never seen before, but predicts it from some rules of
: : thumb he created. Good one Joe.
: Very well Laughing boy.. let's hear what YOUR calculations are..
: no doubt they put the P6 ahead at a Specfp92 of 116million. :P
no, slightly lower than 116 million :P :P
: : : Of course, this also neglects the fact that you'll be able to buy a 604
: : : Mac in a week or so and the P6 boxes will be available about the same time
: : : as WIn95. (Any guesses?)
: : Good for Mac users, but P6's will be released shortly. There's a lot of
: : people working hard at it in Oregon :)
: And Apple, Motorola, and IBM will benefit by producing a chip
: that costs less than half that of the P6. Although it has been a
: tradition of sorts for Intel to release high prices on thier newest, most
: powerful chips, it is my oppinion that this price difference will play a
: big part in the PPC markets favor.
opinion. everyone has some, including me.
: : : We'll hear a lot of messages from Wintel fanatics about how FP doesn't
: : : really matter because most users don't need it (remember the Pentium FP
: : : fiasco?), so the P6 is still just as good as the 604.
: : Floats do matter. P6 will have decent numbers, don't want to baited into
: : speculating numbers along with you, but I'll make a wager with you, it'll
: : be , oh, slightly highers than 116 :) Have a nice day.
: Then.. what? Higher than the 604.. it'll have to be to compete.
: The P6 has been made out by Intel to be a god chip... something that
: will revolutionize the PC industry. I can't see it as that.. although
: Intel and it's advocates have done a good job at trying to make it seem
: as though it is. Intel's chips have never been much up on floating
: point... Motorola has always had the upperhand in that department. So we
: don't see Intel's new design being any different. And with the 604
: blowing away every other aspect of processing in addition to Motorola's
: consitent Floating point dominence, I believe the PowerPC (IBM, Apple,
: and others) will gain a substantial margin on the PC market, perhaps
: begining to sap some of the profits out of the big tippers like Digital
: and Compaq. Well... we'll see. :)
Floats are treated quite differently in 486/Pentiums than P6,
Architecturally, it is very different. All the dynamic data machine stuff
applies just the same to the Integer side as it does to the float side.
P6 is not a God chip. (That is yet to come :P ) P6 dynamically extracts
instructional parallism on the fly out of a rather inefficient
instruction set. As for Digital and Compaq supporting PPC, that is
unlikely, as Compaq still wants to keep IBM at bay because Compaq knows
that though IBM is down, it is still potentially its most potent
competitor. And why would Digital? most everyone knows that they've got
a better RISC chip already. It's a long road ahead for PPC chips.
They've got to be twice as fast a Intel just to entice people to switch
platforms, and that clearly isn't happening. And instead of presenting a
united front to fight Intel/Microsoft together, the IBM/Apple Alliance
spent a year squabbling amongst themselves. That CHRP/PReP thing
should've been out on day one. I'm no X86 fanatic, and programming in
assembly for x86 is really a pain, but I have always enjoyed the power of
an Open system, and the security of knowing that my software/hardware
will almost always work with the newest system out. Also being an
aspiring microprocessor architect, I'm very impressed with the
architectural advances that P6 has managed to come up with. Pentium is
architecturally more like the 486 than P6. I've yet to read up on
details of the 604 (read all about the 601 and 603) so I can't really
comment on it, but unless Moto/IBM/Apple can really get their act
together and get all the people to switch platforms, it may not matter in
a few more years as the newer generations of intel Chips come out when the
speedup advantages of the RISC architecture will probably be nullified.
Just my opinion.
: ditto. ;)
: -Kevin Stone
Jeff Medcalf (214) 280-5970 IBM SDO
> >The results is that the P6 is way behind the 604 in FP performance. I
> >can't see why everyone is so surprised by this result (I got more
> >flames for this than for anything I've ever written). Intel has been
> >bragging about SpecInt for months now and hasn't said a word about
> >SpecFP. Doesn'tthat tell you something?
>
> in specint92, and this is what really matters!!!
Actually, the specint numbers for P6 and PPC604 are:
P6@133, SPECint92 = 200 (approx. from Intel web page)
PPC604@132, SPECint92 = 212 (approx. from IBM web page, linearly translated
from 100MHz to 132MHz.)
So the P6 doesn't kill the 604 at SPECint92. Nice try, though.
> The real power of a processor is in it's specint not specFP!!!
> The specint is the processors power on GENERAL applications, the Specfp
> is on application that only use the FP, how many people have FP only
> applications, well they are not in the higher precentage of the user's
> that's for sure!
> So if you are a scientist that needs a CPU that will give you a number
> 00000007ms faster then the P6 go for the 604 :-)
Look around you, Guy, at the world. What is integer? Man-made abstractions
of the world. What is FP? Everything real: speech, vision, motion. You
can abstract these into integer math, but it's just a way of not coping
with the realities of the world. The beauty of bringing powerful FP
to the desktop is that you begin to open up new fields of programming
that were not feasible before. DOOM is nothing compared to the realistic
3D first-person games we'll see (probably from id :-) five years from
now. And the reason that it is possible to bring realistic sound, vision
and motion is because of very powerful FP math being done by today's CPUs.
--
+-------------+-------------------------------+--------------------------+
|Jeff Medcalf | med...@interlock.dfw.ibm.com | current hats: |
|CTG-Dallas | JKMe...@aol.com | PowerPersonal support |
+-------------+-------------------------------+ SVMP/Tech Eval |
|I not only don't speak for IBM, | Comm Mgr/2 |
|I don't even *know* who speaks for IBM! | |
+---------------------------------------------+--------------------------+
GUY NUSSBAUM
doc...@interramp.com (Joe Ragosta) writes:
>
>In article <3rhedt$e...@ixnews6.ix.netcom.com>, bro...@ix.netcom.com
(GUY
>> Yes, it tells me that the P6 performance is better then that of the
>> 604 in specint92, and this is what really matters!!!
>with the 604 in SPECint (although it's not too far behind).
You want some spec#'s ok
P6 604 620
MHZ 133 100 133
SPECint92 250* 160*(**) 225*(***)
*ESTIMATED
(**) using 1meg L2 cache
(***)No detail about system configuration given!
THE P6 estimates where taken form the intel's home on the net at
http://www.intel.com/
THE 604 and 620 estimates where taken from motorola at
http://www.mot.com/SPS/PowerPC/fs/620.html
>> The specint is the processors power on GENERAL applications, the
>>Specfp is on application that only use the FP, how many people have
>>FP only applications, well they are not in the higher precentage of
>>the user's that's for sure!
>>So if you are a scientist that needs a CPU that will give you a
>>number 00000007ms faster then the P6 go for the 604 :-)
>Did you ever hear of graphics? Do you know why any PowerMac blows the
>socks off Pentia for Photoshop?
Wait are you saying graphics applications only use the FP?
Do you have the exact numbers on this,?
AND don't even get into real world applications here, a 100mhz PowerMAC
w/ the use of the MacOS will perform on real world app's like a 75mhz
pent.
BTW I'm getting my facts where you get yours!!!
AND Yes Mr science I heard about graphics!
1)Only thing is I don't really picture is you as a scientist.
You see in sciense (except for the quantum world) there is no
speculation, I never seen the p6's FP numbers, and neither did you, but
all you MAC people keep on talking about how they are low
Give me some numbers or lets end this thread until the p6 comes out!
TAKE CARE
GUY NUSSBAUM
>Regards, Joe Ragosta -- 100% Chemical and proud of it.
>
>Microsoft Network is prohibited from redistributing this work in any
>form,in whole or in part. Copyright, Joseph Ragosta, 1995. License to
GUY NUSSBAUM
C. Wood) writes:
>
>In article <3rca11$q...@cwis.isu.edu>, kell...@cwis.isu.edu (Inconnu)
writes:
>>In article <3rbgt4$i...@ixnews5.ix.netcom.com>,
>>GUY NUSSBAUM <bro...@ix.netcom.com> wrote:
>
>>:So I would expect windows 95.1!!!! to be the best O/S ever! using
>>:32bit
>>Just curious-- what's the .1 (is there already a revision?, it hasn't
>>been released yet..)
>Just an extrapolation from the release of Windows 3.0...
Something like that!
>>:Is in a few short years, THE P6 and win95 is the future, and there
HEY, HEY ,HEY ,HEY, HEY (just remembered the movie beverly hills cop)
:)
Don't misquote what I said!!!!!!!!
The few short years was used for the comparison of the copland and the
windows platform!!!!!
Not as a prediction of when windows P6 will take over!!!
I can't believe you MAC people you Lie like a RUG :-)
>>:THE P6 and win95 is the future, and there is no stopping it
>Except that the PowerPC 604 is here _now_, and Macintoshes with
>PowerPC will be here in 7 days. The "next" x86 from Intel is _always_
>the processor that "is the future"...
>Just wait for the Wintel transition to RISC.
The P6 and windows is going to be out soon, AND un-like other
people(MAC people) I don't give out SPEC's and DATES where I don't know
them!!!
>Chris
>--
IS nobody talking to you? :-) (you see I can use these same dirty
tricks!)
>Speaking only to myself, of course.
>Chris Wood chr...@lexis-nexis.com Chris...@aol.com
GUY NUSSBAUM
C. Wood) writes:
>
>In article <3rcfoj$9...@ixnews3.ix.netcom.com>, bro...@ix.netcom.com
(GUY NUSSBAUM) writes:
>|> >Let's not get into another specwar, but I think we can all agree
>|> >that the 604@133MHz is faster than the P5@120 (Intel's fastest
>|> >chip on the market)
>
>|> Is the 604 out yet?, NO i didn't think so!
>
>Wake up, Rip Van Winkle! The PPC 604 has been in volume production
>for quite some time; several months. In seven days you will be able
>to buy a PowerPC 604 PCI-expansion bus Macintosh running at 132 MHz.
>
>Chris
Rip Van winkle???, is that the best you could do.
I'm not even going to go there, I'll forget that remark for now!
Chris? is the 604 132mhz out, NO!!!!!!
so what the **** are you talking about? WAKE up and stop talking to
yourself, you'll see that the MAC is nothing but a smiling face!
HERE you go talking to yourself again :-)
>Speaking only to myself, of course.
>Chris Wood chr...@lexis-nexis.com Chris...@aol.com
TAKE CARE
GUY NUSSBAUM
Stuart Woolford
bgr...@scf.nmsu.edu (Bruce Grubb) wrote:
> Macs are NOT made by JUST Apple any more; you have PowerComputing,
> Radius, DayStar, and two asian companies making Macs. Besides the MPC6xx
> line is a MOTOROLA chip not an Apple chip and is found in some high end
> in IBMs.
You mean 'and is found in some mid-low end in IBMs.' - or do you think IBM is
a PC company? (well, one LITTLE part of IBM is...)
> This is all in the Mac-IBM-compare sheet. Reading it saves one from
> having so much hoof and mouth problems.
Is this the same Mac-IBM compare sheet which has been around for years, and
keeps growing more and more ludicrous (sp?) each revision? (PS: I know it
is.. I recognise your name...) - have you made a SINGLE revision sugested
by Intel-based-PC people (carefully avoiding the misnomer IBM PC) yet?
I can remember your old avoidances of reality got quite laughable at times.
> >It is a fact that intel sells more chips then apple, so I would expect
> >the price of the P6 to drop way below that of the 604 and 620 (when the
> >p6 goes 210mhz)
And the 620 becomes reality. Speculation is free, perhaps we should say
when the P7 goes 400MHz ;)
> This does not hold true presently for the P5 vs. the 601. The 601 still
> is about 75% the cost of a P5 of the same MHz. Never mind that in real
> world tests by Ingram the 601 is 33% faster than a P5 of the same MHz.
> All the poorly optimized 386 and 486 code out there has made the 486dx4
> eclipse the Pentium in sales anyhow.
And how much does a 120MHz 601 cost? (hint: you are correct for SOME MHz
speeds, just not ALL)
How about comparing the costs of a Pentium SYSTEM versus an equivalent
PPC SYSTEM?
Thought so.
> >So if your comparing the two chips please I'll talk to you after the
> >chip has been selling for two months and discuss the prices w/ you!
> I'll be glad to talk to you about the P6 and 604 when you get you facts
> right as to who make the 6xx line of chips and who uses and will be using
> them.
Depends on how you consider 'makes'. IBM designed 90% of them. IBM+Motorola
modified them from POWER to PPC, IBM build the early ones (until Motorola
could build IBM designed fabs that could handle them), now I *believe*
motorola makes tham all (does IBM make the few it uses?)
Since you are so woried about this, would you like to list ALL the users
of both Motorola and Intel CPU's? Embeded applications included? I think
the previous poster was more woried about MAJOR PC VENDORS, or have the
Mac cloners got a large marker% already? (PS: there goes one of the last
Mac supposed advantages, the quality of Apple building, as they try to
compete with the cloners...)
Stuart Woolford, bor...@ibm.net
Robert Soderberg
>I really didn't know the big blue, and Texas Instruments where into RAM
I wonder who invented the 35ns 256Mb RAM chip?
Christopher C. Wood
|> In article <3rbgt4$i...@ixnews5.ix.netcom.com>,
|> GUY NUSSBAUM <bro...@ix.netcom.com> wrote:
|> >So I would expect windows 95.1!!!! to be the best O/S ever! using 32bit
|> Just curious-- what's the .1 (is there already a revision?, it hasn't been
|> released yet..)
Just an extrapolation from the release of Windows 3.0...
|> >Is in a few short years, THE P6 and win95 is the future, and there
|> >is no stopping it
Except that the PowerPC 604 is here _now_, and Macintoshes with
PowerPC will be here in 7 days. The "next" x86 from Intel is _always_
the processor that "is the future"...
Just wait for the Wintel transition to RISC.
Chris
--
Speaking only for myself, of course.
Chris Wood chr...@lexis-nexis.com Chris...@aol.com
Mike Schmit
If programmers were truly interested in "fast FP" they would actively
be looking at other methods of optimizing FP operations. For
example, Log Point (in Mt View or Palo Alto) has a patented
technology for performing FP operations with integers that is
faster than an FPU in most cases (cases being x86, PPC, Alpha,
Sparc, etc. ) There is a disadvantage (incompatible data format)
but the advantage is that the FPU can be thrown out (design-wise)
and is right in-line with the RISC philosophy. This is a FP
format designed by programmers, not hardware designers (like the
IEEE format).
But the fact is that the company does not have that many people
interested in its products. (I have no connection with this company).
Mike Schmit
-------------------------------------------------------------------
msc...@ix.netcom.com author:
408-244-6826 Pentium Processor Programming Tools
800-765-8086 ISBN: 0-12-627230-1
-------------------------------------------------------------------
Chad Irby
: Yes, it tells me that the P6 performance is better then that of the 604
: in specint92, and this is what really matters!!!
Gee, then you'll really love the PPC 620, which will be released in
machines at about the same time as the P6, and which should eat the P6
for lunch...
--
cirby
--
Chad Irby / My greatest fear: that future generations will,
ci...@gate.net / for some reason, refer to me as an "optimist."
David Kurtz
NUSSBAUM) wrote:
>In <docjoe-1206...@ip141.wilmington.de.interramp.com>
>doc...@interramp.com (Joe Ragosta) writes:
>>Did you ever hear of graphics? Do you know why any PowerMac blows the
>>socks off Pentia for Photoshop?
>Wait are you saying graphics applications only use the FP?
>Do you have the exact numbers on this,?
>AND don't even get into real world applications here, a 100mhz PowerMAC
>w/ the use of the MacOS will perform on real world app's like a 75mhz
>pent.
>BTW I'm getting my facts where you get yours!!!
>AND Yes Mr science I heard about graphics!
>1)Only thing is I don't really picture is you as a scientist.
>You see in sciense (except for the quantum world) there is no
>speculation, I never seen the p6's FP numbers, and neither did you, but
>all you MAC people keep on talking about how they are low
>Give me some numbers or lets end this thread until the p6 comes out!
Well I am a MAC pursen who is also a sciense studint at UCLA. I hop one
day to use the floteing point capabilaty of the PowerPCs to cintinue my
ed.. eduk.. edica... school.
With these figures below (which I got from my own web page,
http//www.lightside.com/~dkurtz), you can clearly see that the 604 simply
blows the P6 out of the water.
P6 604 620
MHZ 133 100 133
SPECint92 122* 349* i*
*ESTIMATED
Please note the stats for the 620, the chip that (according to my figures)
should redefine the face of computing. There it is, GUY NUSSBAUM:
incontestable fact (it's on the world wide web, after all).
Keep in mind that that your copmuter is worthless if it gets beaten in a
photoshop scroll-bar drag race or by a large bouncer named Larry. Never
sit on the sofa at noontime and may your bong-water never douse your
embers.
--
David A. Kurtz | Portions (c)1995 David A. Kurtz
da...@ucla.edu | The Microsoft Network is prohibited
dku...@lightside.com | from redistributing this work in any http://lightside.com/~dkurtz | form whatsoever.
GUY NUSSBAUM
don't have the time or the patience to answer any more of you wise guys
out there!
1)The P6 is not out yet, it's ESTIMATED performance are
GREATER then 200 specint92.
2)Integer operations make up the majority of instructions in typical
program, and just like RISC instructions they execute in a single clock
cycle this together with the use the P6's ability to execute 5
micro-ops in a clock cycle and the use of modern compliers which try
to use as many of these fast instructions as possible will make the P6
a speed demon!
**************************Bottom line**********************
THE use of win95 and P6 is going to be as fastest combination on the
market!
That's it!!!
And to all you wise guys out there, no matter what you write fact is We
pc users will enjoy the year in which we get better and faster while
you get multi-threaded 8.0 50% native, slow as hell 604's
BYE
GUY NUSSBAUM
Stephen Carter Morgan
>GUY NUSSBAUM (bro...@ix.netcom.com) wrote:
>: Yes, it tells me that the P6 performance is better then that of the 604
>: in specint92, and this is what really matters!!!
>Gee, then you'll really love the PPC 620, which will be released in
>machines at about the same time as the P6, and which should eat the P6
>for lunch...
Sorry, the PPC 620 isn't that much faster than the 604 in SpectInt or SpecFP.
Its primary advantage is that it is a 64-bit CPU rather than the 32-bit CPU.
The P6 and PPC 620 are meant for completely different systems. BTW, does the
620 still run at 25 W?
--
Stephen Carter Morgan
Georgia Institute of Technology, Atlanta Georgia, 30332
uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!gt4148b
Internet: gt4...@prism.gatech.edu
Aron S. Spencer
bro...@ix.netcom.com writes:
p6 620
MHZ 133 120
specint95 250* 225*(**)
*ESTIMATED PERFORMANCE
Actually, according to motorola's & intel's official information:
P6@133MHz: 200 SpecInt92
620@133MHz: 225 PecInt92
Sources:
http://www.intel.com/procs/p6/p6brief.html
http://www.mot.com/SPS/PowerPC/fs/620.html
Try sticking to reality, Guy.
Aron S. Spencer
Radix Consulting
(919) 596-0152 Voice
(919) 596-0081 Fax
aspe...@nando.net
AOL: Asspencer
CIS: 75407,125
James William Stepanek
user GUY NUSS...@ix.netcom.c writes:
>Rip Van winkle???, is that the best you could do.
>I'm not even going to go there, I'll forget that remark for now!
>Chris? is the 604 132mhz out, NO!!!!!!
>so what the **** are you talking about? WAKE up and stop talking to
>yourself, you'll see that the MAC is nothing but a smiling face!
>HERE you go talking to yourself again :-)
Now of course the concept of expecting a civil discussion from this
chump is a bit much, but facts would at least be nice. First he said the
604 wasn't out, now the 132 MHz 604 isn't out. Sorry but you've been
caught lacking as usual. The family of Mac PPC 604 machines to be
released later this month will include a 132MHz PPC 604 machine
(according to MacWeek of May 29). Gee where's your P6? Vaporware?
--------------------------------esp------------------------------------------
James Stepanek NRA,SAF Graduate Student
Materials Science and Engineering Carnegie Mellon University
God created mankind, Samuel Colt made them equal.
Thank God I'm an atheist.
These opinions are mine alone, others can find their own.
-------------------------Don't tread on me-----------------------------------
Christopher C. Wood
[ Bruce Grubb wrote about PowerPC processors found in high-end IBM
products... ]
|> You mean 'and is found in some mid-low end in IBMs.' - or do you
|> think IBM is a PC company? (well, one LITTLE part of IBM is...)
No, he meant high-end IBM products. I recall reading an announcement
from IBM, about a year ago, about new AS400 models; they are
multiprocessor powerPC boxes, and do IBM 370 emulation in software.
IBM said that they get better bang-for-the-buck this way than by
building mainframe processors...
Christopher C. Wood
[ PowerMac 9500 available on June 19th ]
|> No, you'll be able to order it. Actually getting your hands on one
|> might take a while...... Supposedly they're already backordered
|> 3-4 weeks.
|> And the PPC 604 Macs are ungodly expensive as well! ($4000 for
|> 16MB RAM, 1 GB HD, 4X CD-ROM, NO KEYBOARD, NO MONITOR, NO built-in
|> graphics capability is the price that's been bandied around.)
I see. The price is "too high", and there is at least a three-week
queue to buy one, before Apple has announced the machine or started
taking orders.
Which of these two conclusions would you draw: (1) that Apple products
are over-priced niche-market machines that are doomed to irrelevance
already because Microsoft has promised to ship a "bug-free" OS Upgrade
in two months, or (2) Apple products are the hottest, most productive
tools for professionals who know that spending $4000 now will pay them
back right away.
Christopher C. Wood
|> This is my last post in this subject, this is really flaming me, and I
|> don't have the time or the patience to answer any more of you wise guys
|> out there!
|> 1)The P6 is not out yet, it's ESTIMATED performance are
|> GREATER then 200 specint92.
The 604 is out; the numbers I've seen (160 SPECint92 @ 100 MHz; 120MHz
and 132MHz parts are shipping) should yield 213 SPECint92 @ 132MHz.
The floating-point difference will be bigger. Intel is six months
behind.
|> **************************Bottom line**********************
|> THE use of win95 and P6 is going to be as fastest combination on the
|> market!
|> That's it!!!
In 1996, you'll be able to use win95 and P6 and eat the dust of PPC
604 and PPC620 systems from Apple, IBM, and clone vendors...
|> slow as hell 604's
That are faster than P6s and out six months earlier...
|> BYE
That's the most intelligent thing you've said...
Rune Moberg
> There is the key word: Promises...M$ promises alot of things that they
> dont carry thru with.
Although in this case, there is an extensive betatest program going on, so
that most people can see for themselves. Those that aren't cheering for Win95
are simply envious the guys that got into the beta program.
Atleast MS will have 2 operating systems capable of pre-emptive multitasking,
while the Mac still haven't got a single one. Oh, that's right, when Apple
gets their act together, they might become PowerPC compatible and that way
they might be able to run Windows NT which indeed will give them pre-emptive
multitasking (and SMP support). oops.
=\
*=- R.Moberg, author of CD-Player Pro! ftp.cica.indiana.edu:
=/ /win3/sounds/cdppro45.zip
Roy W.Chartier
> I suppose you can supply a reference for this? The P6 won't keep up with
> the 604 in SPECint (although it's not too far behind).
Actually, you are quite mistaken. Although a 133 MHz 604's SPECint92
(estimated) is 5% faster than a 133 MHz P6 (also estimated); it is the 604's
3rd stepping. The P6 on the other hand, when it ships, will be its first
stepping.
What does this mean? That the 604 will peak near 200 MHz when clock-stepping
will no longer be economical. The P6, on the other hand has been designed for
optimal performance between the 200-300 MHz range, peaking likely at 350+
MHz.
Thus, the true competitor for the P6 is the 620, and not the 604.
Roy
Rob Barris
W.Chartier) wrote:
> Thus, the true competitor for the P6 is the 620, and not the 604.
You can say that as many times as you like - but the market only sees
prices and availability. The P6 has neither at the moment - so from that
viewpoint it makes a worthy competitor for the 620...
The 604 is an extremely high performer for its price. It promises to make
PowerPC machines even more competitive with Pentium or P6 based systems in
the very near term, both in the sense of raw performance and
price/performance. Some say >200 SPECInt with the 132MHz 604 - if it's
cheaper than P6 then it sounds like a P6 competitor to me.
Rob Barris Quicksilver Software Inc. rba...@quicksilver.com
* Opinions expressed not necessarily those of my employer *
Kurt Penrose
in article <david-13069...@ts6-3.westwood.ts.ucla.edu> David Kurtz wrote
>
>with these figures below (which I got from my own web page,
>http//www.lightside.com/~dkurtz), you can clearly see that the 604 simply
>blows the P6 out of the water.
>
>
> P6 604 620
>MHZ 133 100 133
>
>SPECint92 122* 349* i*
>*ESTIMATED
If this is the correct benchmark for the 604, then why does everyplace I look
quote a SPECint92 of 160? Even Motorola's own home page has it:
(at http://www.mot.com/SPS/PowerPC/fs/604.html)
PowerPC 604 Major Features
...
Performance: 100 MHz
SPECint92* 160
SPECfp92* 165
if You don't believe MOTOROLA, you can verify these numbers at two other
sights:
http://langmuir.eecs.berkeley.edu:80/www/peterm/processors.html
http://infopad.eecs.berkeley.edu/CIC/summary/summary.html
So now who is lying? Motorola or you? Its okay, everybody knows the truth!
They can look it up and see. You have no credibility when you lie! Were not
as stupid as you think here at comp.sys.intel!
SO I suppose the P6's SPECint92 of 200+ doesn't look so bad! Especially
since the PPC 620 is only estimated (again by Motorola) to be ~SPECint92 225.
Kurt!
(Like there is somebody out there that doesn't already know this info)
Edmond Underwood
>
>Actually, the specint numbers for P6 and PPC604 are:
>
>P6@133, SPECint92 = 200 (approx. from Intel web page)
>PPC604@132, SPECint92 = 212 (approx. from IBM web page, linearly translated
> from 100MHz to 132MHz.)
>
>So the P6 doesn't kill the 604 at SPECint92. Nice try, though.
>
PPC chip at 100 MHz = 140 SPECint
linearly translated to 133 MHz = 186.2
P6 chip estimated 133 MHz = greater than 200.
These are the real chip comparisons. I don't know what hardware you added
or what linear formula you used, but these are the CHIP per CHIP estimated
numbers. It really doesn't matter though, since SPEC does a little more
than just test the chip. So, I don't know what either complete system
numbers would give.
>> The real power of a processor is in it's specint not specFP!!!
>> The specint is the processors power on GENERAL applications, the Specfp
>> is on application that only use the FP, how many people have FP only
>> applications, well they are not in the higher precentage of the user's
>> that's for sure!
>> So if you are a scientist that needs a CPU that will give you a number
>> 00000007ms faster then the P6 go for the 604 :-)
>
>Look around you, Guy, at the world. What is integer? Man-made abstractions
>of the world. What is FP? Everything real: speech, vision, motion. You
>can abstract these into integer math, but it's just a way of not coping
>with the realities of the world. The beauty of bringing powerful FP
>to the desktop is that you begin to open up new fields of programming
>that were not feasible before. DOOM is nothing compared to the realistic
>3D first-person games we'll see (probably from id :-) five years from
>now. And the reason that it is possible to bring realistic sound, vision
>and motion is because of very powerful FP math being done by today's CPUs.
>
You ought to dump your current programs some time. Tell me how many times
you see the instructions fmul, fdiv, fadd, fistp, fld, fild, etc. Basically,
anything with an 'f' in front of it. This is a rarity. I dont see the need
to have every data variable estimated to the 0.xxx.... th place. You will
have to explain to me this need some time. 3-D rendering would some double
and float numbers, especially the high-end stuff like OpenGL. What other
kinds of apps could (other than obvious math oriented apps) really need
float and double variables?
>--
>+-------------+-------------------------------+--------------------------+
>|Jeff Medcalf | med...@interlock.dfw.ibm.com | current hats: |
>|CTG-Dallas | JKMe...@aol.com | PowerPersonal support |
>+-------------+-------------------------------+ SVMP/Tech Eval |
>|I not only don't speak for IBM, | Comm Mgr/2 |
>|I don't even *know* who speaks for IBM! | |
>+---------------------------------------------+--------------------------+
---------------------------------------------------------------------------------
Edmond Underwood
Systems Management Group
Computing & Network Services (University of Colorado)
E-mail: unde...@Colorado.Edu
James William Stepanek
user GUY NUSS...@ix.netcom.c writes:
>This is my last post in this subject, this is really flaming me, and I
>don't have the time or the patience to answer any more of you wise guys
>out there!
[translation: I'me getting my ass kicked so it's time to leave]
>1)The P6 is not out yet, it's ESTIMATED performance are
>GREATER then 200 specint92.
Ahh, the wonders of vaporware.
>2)Integer operations make up the majority of instructions in typical
>program, and just like RISC instructions they execute in a single clock
>cycle this together with the use the P6's ability to execute 5
>micro-ops in a clock cycle and the use of modern compliers which try
>to use as many of these fast instructions as possible will make the P6
>a speed demon!
Yes it's comparable to a PPC604 (which is already selling to consumers-
where are your P6s?)
>**************************Bottom line**********************
[shouldn't the bottom line be at the bottom?]
>THE use of win95 and P6 is going to be as fastest combination on the
>market!
Something from Microsoft that's fast?!? Right.
>That's it!!!
>And to all you wise guys out there, no matter what you write fact is We
>pc users will enjoy the year in which we get better and faster while
>you get multi-threaded 8.0 50% native, slow as hell 604's
As usual you know not of what you speak. Gee, Win 95 finally gives you
PC users an interface almost as good as ours only a few years late. You
also get about as much effectiveness in terms of PMT as we'll have in
sys 8 since most of the software available for your machines won't be
able to take advantage of the new features. Also Sys 8 is supposed to be
mostly native, much more than 50%.
Anyone who believes that a 604 is slow has a really funny world view. Oh
and since the P6 is nowhere to be seen why don't we compare to the 620
which will kick it's ass.
>BYE
>GUY NUSSBAUM
[Yipe yipe yipe, running off like a whipped dog with his tail between
his legs]