CPU MHz in A1200+
Wells Fargo Bank
my thoughts on this point.
It would make good business sense to me for AT to include a 50Mhz 030 with
the next generation of A1200 rather than a 40Mhz. For a ocuple of reasons.
1. The 030/50 is barely acceptable for a number of different applications
and is just beginning to reach slow 486 speeds.
2. Any 030 upgrade from 030/40 won't be enough to make it worthwhile or
cost effective for the end-user.
3. Using the 030/50 would allow purchasers of new Amigas to be able to
concentrate on 040 and 060 and PPC CPU upgrades which will benefit
the Amiga in the long run.
Please consider these things AT. An 030/40 seems to be a less productive
choice. If the cost is more, I an pretty sure the end users would be happy
to pay an extra $30 or so for the extra 10MHz of speed. I would!
PS. Consider including a full 68030 with MMU for the sake of runnic UNIX
and other things that require MMU. Now is the time to show what the Amiga
can do, not give just the bare minimum like Commodore always did. Oh, and
are you including the FPU socket on the A1200+ motherboard as I requested? :)
Good work, and God Bless! AT, I'm behind you all the way. Get a pooled
funded Advertising Campaign going in the US. You can continue the way you
are in the US and make a 'little' profit, or go for it, and make 'a lot' of
profit! Up to you guys. The A1200+ with the Surfer package could be your
first workable door into the US market. Mainstream Magazine Ads, and as some
have suggested, possibly an Infomercial.
Rob Bamford
(wfb...@netcom.com)
AMIGA - "Back for the Future!"
Michael van Elst
>It would make good business sense to me for AT to include a 50Mhz 030 with
>the next generation of A1200 rather than a 40Mhz. For a ocuple of reasons.
It doesn't really make any sense. The 50MHz version is MUCH more
expensive than the 40MHz version. Why ? It's a PGA package and it
is a full 68030. The 40MHz version is a small plastic package and
an 68EC030. There is no 50MHz EC030.
>1. The 030/50 is barely acceptable for a number of different applications
> and is just beginning to reach slow 486 speeds.
The speed difference doesn't justify the cost.
>2. Any 030 upgrade from 030/40 won't be enough to make it worthwhile or
> cost effective for the end-user.
Means they have to go for 040 or 060.
>3. Using the 030/50 would allow purchasers of new Amigas to be able to
> concentrate on 040 and 060 and PPC CPU upgrades which will benefit
> the Amiga in the long run.
Where is the logic in that sentence ? Do you mean with a 40MHz 030
they must upgrade first to a 50MHz one ? Why not start with a 7MHZ 030
then, would give you a hell of possible upgrades.
>Please consider these things AT. An 030/40 seems to be a less productive
>choice. If the cost is more, I an pretty sure the end users would be happy
>to pay an extra $30 or so for the extra 10MHz of speed. I would!
$30 ? Talk about $200.
--
Michael van Elst
Internet: mle...@serpens.rhein.de
"A potential Snark may lurk in every tree."
Wells Fargo Bank
: wfb...@netcom.com (Wells Fargo Bank) writes:
: >It would make good business sense to me for AT to include a 50Mhz 030 with
: >the next generation of A1200 rather than a 40Mhz. For a ocuple of reasons.
: It doesn't really make any sense. The 50MHz version is MUCH more
: expensive than the 40MHz version. Why ? It's a PGA package and it
: is a full 68030. The 40MHz version is a small plastic package and
: an 68EC030. There is no 50MHz EC030.
They also ave full 68030's at 40Mhz! I hope they use it and include a FPU
socket on the motherboard.
: >1. The 030/50 is barely acceptable for a number of different applications
: > and is just beginning to reach slow 486 speeds.
: The speed difference doesn't justify the cost.
This MAY be so... But I doubt it.
: >2. Any 030 upgrade from 030/40 won't be enough to make it worthwhile or
: > cost effective for the end-user.
: Means they have to go for 040 or 060.
: >3. Using the 030/50 would allow purchasers of new Amigas to be able to
: > concentrate on 040 and 060 and PPC CPU upgrades which will benefit
: > the Amiga in the long run.
: Where is the logic in that sentence ? Do you mean with a 40MHz 030
: they must upgrade first to a 50MHz one ? Why not start with a 7MHZ 030
: then, would give you a hell of possible upgrades.
The logic is that if the owner of a 030/40 wants more speed and sees that
a comparitively cheaper 030/50 accelerator is available versus a more
expensive 040 or 060 accelerator, they may be more prone to go for the
small increase for less money. Granted, my thought came from one particular
direction, and other considerations do need to be seen. Just my ideas...
Nobody HAS to impliment them or agree with them.
: >Please consider these things AT. An 030/40 seems to be a less productive
: >choice. If the cost is more, I an pretty sure the end users would be happy
: >to pay an extra $30 or so for the extra 10MHz of speed. I would!
: $30 ? Talk about $200.
I can purchase an 030/50 new for $99 +s&h. The difference from $99 and what
the 030/40 costs can't be much more than $30, and definetly no where near
$200! Come on now.
: --
: Michael van Elst
: Internet: mle...@serpens.rhein.de
: "A potential Snark may lurk in every tree."
Rob Bamford
(wfb...@netcom.com)
Michael van Elst
>They also ave full 68030's at 40Mhz! I hope they use it and include a FPU
>socket on the motherboard.
Same argument here. The full 68030 is much more expensive than the EC030.
>: The speed difference doesn't justify the cost.
>This MAY be so... But I doubt it.
There is strong evidence.
>I can purchase an 030/50 new for $99 +s&h. The difference from $99 and what
>the 030/40 costs can't be much more than $30, and definetly no where near
>$200! Come on now.
You simply forget that a) the prices YOU pay for the chips have nothing
to do with what AT has to pay for them (they pay less and probably won't
use refurbished parts), b) the prices they pay are a factor 3-5 lower
than what a customer has to be charged later and c) that the faster CPU also
requires faster memory and logic to be useful.
$200 is more like it.. believe me.
Regards,
Ross Deeley
> It would make good business sense to me for AT to include a 50Mhz 030 with
> the next generation of A1200 rather than a 40Mhz. For a ocuple of reasons.
>
> 1. The 030/50 is barely acceptable for a number of different applications
> and is just beginning to reach slow 486 speeds.
>
> 2. Any 030 upgrade from 030/40 won't be enough to make it worthwhile or
> cost effective for the end-user.
>
> 3. Using the 030/50 would allow purchasers of new Amigas to be able to
> concentrate on 040 and 060 and PPC CPU upgrades which will benefit
> the Amiga in the long run.
>
> Please consider these things AT. An 030/40 seems to be a less productive
> choice. If the cost is more, I an pretty sure the end users would be happy
> to pay an extra $30 or so for the extra 10MHz of speed. I would!
It may be $30 to AT, but by the time you add markups down the line it
will turn out say $100 to the purchaser - that could mean the difference
between somebody buying an Amiga or a PC. And the difference in
perfomance is not that noticeable.
> PS. Consider including a full 68030 with MMU for the sake of runnic UNIX
Generally the cheaper 40Mhz 030 is the EC model without the MMU, and the
50Mhz one does include a MMU, whioch is why they are quite a bit more
expensive.
The only reason I could see AT including 50Mhz full 030's was if Motorola
had a heap of them they couldn'r get rid of.
/=============================================================\
| Ross Deeley voice +646-877-5959 |
| |
| BASEBaud BBS 1200-14k4 +646-877-4294 |
| |
| ros...@inhb.co.nz /[][] Ross |
| |
| "The earth is but one country, and mankind its citizens." |
| -- Baha'u'llah |
| |
\=============================================================/
Lasse Olsen
: On Tue, 2 Jan 1996, Wells Fargo Bank wrote:
: > It would make good business sense to me for AT to include a 50Mhz 030 with
: > the next generation of A1200 rather than a 40Mhz. For a ocuple of reasons.
: >
: > 1. The 030/50 is barely acceptable for a number of different applications
: > and is just beginning to reach slow 486 speeds.
: >
: > 2. Any 030 upgrade from 030/40 won't be enough to make it worthwhile or
: > cost effective for the end-user.
: >
: > 3. Using the 030/50 would allow purchasers of new Amigas to be able to
: > concentrate on 040 and 060 and PPC CPU upgrades which will benefit
: > the Amiga in the long run.
: >
: > Please consider these things AT. An 030/40 seems to be a less productive
: > choice. If the cost is more, I an pretty sure the end users would be happy
: > to pay an extra $30 or so for the extra 10MHz of speed. I would!
: It may be $30 to AT, but by the time you add markups down the line it
: will turn out say $100 to the purchaser - that could mean the difference
: between somebody buying an Amiga or a PC. And the difference in
: perfomance is not that noticeable.
:
: > PS. Consider including a full 68030 with MMU for the sake of runnic UNIX
: Generally the cheaper 40Mhz 030 is the EC model without the MMU, and the
: 50Mhz one does include a MMU, whioch is why they are quite a bit more
: expensive.
: The only reason I could see AT including 50Mhz full 030's was if Motorola
: had a heap of them they couldn'r get rid of.
Could they have a heap of old 20MHz 040s lying around?
I'll say a prayer.
Cheers...
Ross Deeley
> : The only reason I could see AT including 50Mhz full 030's was if Motorola
> : had a heap of them they couldn'r get rid of.
>
> Could they have a heap of old 20MHz 040s lying around?
> I'll say a prayer.
I thought the slowest 040 was 25Mhz, and they were a bit hot to put in a
low profile case without a fan. (Of course if the "1200+/1300 is in a
pizza box it might be possible.)
Mind you we don't want any crippled (MMU less, FPU less) 040's that might
be used in washing machines...
Lasse Olsen
: On 4 Jan 1996, Lasse Olsen wrote:
: > : The only reason I could see AT including 50Mhz full 030's was if Motorola
: > : had a heap of them they couldn'r get rid of.
: >
: > Could they have a heap of old 20MHz 040s lying around?
: > I'll say a prayer.
: I thought the slowest 040 was 25Mhz, and they were a bit hot to put in a
: low profile case without a fan. (Of course if the "1200+/1300 is in a
: pizza box it might be possible.)
The first version of the 040 was cleared at 20MHz, which
is also the slowest stable running-speed of this processor.
Cheers...
Jyrki Saarinen
> Could they have a heap of old 20MHz 040s lying around?
> I'll say a prayer.
A 040/20 is not much faster than a 030/50.
I just can get it why faster Motorola CPUs cost so much
in Amigas, not Macs.. (MvE, please give some numbers)
I dont see how a 33MHz 3.3v EC or LC 040 could be expensive.
Maybe it is the stuff around it that is needed.
-- _
a Stellar programmer _ //
"Amiga - back for the future" \X/
Jyrki Saarinen
> I thought the slowest 040 was 25Mhz, and they were a bit hot to put in a
> low profile case without a fan. (Of course if the "1200+/1300 is in a
> pizza box it might be possible.)
There are low power 3.3v versions of the 68040.
> Mind you we don't want any crippled (MMU less, FPU less) 040's that
> might be used in washing machines...
What is the point of having a FPU in a low-end machine?
John Hendrikx
>> Could they have a heap of old 20MHz 040s lying around?
>> I'll say a prayer.
jS> A 040/20 is not much faster than a 030/50.
Infact, it is slower, except maybe at FPU stuff :-)
Grtz John
-- Via Xenolink 1.985B1, XenolinkUUCP 1.1
Michael van Elst
> jS> A 040/20 is not much faster than a 030/50.
>Infact, it is slower, except maybe at FPU stuff :-)
Unless the 040/20 system has other deficiencies (like slow memory)
it is most often a bit faster than the 030/50 and much faster
for FPU stuff.
Mski
>wfb...@netcom.com (Wells Fargo Bank) writes:
>>It would make good business sense to me for AT to include a 50Mhz 030 with
>>the next generation of A1200 rather than a 40Mhz. For a ocuple of reasons.
>It doesn't really make any sense. The 50MHz version is MUCH more
>expensive than the 40MHz version. Why ? It's a PGA package and it
>is a full 68030. The 40MHz version is a small plastic package and
>an 68EC030. There is no 50MHz EC030.
>>1. The 030/50 is barely acceptable for a number of different applications
>> and is just beginning to reach slow 486 speeds.
>The speed difference doesn't justify the cost.
>>2. Any 030 upgrade from 030/40 won't be enough to make it worthwhile or
>> cost effective for the end-user.
>Means they have to go for 040 or 060.
>>3. Using the 030/50 would allow purchasers of new Amigas to be able to
>> concentrate on 040 and 060 and PPC CPU upgrades which will benefit
>> the Amiga in the long run.
>Where is the logic in that sentence ? Do you mean with a 40MHz 030
>they must upgrade first to a 50MHz one ? Why not start with a 7MHZ 030
>then, would give you a hell of possible upgrades.
>>Please consider these things AT. An 030/40 seems to be a less productive
>>choice. If the cost is more, I an pretty sure the end users would be happy
>>to pay an extra $30 or so for the extra 10MHz of speed. I would!
>$30 ? Talk about $200.
>--
> Michael van Elst
>Internet: mle...@serpens.rhein.de
> "A potential Snark may lurk in every tree."
Where do you get your numbers from. $200 for 030/50 mayby 2-3 years
age but not now. I don know about now but 1 year ago i bought brand
new 030/50/mmu for $99 from: Memory Word
3392 Progres Dr.
Units B&C
Bansalem, PA 19020-5899
Attn: AmigaDept.
phon orders 215-244-7930
Gregory Scribner
> FPU
> >socket on the motherboard.
>
> Same argument here. The full 68030 is much more expensive than the
> EC030.
>
> >: The speed difference doesn't justify the cost.
> >This MAY be so... But I doubt it.
>
> There is strong evidence.
>
> >I can purchase an 030/50 new for $99 +s&h. The difference from $99
> and
> what
> >the 030/40 costs can't be much more than $30, and definetly no where
> near
> >$200! Come on now.
>
> You simply forget that a) the prices YOU pay for the chips have nothing
> to do with what AT has to pay for them (they pay less and probably
> won't
> use refurbished parts), b) the prices they pay are a factor 3-5 lower
> than what a customer has to be charged later and c) that the faster CPU
> also
> requires faster memory and logic to be useful.
>
> $200 is more like it.. believe me.
>
> Regards,
> Michael van Elst
>
> Internet: mle...@serpens.rhein.de
> "A potential Snark may lurk in every
> tree."
>
> * Origin: dis- (1:202/121.2)
> PATH: 202/121
>
>
However, since AT has to redesign the A1200 for a 40Mhz part(if a 40Mhz 68030
is used), I cannot see the cost of using a 50Mhz part to be more then the cost
difference of a faster CPU. I haven't really looked into it, but I would think
that a 10Mhz difference isn't going to require faster ram or logic than
needed for a 40Mhz part.
It also seems that people believe that any difference in the price of making a
computer will mean a major increase in price by the time it reaches the consumer.
Well, just for the hell of it I pick up a local computer magazine(Aug. 95 edition)
sitting next to me and thumbed through it. In it I found a local computer dealer
selling Pentium systems. The difference between a P90 and P75 with everthing
else being the same was $100. Flipping futher through the magazine I looked
at CPU prices. The diference between P75 and P90 was just about a hundred bucks.
So, do economics work different in the PC clone world or is everyone so concerned
about any increase in cost that they settle for less to satisfy a perceived price
increase to the consumer.
And can someone explain to me on a component level the cost differences
of a mini-tower A1200 and a wedge A1200?
Greg Scribner
| AmiQWK 2.9 - FREEWARE |
... I only shoot IBM's to put them out of their misery.
--
__
__ /// Fido: 1:202/122.0 -= The Mouse Trap BBS =-
\\X// 1:202/121.0 San Diego, CA USA
\X/ Data: (619)-464-2134 Serving Amiga Owners Since 1987
Ruud Dingemans
jS>> A 040/20 is not much faster than a 030/50.
JH> Infact, it is slower, except maybe at FPU stuff :-)
You may be right John, but how d'you measure the real-life
performance? AFAIK, 20 Mhz 040 computers or accelerators weren't made
(okay, maybe the Mac Centris 610) and 'estimates' aren't exactly hard
evidence.
Regards, Ruud
rdi...@grafix.xs4all.nl
-- Via Xenolink 1.981, XenolinkUUCP 1.1
Jyrki Saarinen
> jS> A 040/20 is not much faster than a 030/50.
>
With decent memory system I would expect a 040/20 to
be faster in integer mostly than a 030/50, but
not much.
Thomas Karlsen
> : The only reason I could see AT including 50Mhz full 030's was if Motorola
> : had a heap of them they couldn'r get rid of.
>
> Could they have a heap of old 20MHz 040s lying around?
> I'll say a prayer.
> Cheers...
>
>
In the end you'll see that they have included a 25Mhz 030 just to save
some bucks.. Wouldn't surprise me if they did..
Michael van Elst
>Where do you get your numbers from. $200 for 030/50 mayby 2-3 years
>age but not now. I don know about now but 1 year ago i bought brand
>new 030/50/mmu for $99 from: Memory Word
If AT had to pay $99 then the end user price would be even
more than $200. Besides, a 50MHz 030 also requires fast memory
which adds even more to the cost.
Michael van Elst
>However, since AT has to redesign the A1200 for a 40Mhz part(if a 40Mhz 68030
>is used), I cannot see the cost of using a 50Mhz part to be more then the cost
>difference of a faster CPU.
Again. This is NOT the price difference you will have to pay for
single parts.
Lasse Olsen
: > To clearify (hmm joyful reading ;)) the issue I droped by
: > Motorolas price-book today. (http://motserv.indirect.com/cgi-bin/pg)
: > Interresting to note is the 33Mhz version of the 020.
: Most interesting list :)
: The main point is that a 40MHz 68EC030 is cheaper by a factor of 3,
: mainly because the 40MHz 68030 just exists in PGA package.
Not by a factor three though. The plastic 030/40 is $48
in Qts of 36 - the EC040/20 is only $76 in Qts of 24.
: A 33MHz
: 68030 (in PLCC package) is just only slightly more expensive. So you
: could argue wether you want a 40MHz CPU without MMU or a 33MHz one
: with MMU. But it is a bit difficult to argue why a slower machine
: has to cost more.
Yep. :)
: Another interesting fact is that a 25MHz 040 is only slightly more
: expensive than a 50MHz 030 and a 33MHz LC040 is even cheaper.
: This basically rules out the 50MHz 030 for everything and the choice
: is between 33MHz 68030, 40MHz 68EC030 (about same cost) or a more
: expensive 040 version.
This probably has something to do with ATs choice of the 40Mhz.
: Also, the $100 more for the 040 are what the manufacturer would have to
: pay (well, probably less in large quantity). You can't expect that the end
: customer had to pay the same prices even when you just consider the cost
: of the CPU chip only (there is more... faster logic and RAM, cooling, etc).
You're right, but the price of the cheapest 040 is only %40
or so more than the very cheap 030/40.
I sure wouldn't mind spending a little bit more on a A1200/040.
Of course, I'm not sure of exactly how much extra glue logic
it would take to acchive this. And you'll loose the ability
to upgrade with a FPU. Might even turn out to be too hot.
Here's some more info on give-away 020s:
Device MRPS PAL C Qty Price Description
MC68EC020FG16 GGBB MAAR 1 66 15.9100 - 32-BIT EMBEDDED CTRL PQFP
MC68EC020FG25 GGBB MAAR 1 66 17.8100 - 32-BIT EMBEDDED CTRL PQFP
MC68EC020RP16 GGBB MAAR 2 1 41.9000 - 32-BIT EMBEDDED CTRL PPGA
MC68EC020RP25 GGBB MAAR 1 1 50.3200 - 32-BIT EMBEDDED CTRL PP
And some DSPs:
XC56001AFC27 DPAA TCCB 1 36 27.9000 - DSP24 BIT
XC56001AFC33 DPAA TCCB N 36 33.4900 - DSP24 BIT
XC56001AFE27 DPAA TCCB N 36 31.7000 - DSP24 BIT
XC56001AFE33 DPAA TCCB N 36 38.0100 - DSP24 BIT
XC56001ARC27 DPAA TCCB 1 14 70.8700 - DSP
XC56001ARC33 DPAA TCCB N 14 82.9800 - DSP
XC56002FC66 DPAA TCCJ 1 36 30.5400 - DSP-66 MHZ DSP56002 PQFP
XC56002PV40 DPAA TCCJ N 60 25.0600 - I/C MONOLITHIC40MHZ DSP
XC56002PV66 DPAA TCCJ N 60 29.0600 - DSP
XC56002PV80 DPAA TCCJ N 60 36.0100 - DSP I/C MONOLITHIC80 MHZ
XC56004FJ50 DPAA TCCQ N 84 19.0600 - DSP
XC56004FJ66 DPAA TCCQ N 84 22.0100 - DSP24-BIT AUDIO DSP
XC56005PV50 GGCC TCCR N 60 27.0600 - 56005 DSP
XC56301PW66 DPAA TCCM N 36 70.0200 - 24-BIT HIGH PERFORMCE DSP
DSP56001AFC27 DPAA TCCB N 1 27.9000 - DSP
DSP56001AFC33 DPAA TCCB N 1 33.4900 - DSP
DSP56001AFE27 DPAA TCCB N 1 31.7000 - DSP
DSP56001AFE33 DPAA TCCB N 1 38.0100 - DSP
DSP56001ARC27 DPAA TCCB N 1 70.8700 - DSP
DSP56001ARC33 DPAA TCCB N 1 82.9800 - DSP
DSP56002EVM DPZZ TCCX 1 1 149.950 - DSP
DSP56002FC40 DPAA TCCJ 1 36 24.0100 - MPG DSP
DSP56002FC66 DPAA TCCJ N 36 30.5400 - DSP
DSP56002RC40 DPAA TCCJ 2 14 74.5500 - 56002 MPG DSP
DSP56L002FC40 DPAA TCCJ 2 36 24.0100 - MPG DSP
DSP56004ADSA DPZZ TCCX 2 1 1500.08 - DSP
And the new super-fast DSP from Motorola:
DSP56300CLASB DPZZ TCCX N 1 300.020 - DSP-MAC (80MHz, soon 100Mhz)
Happy readin'.
Cheers...
Michael van Elst
> Not by a factor three though. The plastic 030/40 is $48
> in Qts of 36 - the EC040/20 is only $76 in Qts of 24.
Right. That's an XC chip though. The MC one costs a little
bit more and a version with MMU (LC040) is more like $120
> This probably has something to do with ATs choice of the 40Mhz.
That's what I think too. The EC030/40 is the best _cheap_ CPU
for an A1200 class machine.
> I sure wouldn't mind spending a little bit more on a A1200/040.
> Of course, I'm not sure of exactly how much extra glue logic
> it would take to acchive this. And you'll loose the ability
> to upgrade with a FPU. Might even turn out to be too hot.
Exactly.
> XC56002PV80 DPAA TCCJ N 60 36.0100 - DSP I/C MONOLITHIC80 MHZ
That's ultra cheap. I bought 56002's when they did cost $200 for the
33MHz version :)
> DSP56002EVM DPZZ TCCX 1 1 149.950 - DSP
This is the evaluation kit. DSP + 32k words RAM + 16bit stereo codec.
Nice thingie and I might have some Amiga software for it in the future.
Michael van Elst
> To clearify (hmm joyful reading ;)) the issue I droped by
> Motorolas price-book today. (http://motserv.indirect.com/cgi-bin/pg)
> Interresting to note is the 33Mhz version of the 020.
Most interesting list :)
The main point is that a 40MHz 68EC030 is cheaper by a factor of 3,
mainly because the 40MHz 68030 just exists in PGA package. A 33MHz
68030 (in PLCC package) is just only slightly more expensive. So you
could argue wether you want a 40MHz CPU without MMU or a 33MHz one
with MMU. But it is a bit difficult to argue why a slower machine
has to cost more.
Another interesting fact is that a 25MHz 040 is only slightly more
expensive than a 50MHz 030 and a 33MHz LC040 is even cheaper.
This basically rules out the 50MHz 030 for everything and the choice
is between 33MHz 68030, 40MHz 68EC030 (about same cost) or a more
expensive 040 version.
Also, the $100 more for the 040 are what the manufacturer would have to
pay (well, probably less in large quantity). You can't expect that the end
customer had to pay the same prices even when you just consider the cost
of the CPU chip only (there is more... faster logic and RAM, cooling, etc).
Regards,
Lasse Olsen
: > jS> A 040/20 is not much faster than a 030/50.
: >
: > Infact, it is slower, except maybe at FPU stuff :-)
: With decent memory system I would expect a 040/20 to
: be faster in integer mostly than a 030/50, but
: not much.
Even if it was the same speed the end consumer would
bennefit with the software developers free to
concentrate on the 040. Every programmer would be forced
to learn how to handle the 040 cache, FPU etc.
Cheers...
Lasse Olsen
: >Where do you get your numbers from. $200 for 030/50 mayby 2-3 years
: >age but not now. I don know about now but 1 year ago i bought brand
: >new 030/50/mmu for $99 from: Memory Word
: If AT had to pay $99 then the end user price would be even
: more than $200. Besides, a 50MHz 030 also requires fast memory
: which adds even more to the cost.
To clearify (hmm joyful reading ;)) the issue I droped by
Motorolas price-book today. (http://motserv.indirect.com/cgi-bin/pg)
Interresting to note is the 33Mhz version of the 020.
I've highlighted some (single or groups) potential Amiga-parts.
Definitions at the bottom.
----------
MOTOROLA SEMICONDUCTOR
NORTH AMERICAN *
PRICE BOOK
and
(Product Guide)
January 3, 1996
Device MRPS PAL C Qty Price Description
MC68020FC16E GGBB MAAS 2 36 38.0000 - 32-BIT MPU16MHZ PQFP
MC68020FC20E GGBB MAAS 1 36 45.1100 - 32-BIT MPU20MHZ PQFP
MC68020FC25E GGBB MAAS 2 36 49.8000 - 32-BIT MPU25MHZ PQFP
MC68020FC33E GGBB MAAS 2 36 59.3000 - 32-BIT MPU33MHZ PQFP
MC68020FE16E GGBB MAAS 2 36 53.0000 - NO NEW DESIGNS
MC68020FE20E GGBB MAAS 2 36 60.6000 - NO NEW DESIGNS
MC68020FE25E GGBB MAAS 2 36 67.8000 - NO NEWS DESIGNS
MC68020RC16E GGBB MAAS 1 1 125.800 - 32BIT MPU HCMOS16MHZ PGA
MC68020RC20E GGBB MAAS 1 1 135.480 - 32BIT MPU HCMOS20MHZ PGA
*MC68020RC25E GGBB MAAS 1 1 144.430 - 32BIT HCMOS MPU25MHZ PGA
MC68020RC33E GGBB MAAS 1 1 165.800 - 32BIT MPU HCMOS33MHZ PGA
MC68020RP16E GGBB MAAS 1 1 54.4200 - 2BIT MPU HCMOS16MHZ PGA
MC68020RP20E GGBB MAAS 1 1 55.7900 - 32BIT MPU HCMOS20MHZ PGA
MC68020RP25E GGBB MAAS 1 1 57.9000 - 32BIT MPU HCMOS25MHZ PGA
Device MRPS PAL C Qty Price Description
MC68030FE16B GGCC MACB 2 36 48.7500 - 32-BIT MPU16MHZ CQFP
MC68030FE16C GGCC MACB N 36 49.0900 - 32BIT ON-CHIP CACHE & MMU
MC68030FE20C GGCC MACB N 36 49.0900 - 32BIT,ON-CHIP CACHE & MMU
MC68030FE25C GGCC MACB N 36 63.7900 - 32BIT,ON-CHIP CACHE & MMU
*MC68030FE33C GGCC MACB N 36 63.7900 - 32BIT,ON-CHIP CACHE & MMU
MC68030RC16C GGCC MACB N 1 125.950 - 32BIT,ON-CHIP CACHE & MMU
MC68030RC20C GGCC MACB N 1 125.950 - 32BIT ON-CHIP CACHE & MMU
MC68030RC25C GGCC MACB N 1 143.510 - 32BIT ON-CHIP CACHE & MMU
MC68030RC33C GGCC MACB N 1 143.510 - 32BIT ON-CHIP CACHE & MMU
MC68030RC40B GGCC MACB 2 1 157.400 - 32BIT HCMOS MPU40MHZ PGA
*MC68030RC40C GGCC MACB N 1 158.490 - 32BIT,ON-CHIP CACHE & MMU
MC68030RC50B GGCC MACB 1 1 176.340 - 32BIT HCMOS MPU50MHZ PGA
MC68030RC50C GGCC MACB N 1 177.560 - 32BIT,ON-CHIP CACHE & MMU
MC68030RP16B GGCC MACB N 1 82.1100 - 32BIT,ON-CHIP CACHE & MMU
MC68030RP20B GGCC MACB N 1 82.1100 - 32BIT,ON-CHIP CACHE & MMU
MC68030RP25B GGCC MACB 2 1 95.2100 - 32BIT,ON-CHIP CACHE & MMU
*MC68030RP33B GGCC MACB 1 1 95.2100 - 32BIT,ON-CHIP CACHE & MMU
MC68837KBB GGDD MAFH N 1 70.0300 - FDDI ELM CHIP
MC68EC030FE25B GGCC MACB 1 36 34.0000 - 32BIT EMBEDDED CTRL CQFP
MC68EC030FE25C GGCC MACB 1 36 34.0000 - 32BIT,ON-CHIP CACHE & MMU
*MC68EC030FE40C GGCC MACB N 36 47.8800 - 32BIT,ON-CHIP CACHE & MMU
MC68EC030RP25B GGCC MACB 1 1 51.2200 - 32BIT EMBEDDED CTRL PGA
MC68EC030RP25C GGCC MACB N 1 34.2400 - HPESD MPU HIGH END
MC68EC030RP40B GGCC MACB 1 1 66.2900 - 32BIT EMBEDDED CTRL PGA
MC68EC030RP40C GGCC MACB N 1 47.8800 - HPESD MPU HIGH END
Device MRPS PAL C Qty Price Description
*MC68040FE25 GGCC MACC N 24 177.630 - 32BIT,FPU,CACHE,MMU
MC68040FE33 GGCC MACC N 24 212.880 - 32BIT,FPU,CACHE,MMU
MC68040RC25 GGCC MACC N 1 207.760 - 32BIT,FPU CACHE,MMU
MC68040RC33 GGCC MACC 1 1 244.910 - 32BIT,FPU,CACHE,MMU
MC68040RC40 GGCC MACC N 1 382.400 - 32BIT,FPU,CACHE,MMU
MC68882FN33A GGCC MACG 1 1 52.6800 - ENHANCED FPCP33MHZ PLCC
MC68EC040FE20 GGCC MACS N 24 77.0400 - 32BIT MPU EMBEDDED
MC68EC040FE25 GGCC MACS N 24 77.0400 - 32BIT MPU EMBEDDED
MC68EC040FE33 GGCC MACS N 24 108.100 - 32BIT MPU EMBEDDED
MC68EC040RC20 GGCC MACS N 1 99.2200 - 32BIT MPU EMBEDDED
MC68EC040RC25 GGCC MACS N 1 99.2200 - 32BIT MPU EMBEDDED
MC68EC040RC33 GGCC MACS N 1 131.860 - 32BIT MPU EMBEDDED
MC68EN360FE25C GGZZ MAFO 1 24 92.6800 - 68360 HI-END MPU
MC68HC16Z1CFC1 GALA TBLA 1 36 17.0400 - 16BIT MCU,RAM,10BIT A/D
MC68LC040FE20 GGCC MACS N 24 121.740 - 32BIT MPU EMBEDDED
MC68LC040FE25 GGCC MACS N 24 121.740 - 32BIT MPU EMBEDDED
MC68LC040FE33 GGCC MACS N 24 140.190 - 32BIT MPU EMBEDDED
MC68LC040RC20 GGCC MACS N 1 146.240 - 32BIT MPU EMBEDDED
MC68LC040RC25 GGCC MACS N 1 146.240 - 32BIT MPU EMBEDDED
MC68LC040RC33 GGCC MACS N 1 165.710 - 32BIT MPU EMBEDDED
Device MRPS PAL C Qty Price Description
XC68040FE25M GGCC MACC N 24 176.410 - MPG32-BIT MPU
XC68040FE33M GGCC MACC 2 24 211.410 - MPG32-BIT MPU
XC68040RC25M GGCC MACC 1 1 206.330 - HIGH END MPU
XC68040RC33M GGCC MACC 1 1 243.220 - MPG32-BIT MPU
XC68040RC40M GGCC MACC 1 1 379.750 - MPG32-BIT MPU
*XC68EC040FE20B GGCC MACS N 24 76.5100 - 68040, NO FPU, NO MMU
XC68EC040FE25B GGCC MACS 1 24 76.5100 - 68040, NO FPU, NO MMU
XC68EC040FE33B GGCC MACS 1 24 107.350 - 68040, NO FPU, NO MMU
*XC68EC040RC20B GGCC MACS N 1 98.5300 - 68040, NO FPU, NO MMU
XC68EC040RC25B GGCC MACS 1 1 98.5300 - 68040, NO FPU, NO MMU
XC68EC040RC33B GGCC MACS 1 1 130.960 - 68040, NO FPU, NO MMU
XC68HC711K4CFN GAGR TBHR 2 15 37.9400 - 8-BIT OTP,768RAM,A/D,EE
*XC68LC040FE20B GGCC MACS N 24 120.900 - 68040, NO FPU
XC68LC040FE25B GGCC MACS N 24 120.900 - 68040,NO FPU
XC68LC040FE33B GGCC MACS N 24 139.220 - 68040,NO FPU
*XC68LC040RC20B GGCC MACS N 1 145.230 - 68040, NO FPU
XC68LC040RC25B GGCC MACS 1 1 145.230 - 68040, NO FPU
XC68LC040RC33B GGCC MACS 2 1 164.570 - 68040, NO FPU
Device MRPS PAL C Qty Price Description
MC68882FN16A GGCC MACG 2 1 40.5100 - ENHANCED FPCP16MHZ PLCC
MC68882FN20A GGCC MACG 2 1 44.6100 - ENHANCED FPCP20MHZ PLCC
MC68882FN25A GGCC MACG 1 1 48.5600 - ENHANCED FPCP25MHZ PLCC
*MC68882FN33A GGCC MACG 1 1 52.6800 - ENHANCED FPCP33MHZ PLCC
*MC68882FN40A GGCC MACG 2 1 81.0700 - ENHANCED FPCP40MHZ PLCC
MC68882RC16A GGCC MACG 1 1 69.3200 - ENHANCED FPCP16MHZ PGA
MC68882RC20A GGCC MACG 2 1 69.3200 - ENHANCED FPCP20MHZ PGA
MC68882RC25A GGCC MACG 1 1 78.9500 - ENHANCED FPCP25MHZ PGA
*MC68882RC33A GGCC MACG 1 1 78.9500 - ENHANCED FPCP33MHZ PGA
MC68882RC40A GGCC MACG 2 1 103.690 - ENHANCED FPCP40 MHZ PGA
*MC68882RC50A GGCC MACG 2 1 112.320 - ENHANCED FPCP50 MHZ PGA
DEVICE SUFFIX DESIGNATORS
MPU, MCU, DSP AND MEMORIES
SUFFIX DEFINITION
A,B,C When followed by a package suffix, indicates
improved device or die shrink
C When followed by a package suffix, indicates
industrial temperature range (follows package
suffix for FSRAMs)
EM QFP 240-pin plastic.
FB PQFP (Plastic Quad Flat Pack) package.
FC,FG,F,U PQFP (Plastic Quad Flat Pack) package, gullwing
FE,FJ,FS CQFP (Ceramic Quad Flat Pack) package, gullwing
FM Molded Carrier Ring (MCR)
FN PLCC (Plastic Lead Chip Carrier) package
FT 28 x 28MM Quad Flat Pack (QFP)
I 0 to 85° C Temp Range
J SOJ package (350-mil for 4M DRAM, 300-mill
for all other memories)
KB Ceramic PGA w/ceramic lid
L Ceramic dual-in-line side-brazed package, or
for 48-pin devices or larger, refers to select
plate tin finish leads (microprocessors); SIP
(Single In-Line Package) package (memory
modules)
LC For 48-pin devices or larger, refers to full gold
finish leads.
LH Low-height SIP (Single In-Line Package) package
M -40 to 125° C
N Narrow SOJ package (new notation)
NJ Narrow SOJ package (old notation)
P Plastic dual-in-line package
R PGA (Pin Grid Array) package, select plate tin
finish leads
RC PGA (Pin Grid Array) package, full gold finish
leads.
RL Glass seal cceramic PGA (Pin Grid Array) package,
tinned leads
RP Plastic PGA (Pin Grid Array) package
R2 13" tape & reel indicator, used only following a
surface-mount package suffix
S Ceramic dual-in-line package (microprocessors);
SIMM (Single In-Line Memory Module) package,
gold pads
SH Low-hight SIMM (Single In-Line Memory Module)
package
SHG Low-Height SIMM (Single In-Line Memory Module)
package, gold pads
TH 16 x 16mm Quad Flat Pack (QFP)
V -40 to +105° C
WJ 400-mil SOF package for FSRAMs
Z Chip carrier package (microprocessors); Zig-zag
leaded module (memory modules)
ZB LCC (Leadless Chip Carrier) package
XC Pilot Production Prototype
-----------------
Cheers...
Lasse Olsen
: > I thought the slowest 040 was 25Mhz, and they were a bit hot to put in a
: > low profile case without a fan. (Of course if the "1200+/1300 is in a
: > pizza box it might be possible.)
: There are low power 3.3v versions of the 68040.
: > Mind you we don't want any crippled (MMU less, FPU less) 040's that
: > might be used in washing machines...
: What is the point of having a FPU in a low-end machine?
If you are talking as-low-as-you-get- end machine your point
is valid, but I think most people would not hesitate too
much with a +$50 bill for a decent FPU.
Think of what software developers would have to say about
having a FPU as standard.
Cheers...
Ruud Dingemans
>> jS> A 040/20 is not much faster than a 030/50.
>> Infact, it is slower, except maybe at FPU stuff :-)
MVE> Unless the 040/20 system has other deficiencies (like slow memory) it
MVE> is most often a bit faster than the 030/50 and much faster for FPU
Well, maybe. After I replied to John I did some digging: the (AFAIK)
only 20 Mhz 040 machine ever made was the Mac Centris 610; i looked it
up, and according to benchmark tests some 50 Mhz 68030/68882
accelerator boards for the Mac II were a little bit (but only a
little) faster than this machine. Not sure about the FPU side of
things, though.
Jyrki Saarinen
> If you are talking as-low-as-you-get- end machine your point
> is valid, but I think most people would not hesitate too
> much with a +$50 bill for a decent FPU.
Where "most people" use the FPU?
> Think of what software developers would have to say about
> having a FPU as standard.
What is the difference? Compiliers can produce FPU code
or non-FPU code.
Lasse Olsen
: >Michael van Elst (mle...@serpens.rhein.de) wrote:
: >: This is the evaluation kit. DSP + 32k words RAM + 16bit stereo codec.
: >: Nice thingie and I might have some Amiga software for it in the future.
: > By all means, please do.
: > I heard talk of some DSP boards comming for the Amiga some
: > time ago - are they still vapour-ware?
: AFAIK you could buy the Delfina board for some time.
Ah, that's right.
The Finish one with the 56002 at approx $800 (?)
Cheers...
Michael van Elst
>Michael van Elst (mle...@serpens.rhein.de) wrote:
>: This is the evaluation kit. DSP + 32k words RAM + 16bit stereo codec.
>: Nice thingie and I might have some Amiga software for it in the future.
> By all means, please do.
> I heard talk of some DSP boards comming for the Amiga some
> time ago - are they still vapour-ware?
AFAIK you could buy the Delfina board for some time.
--
Lasse Olsen
: On 9 Jan 1996, Michael van Elst wrote:
: > That's what I think too. The EC030/40 is the best _cheap_ CPU
: > for an A1200 class machine.
: >
: > > I sure wouldn't mind spending a little bit more on a A1200/040.
: > > Of course, I'm not sure of exactly how much extra glue logic
: > > it would take to acchive this. And you'll loose the ability
: > > to upgrade with a FPU. Might even turn out to be too hot.
: The best choice would be to have the A1200+ equipped with an 40Mhz EC030
: chip and a Processor Direct Slot like in the A3000/A4000 models. Then you
: can install a 040 or a PPC if you like, when you like. The PDS should be
: located near the CPU and the old trapdoor or similar slot should still be
: there aswell for other cards.
Forgive me if I'm misunderstanding you here, but, the A1200s
expansion-slot _is_ just that - a localbus, or PDS as you say.
Although it's only 150 pins as opposed to the A3/4000s 200.
Cheers...
nyle landas
: : >Where do you get your numbers from. $200 for 030/50 mayby 2-3 years
: : >age but not now. I don know about now but 1 year ago i bought brand
: : >new 030/50/mmu for $99 from: Memory Word
: : If AT had to pay $99 then the end user price would be even
: : more than $200. Besides, a 50MHz 030 also requires fast memory
: : which adds even more to the cost.
Keep in mind that was for a consumer to buy one from a small computer mail
order house. Imagine how cheap 50MHz 030s would be in quantity.
Considering Memory World probably only bought a few. Since Simms will be
onboard the A1200+ MB and 70ns RAM is standard selling now - I don't think
the RAM will cost any more.
-Nyle
--
---
Nyle F. Landas
ny...@servtech.com http://www.servtech.com/public/ami4000
ftp://ftp.servtech.com/pub/users/ami4000/
Amiga Users Page - http://www.servtech.com/public/ami4000/AmiUsers.html
Build a better mouse trap and Bill Gates will steal all the cheese.
Michael van Elst
>Keep in mind that was for a consumer to buy one from a small computer mail
>order house. Imagine how cheap 50MHz 030s would be in quantity.
Since you probably know that why don't you post figures ?
>Considering Memory World probably only bought a few. Since Simms will be
>onboard the A1200+ MB and 70ns RAM is standard selling now - I don't think
>the RAM will cost any more.
It is basically the difference between 70ns RAM and 60ns RAM. Otherwise
your 50MHz 030 will be most often _slower_ than the 40MHz one.
Michael van Elst
> Ah, that's right.
> The Finish one with the 56002 at approx $800 (?)
$800 ? Can't remember that. Wasn't it more like $400 ?
Thomas Karlsen
> That's what I think too. The EC030/40 is the best _cheap_ CPU
> for an A1200 class machine.
>
> > I sure wouldn't mind spending a little bit more on a A1200/040.
> > Of course, I'm not sure of exactly how much extra glue logic
> > it would take to acchive this. And you'll loose the ability
> > to upgrade with a FPU. Might even turn out to be too hot.
The best choice would be to have the A1200+ equipped with an 40Mhz EC030
chip and a Processor Direct Slot like in the A3000/A4000 models. Then you
can install a 040 or a PPC if you like, when you like. The PDS should be
located near the CPU and the old trapdoor or similar slot should still be
there aswell for other cards.
/Thomas
Lasse Olsen
: In a message of 07 Jan 96 Michael Van Elst wrote to All:
: >> jS> A 040/20 is not much faster than a 030/50.
: >> Infact, it is slower, except maybe at FPU stuff :-)
: MVE> Unless the 040/20 system has other deficiencies (like slow memory) it
: MVE> is most often a bit faster than the 030/50 and much faster for FPU
: Well, maybe. After I replied to John I did some digging: the (AFAIK)
: only 20 Mhz 040 machine ever made was the Mac Centris 610; i looked it
: up, and according to benchmark tests some 50 Mhz 68030/68882
: accelerator boards for the Mac II were a little bit (but only a
: little) faster than this machine. Not sure about the FPU side of
: things, though.
That's one major point. A 040/20 would be much faster in FPU.
Even more interresting it becomes if we consider an imagenary
A4000-030/50 including the 50Mhz FPU.
This turns out to be even more expensive than a fully fledged 040/33.
Cheers...
Glenn Saunders
: Hardly. If you want an expandable machine you have to buy one with
: slots.
A machine that can't expand hasn't much of a market these days anyway IMHO.
Lasse Olsen
: > Not by a factor three though. The plastic 030/40 is $48
: > in Qts of 36 - the EC040/20 is only $76 in Qts of 24.
: Right. That's an XC chip though. The MC one costs a little
: bit more and a version with MMU (LC040) is more like $120
Yep.
But many A4000/040s were XCs - no problemo.
: > This probably has something to do with ATs choice of the 40Mhz.
: That's what I think too. The EC030/40 is the best _cheap_ CPU
: for an A1200 class machine.
Yes. I belive an 030/50 with FPU would be a bad bet costwise.
The full 040/33 would be cheaper.
: > XC56002PV80 DPAA TCCJ N 60 36.0100 - DSP I/C MONOLITHIC80 MHZ
: That's ultra cheap. I bought 56002's when they did cost $200 for the
: 33MHz version :)
That's life. :)
: > DSP56002EVM DPZZ TCCX 1 1 149.950 - DSP
: This is the evaluation kit. DSP + 32k words RAM + 16bit stereo codec.
: Nice thingie and I might have some Amiga software for it in the future.
By all means, please do.
I heard talk of some DSP boards comming for the Amiga some
time ago - are they still vapour-ware?
Cheers...
Michael van Elst
>The best choice would be to have the A1200+ equipped with an 40Mhz EC030
>chip and a Processor Direct Slot like in the A3000/A4000 models.
Huh ? The trapdoor _is_ a CPU slot.
>located near the CPU and the old trapdoor or similar slot should still be
>there aswell for other cards.
Hardly. If you want an expandable machine you have to buy one with
slots.
--
G. Baldwin
: Jyrki Saarinen (jsaa...@kone.fipnet.fi) wrote:
: : > I thought the slowest 040 was 25Mhz, and they were a bit hot to put in a
: : > low profile case without a fan. (Of course if the "1200+/1300 is in a
: : > pizza box it might be possible.)
: : There are low power 3.3v versions of the 68040.
: : > Mind you we don't want any crippled (MMU less, FPU less) 040's that
: : > might be used in washing machines...
: : What is the point of having a FPU in a low-end machine?
: If you are talking as-low-as-you-get- end machine your point
: is valid, but I think most people would not hesitate too
: much with a +$50 bill for a decent FPU.
: Think of what software developers would have to say about
: having a FPU as standard.
: Cheers...
Speak of the devil, didn't I remeber C= making a machine called the
A300? C= wanted to make the cheapest computer they could, because
people out there didn't want a used cheap computer, they wanted a new
one. Too bad that this idea couldn't be done, and that A300 was
renamed the A600.
Point one - the Amiga line has some real tightwads out there, who
would bitch if you put in a bbu clock, let alone an FPU. Lets get an
MMU standard before we get an FPU in there...
Point two - the memory subsystem of an 040 (standard, lowcost,
economy, or cold fire) is a LOT more complicated than that of an 030.
Case in point - look at the 3640 (Commodore's 040 card for the
A3k/A4k). It has a memory bus from hell, and it was STILL expensive
to make. How do you expect to make a low end computer when you have
to deal with that? You can't.
Point three - the A1200 MB is designed to handle an 030 (the EC020 was
a last minute hack by C= to lower the cost). One would have to think
that it would be better if Escom/AT just slapped an 030/33 or 030/50
in there (030/40 has no MMU - we need that for VMM in AmigaOS 4.x)
instead of worrying about designing a new 040 MB. I would much rather
see the next low end machine after the A1200+ have a PPC 603 (the
laptop/economy PPC chip) than have a clunky old 040 or expensive 060.
Any complaints?
--Greg Baldwin (dri...@eskimo.com)--
*Avid Amiga user & user since 1987*
*Tyranical EfNet #Amiga OP 'Drizzit'*
Thomas Tavoly
In article <4d1tjc$g...@nnrp1.news.primenet.com> Glenn Saunders <kri...@primenet.com> writes:
> The mortal Michael van Elst wrote:
> : slots.
>
If AT has any sense they will stop the gap between the A1200 and A4000T
with an expandable desktop machine with detached keyboard. If they have real
sense they might implement the A1000 keyboard bay idea :)
. Thomas Tavoly + A4000 = aTm...@amiga.ow.nl _ .
. WWW - http://www.cistron.nl/~ttavoly _ // .
. Finger - tta...@cistron.nl [.sig v4.0] \X/ .
... >>> Amiga - Back for the future! <<< ...
Bernd Bernie Meyer
>Thomas Karlsen (sle...@algonet.se) wrote:
>: The best choice would be to have the A1200+ equipped with an 40Mhz EC030
>: chip
Yeah, great. Say goodbye to protected memory and virtual memory. And to
good floaing point performance....
>:and a Processor Direct Slot like in the A3000/A4000 models. Then you
>: can install a 040 or a PPC if you like, when you like.
And I bet you'd just LOVE to have a PPC with some sort of horrendous glue
logic to make it look like an 030 to the rest of the world. Which means
32 bit data paths. And unless you include it on the add on module, no
cache. GREAT performance is to be expected from a RISC CPU with horrible
memory badnwidth....
Bernie
--
==============================================================================
I'm using an experimental version of "nn" to read news. So if anything seems
strange about the headers of this mail or post, please inform me. Thanks!
Lasse Olsen
: : : > I thought the slowest 040 was 25Mhz, and they were a bit hot to put in a
: : : > low profile case without a fan. (Of course if the "1200+/1300 is in a
: : : > pizza box it might be possible.)
: : : There are low power 3.3v versions of the 68040.
: : : > Mind you we don't want any crippled (MMU less, FPU less) 040's that
: : : > might be used in washing machines...
: : : What is the point of having a FPU in a low-end machine?
: : If you are talking as-low-as-you-get- end machine your point
: : is valid, but I think most people would not hesitate too
: : much with a +$50 bill for a decent FPU.
: : Think of what software developers would have to say about
: : having a FPU as standard.
: : Cheers...
: Speak of the devil, didn't I remeber C= making a machine called the
: A300? C= wanted to make the cheapest computer they could, because
: people out there didn't want a used cheap computer, they wanted a new
: one. Too bad that this idea couldn't be done, and that A300 was
: renamed the A600.
Whelp.. ;)
: Point one - the Amiga line has some real tightwads out there, who
: would bitch if you put in a bbu clock, let alone an FPU. Lets get an
: MMU standard before we get an FPU in there...
There is no reason as to why both couldn't be implemented -
except a minute cost-increase.
: Point two - the memory subsystem of an 040 (standard, lowcost,
: economy, or cold fire) is a LOT more complicated than that of an 030.
: Case in point - look at the 3640 (Commodore's 040 card for the
: A3k/A4k). It has a memory bus from hell, and it was STILL expensive
: to make. How do you expect to make a low end computer when you have
: to deal with that? You can't.
Might be a point. But, the 3640 was designed my a team
having less-than-satisfactory experience with 040s -
more than 4 years ago.
This is 1996.
: Point three - the A1200 MB is designed to handle an 030
Sidenote: So is the A4000 MB.
: (the EC020 was
: a last minute hack by C= to lower the cost). One would have to think
: that it would be better if Escom/AT just slapped an 030/33 or 030/50
: in there (030/40 has no MMU - we need that for VMM in AmigaOS 4.x)
: instead of worrying about designing a new 040 MB. I would much rather
: see the next low end machine after the A1200+ have a PPC 603 (the
: laptop/economy PPC chip) than have a clunky old 040 or expensive 060.
If ADos 4 is PPC only (as it looks like today) a MMU in
the 030 will be of no value anyway, so forget the 33Mhz.
The 50Mhz is far too expensive if Motorola doesn't wrap it
in a new package.
: Any complaints?
Nah... :)
Cheers...
Lasse Olsen
: >Thomas Karlsen (sle...@algonet.se) wrote:
: >: The best choice would be to have the A1200+ equipped with an 40Mhz EC030
: >: chip
: Yeah, great. Say goodbye to protected memory and virtual memory. And to
: good floaing point performance....
First of all, nice to see you again Bernd - to bad I'll
be gone in a couple of days. :)
Second, I saw you included my sig up there although I didn't
write the above.
So, to the matter.
Why is the EC030 a bad choice for the A1200+?
Prot. mode and MMU wont be necessary until the new OS anyway.
Do you have a better idea?
: >:and a Processor Direct Slot like in the A3000/A4000 models. Then you
: >: can install a 040 or a PPC if you like, when you like.
: And I bet you'd just LOVE to have a PPC with some sort of horrendous glue
: logic to make it look like an 030 to the rest of the world. Which means
: 32 bit data paths. And unless you include it on the add on module, no
: cache. GREAT performance is to be expected from a RISC CPU with horrible
: memory badnwidth....
Nope, I'll all be included on the accelerator card.
The only part which might suffer a bit is the internal
IDE interface. But by then, you'll probably go with the
on-board FSCSI-II anyway.
Cheers...
Shane Kuntz
|> Thomas Karlsen <sle...@algonet.se> writes:
|>
|> >The best choice would be to have the A1200+ equipped with an 40Mhz EC030
|>
|> Huh ? The trapdoor _is_ a CPU slot.
|>
|> >located near the CPU and the old trapdoor or similar slot should still be
|> >there aswell for other cards.
|>
What AT should look into is a standard slot that the 1200/4000 can both use.
It may increase the size of the trap door in the 1200 but it would cut the
price of upgrade cards as developers would only have to develop one standard.
I would also recommend that whichever processor they go with be on the
expansion board. When you wanted to upgrade you trade in your old board
and buy the new one minus the trade up offer. This would benefit users,
because the upgrade path would be cheaper with the trade up and it would
benefit AT because they would have a stock of processor boards they could
throw into new machines (eliminating manufacturing of new processor boards
every time) and offer them at a discount.
One last thing I would do is put a option in the 1200 to hook up an external
mini-tower design with a few zorro/pci slots. Similar to the optional port the
1200's have on the back now (gvp used it for an external scsi connector).
Just my 10 cents.
|> Hardly. If you want an expandable machine you have to buy one with
|> slots.
|>
With my options.. one could buy a 1200 and expand it into near 4000t
capability. Also, one could trick out a 1200 with a PPC board and then
buy a 4000T with no processor and use the processor board out of the 1200
when/if he/she wanted to.
Oop.. make that 15 cents.
Shane
nyle landas
: >Keep in mind that was for a consumer to buy one from a small computer mail
: >order house. Imagine how cheap 50MHz 030s would be in quantity.
: Since you probably know that why don't you post figures ?
Do you honestly believe that AmiT ordering 100,000 030s is going to pay the
same as MEmory World ordering 100 or less? I don't have the figures but
volume discounts are common on computer parts especially. Prices are also
less to resellers than to consumers.
: >Considering Memory World probably only bought a few. Since Simms will be
: >onboard the A1200+ MB and 70ns RAM is standard selling now - I don't think
: >the RAM will cost any more.
: It is basically the difference between 70ns RAM and 60ns RAM. Otherwise
: your 50MHz 030 will be most often _slower_ than the 40MHz one.
Why do they ship so many High end PC compatibles with 70ns then? Just
curious.
-Nyle
--
----
Nyle F. Landas http://www.servtech.com/public/ami4000/
ny...@servtech.com ftp://ftp.servtech.com/pub/users/ami4000/
Amiga Users Page - http://www.servtech.com/public/ami4000/AmiUsers.html
Amiga Dealers Page - http://www.servtech.com/public/ami4000/AmiDlrs.html
Build a better mousetrap and Bill Gates will steal all the Cheese.
Jyrki Petsalo
>> The Finish one with the 56002 at approx $800 (?)
> $800 ? Can't remember that. Wasn't it more like $400 ?
Yes, currently it is $400 plus VAT (22% incide EC, VAT-free for
companies) plus postage. Price reduction is directed for software
developers, but others can make use of it too.
-JP
----------------------------------------------------------------------------
Delfina DSP. Discover the Sound of Power.
Jyrki Petsalo
expansion-slot _is_ just that - a localbus, or PDS as you say.
Although it's only 150 pins as opposed to the A3/4000s 200.
Cheers...
Is A1200 expansion slot pin list available anywhere?
-JP
Wells Fargo Bank
: Michael van Elst (mle...@serpens.rhein.de) wrote:
: : ny...@cyber1.servtech.com (nyle landas) writes:
: : >Keep in mind that was for a consumer to buy one from a small computer mail
: : >order house. Imagine how cheap 50MHz 030s would be in quantity.
: : Since you probably know that why don't you post figures ?
: Do you honestly believe that AmiT ordering 100,000 030s is going to pay the
: same as MEmory World ordering 100 or less? I don't have the figures but
: volume discounts are common on computer parts especially. Prices are also
: less to resellers than to consumers.
: : >Considering Memory World probably only bought a few. Since Simms will be
: : >onboard the A1200+ MB and 70ns RAM is standard selling now - I don't think
: : >the RAM will cost any more.
: : It is basically the difference between 70ns RAM and 60ns RAM. Otherwise
: : your 50MHz 030 will be most often _slower_ than the 40MHz one.
: Why do they ship so many High end PC compatibles with 70ns then? Just
: curious.
The bus which the PC's usually access memory from are only upwards of 33MHz
which 70ns RAM is plenty capable of handling. This is my understanding.
You will sometimes see 60ns RAM on some PC's and I always wonder if it is
actually taking advantage of the faster RAM. On the Amiga, I regularly
hear people saying that if you are going to go with a CPU faster than a
33Mhz 030 that you should get 60ns RAM... Obviously the Amiga can access
RAM faster than a PC somewhere.
Also, keep in mind that a 486DX2-66 is only being accessed from the Mother
Board at 33MHz... Still within the 70ns range. Same with 486DX4-100, but
its only 25MHz in this case. A 486DX4-120 is 30MHz and so on. The only
possible exception I am aware of is the AMD (I believe) 486DX2-80 which
would access at 40MHz and would seemingly be able to take advantage of
60ns RAM. Obviously, I am not aware of how this compares in the Pentium
realm, but I still only see systems selling with 70ns RAM with an
occasional 60ns, again, is it taking advantage of the faster RAM?
Basically, it appears that the Amiga has the ability to access RAM faster
IF a faster CPU is being used than the typical PC does.
Rob Bamford
(wfb...@netcom.com)
AMIGA - "Back for the Future!"
They going to make the RAM slots in the new A1200 variant EDO compliant?
Peter Flynn
> The only reason I could see AT including 50Mhz full 030's was if
Motorola
> had a heap of them they couldn'r get rid of.
If they can't produce an Amiga 1200+ with a full 50 mhz 030 with MMU then
they should do an 040 instead.
Charge a bit more but at least give the enthusiasts something to upgrade
to. (How about both a 40 mhz ec version and another version with 50 mhz
68030 and 68882?)
There are a lot of A1200 owners with fast ram trapdoor expansions who may
want an A1200+ with faster cpu, mmu, coprocessor, WB3.1 and CDROM. In
short
all the things an A1200 with fastram presently lacks.
If the A1200+ is a 40mhz 030 without mmu, the potential upgrader will
probably be better off expanding an old A1200 by adding a 50 mhz 030 or a
25 or 33mhz 040 (or 060) that are now coming on the market.
The potential upgrader can add a better processor, WB3.1 and a CDROM
anytime
finances allow. The end result is the third party suppliers making sales
and
not AT.
Some A1200 owners have had 50 mhz 030s since 1993. Something as good or
better is expected in a "+" model. It is more than 3 years since the
release of the A1200.
If high cost of 50mhz 030 is such a problem then it may be better to keep
the 020. It may be possible to switch to a 28mhz 020 (not much slower
than
a 40mhz 030) and concentrate on reducing production costs by custom chip
integration, substitution of 3.5" hard drives for 2.5" and anything else
that might reduce costs in the shops.
* AmyBW v2.13 *
~~~ Blue Wave/RA
Bernd Bernie Meyer
> I would also recommend that whichever processor they go with be on the
> expansion board. When you wanted to upgrade you trade in your old board
> and buy the new one minus the trade up offer. This would benefit users,
> because the upgrade path would be cheaper with the trade up and it would
> benefit AT because they would have a stock of processor boards they could
> throw into new machines (eliminating manufacturing of new processor boards
> every time) and offer them at a discount.
There is two problems with this approach:
a) Prices drop through the floor REAL FAST. Since 1990, I have quite often
spent about $700 on a PC processor board. This got me the "near the top
of the line board", any year. Usually, the resale value of my old board
was about $100 or $150. In the private market, without any administration,
storage or markup costs.
Now think about AT buying back old CPU boards --- they have to pay all
those costs, so all you will probably get for your old board, even if
it was pretty recent, is in the $50 to $100 ballpark. Which really isn't
worth the bother.
Also, many of these old CPU modules would never be sold again --- would you
buy a 286 motherboard these days? Not in a time where you have to pay to
have them disposed properly....
b) The PC world has been doing exactly that for quite some time. The "processor
card" is called a motherboard, but up until recently, a processor card
plus a bus was all it was. Nowadays, harddisk and floppy interfaces are
included, as are standard IO ports. But that's a pretty new development.
Thomas Tavoly
In article <DL0M6...@wombat.hanse.de> ro...@wombat.hanse.de (Bernd "Bernie" Meyer) writes:
> lol...@hsr.no (Lasse Olsen) writes:
>
> >Thomas Karlsen (sle...@algonet.se) wrote:
>
> >: chip
>
> good floaing point performance....
Why? The current A1200 does not have MMU or FPU either. The OS is not laid
out to fully use memory protection or virtual memory either. FPU is not
used by many programs, except raytracers and the like. People with those
requirements have a higher spec Amiga or according accelerator board
anyway. It is a good (marketing) choice until the PowerAmiga arrives.
> And I bet you'd just LOVE to have a PPC with some sort of horrendous glue
> logic to make it look like an 030 to the rest of the world.
An 040 actually, with Phase 5's 68k emulator software. It will make the
A1200 faster by some factors anyway, and that's what matters.
> Which means
> 32 bit data paths. And unless you include it on the add on module, no
> cache. GREAT performance is to be expected from a RISC CPU with horrible
> memory badnwidth....
What would you rather have for a stop gap 'til 97: An A1200 that's 4-8
times faster or an A1200 that comes a year later with redesigned chipset,
OS and lost compatibility that's 16 times faster?
Nathanael J Henderson
: Do you honestly believe that AmiT ordering 100,000 030s is going to pay the
: same as MEmory World ordering 100 or less? I don't have the figures but
: volume discounts are common on computer parts especially. Prices are also
: less to resellers than to consumers.
Yes. Very large purchasers of parts pay much less then smaller ones
per part, sometimes as little as half as much.
Shaun C. Murray
>Also, keep in mind that a 486DX2-66 is only being accessed from the Mother
>Board at 33MHz... Still within the 70ns range. Same with 486DX4-100, but
>its only 25MHz in this case.
Sorry, this is another Intel marketting ploy. The DX4 chips are clock tripled.
The 100Mhz uses a clock tripled 33.2 Mhz giving 99.6Mhz. I presume Intel
have some clever reason why they call it the DX"4".
--
Shaun C. Murray | s...@mfltd.co.uk (work)
Micro Focus Ltd, UK. http://www.mfltd.co.uk | s...@doobie.demon.co.uk (home)
Michael van Elst
> What AT should look into is a standard slot that the 1200/4000 can both use.
> It may increase the size of the trap door in the 1200 but it would cut the
> price of upgrade cards as developers would only have to develop one standard.
I doubt that this will help. A3000s and A4000s have the same CPU slot,
but inventive people made A4000-only CPU cards.
Besides, the form factor requirements for A4000 and A1200 are different.
> One last thing I would do is put a option in the 1200 to hook up an external
> mini-tower design with a few zorro/pci slots. Similar to the optional port the
> 1200's have on the back now (gvp used it for an external scsi connector).
Hmm. That's what you could do now, but it wouldn't be cheap.
Michael van Elst
>Why do they ship so many High end PC compatibles with 70ns then? Just
>curious.
To make more profit. Many people know that they have to buy a certain
amount of RAM and that they may need "cache". They rarely ask for the
memory size, speed, architecture, etc..
Regards,
Ben Hutchings
On Sun, 14 Jan 1996, Wells Fargo Bank wrote:
[...]
> Also, keep in mind that a 486DX2-66 is only being accessed from the Mother
> Board at 33MHz... Still within the 70ns range. Same with 486DX4-100, but
> its only 25MHz in this case. A 486DX4-120 is 30MHz and so on. The only
Nooo! The DX4 is only clock-tripled. The 4 supposedly stands for "4th
generation". So the DX4-120 has a 40 MHz motherboard. Don't about the
local bus though; I thought PCI only goes up to 33 MHz.
> possible exception I am aware of is the AMD (I believe) 486DX2-80 which
> would access at 40MHz and would seemingly be able to take advantage of
> 60ns RAM. Obviously, I am not aware of how this compares in the Pentium
> realm, but I still only see systems selling with 70ns RAM with an
> occasional 60ns, again, is it taking advantage of the faster RAM?
Considering most PC systems have a substantial L2 cache (256K) and the
CPU cache on the 486/586 is supposed to get about a 90% hit-rate on it's
own, I think the main memory speed won't make too much difference.
> Basically, it appears that the Amiga has the ability to access RAM faster
> IF a faster CPU is being used than the typical PC does.
Yes, but it has to do it more often thanks to the lack of L2 cache.
Ben.
--
Ben Hutchings, student. Finger me on worc...@sable.ox.ac.uk for more rubbish.
email: benjamin....@worc.ox.ac.uk WWW: http://sable.ox.ac.uk/~worc0223
Speak out against OUCS .sig censorship! See my web pages for details
Mat Bettinson
PF> If the A1200+ is a 40mhz 030 without mmu, the potential upgrader will
PF> probably be better off expanding an old A1200 by adding a 50 mhz 030 or
PF> a 25 or 33mhz 040 (or 060) that are now coming on the market.
Come on mate, get a grip.
The base model machine will have to sell at the same price as the current
model (definately not MUCH more). That means that whatever costs have been
saved by the production ramping up etc and a more efficient MB design, will
go towards the better hardware
A 1200 with a big time CPU will cost a fortune. Not only that but it's not
instantly apparent what improvement there is over a 40Mhz EC30 or the like.
Far far better to divert any extra revenue that may be available to more
memory and cheaper HD options plus the usual nicities that we'd like to see
including a built-in clock etc.
PF> Some A1200 owners have had 50 mhz 030s since 1993. Something as good or
PF> better is expected in a "+" model. It is more than 3 years since the
PF> release of the A1200.
Yes and they paid as much as the machine was worth all over again. Now with
the 68040 and 68060 accelerators, you can upgrade your CPU to these levels
with only 1.5 to 2 times the cost of the original machine.
Keen Amiga enthusiasts that pour money into their machine are not indicative
of the mass market that AT needs to capture to survive.
I wish it wasn't so but reality comes knocking on the door.
.--, /\ _ _
_ ___ |CU| / \ |\/| | | _ |_|
/\/\(-) | @ `--' /----\ | | | |_| | | Staff technical writer
/ \
< Opinions expressed by me are not necessarily those of my employer >
Bernd Bernie Meyer
>Sorry, this is another Intel marketting ploy. The DX4 chips are clock tripled.
>The 100Mhz uses a clock tripled 33.2 Mhz giving 99.6Mhz. I presume Intel
>have some clever reason why they call it the DX"4".
They argue it is giving performance equivalent to a clock-quadrupled 486,
because the Intel DX4 has a 16k cache instead of the DX and DX2's 8k.
Of course, AMD (with their 8k cache DX4s) will have a tough time explaining...
Alan L.M. Buxey
: Why do they ship so many High end PC compatibles with 70ns then? Just
: curious.
simple...same answer as why they sell people PCs in the first place with
low UARTs etc - ripping people off...
alan
--
-T H E A L A N T I M E S- FREE Distribution: >1,000,000 copies worldwide!
Date: Wed Jan 17 12:23:50 GMT 1996 Issue:01/00/00 NOT to be carried on M.S.N.!
** Confidential! ** {begin 644 address.txt} [1mWork it out... [0m
{D:'1T<#HO+W=W=RYS=7-S97@N86,N=6LO57-E<G,O:V-C:3$*}{ end}Replies Appreciated :)
IO9...@maine.maine.edu
thread. Us consumers pay a lot of money (700 bucks?) for
a decent chip one year, only to see it have a 3rd of that
value the next year. Why can't we have a OS that is able
to use banks of old CPUs in parallel? I'm sorry, but this
cannot be that hard to do. I think the lack of multi-CPU
OS'es and multi CPU system boards is directly related to
some evil scheme by Intel...
It HAS to be true. :)
-StrangeBrew
Wells Fargo Bank
: In article <wfblanDL...@netcom.com>, wfb...@netcom.com says...
: >Also, keep in mind that a 486DX2-66 is only being accessed from the Mother
: >Board at 33MHz... Still within the 70ns range. Same with 486DX4-100, but
: >its only 25MHz in this case.
: Sorry, this is another Intel marketting ploy. The DX4 chips are clock tripled.
: The 100Mhz uses a clock tripled 33.2 Mhz giving 99.6Mhz. I presume Intel
: have some clever reason why they call it the DX"4".
My mistake... This is correct. The Mother Board is accessing at 33Mhz in this
case too then.
On a different reply, someone seemed to indicate that at 33MHz, 30ns RAM is
really 'needed'. In strict theory, without a 'real world' situation, this
is of course true. However, when you have to factor in the workings of your
computer, layers, intuition, BIOS, and whatever else interacts and sets
between your CPU and RAM, this goes down to the 70ns/60ns arena.
Rob Bamford
(wfb...@netcom.com)
Wells Fargo Bank
: I just had a absolutely STUNNING jump of logic, reading this
HERE HERE!!
I was uplifted by the rumors a while back of two different companies (one
unfortunetly famous for vapour-ware) developing Multi-CPU boards for Amigas.
This IS the way for the future. And if for no other reason, to give our
old CPUs a home as a Co-Processor of sorts.
I love my CSA Derringer accelerator board because it goes half-way to this
by at least making my older 68000 CPU available for compatbility reasons.
True, it can't be accesses as a Co-processor or Parallell Processor while
the 68030 is being used, but I have the use of my old CPU still. The next
step would be to make the CPU useable WHILE the 68030 is being used, in
whatever manner it can be used.
As you say, "I'm sorry, but this cannot be that hard to do."!!
Other than Dave Haynie (I believe it was) saying this was next to impossible,
I have heard no other argument against its viability. And for some reason I
feel that Dave (it it was even him) may have misunderstood, or something.
I think he was saying that a TOTAL re-work of the Amiga-OS, etc would have to
be done to have a Multi-CPU Mother Board, etc, which is years away, if at all.
But I, and the previous person, are refering to add-on cards that could have
the necessary logic onboard to work with the OS as it currently stands. For
the 2nd CPU to be transparent to the Amigas OS. It can be done.
How about it developers. Who will be the first?
Rob Bamford
(wfb...@netcom.com)
Mark Melton
>IO9...@MAINE.MAINE.EDU wrote:
>I was uplifted by the rumors a while back of two different companies (one
>unfortunetly famous for vapour-ware) developing Multi-CPU boards for Amigas.
>This IS the way for the future. And if for no other reason, to give our
>old CPUs a home as a Co-Processor of sorts.
>I love my CSA Derringer accelerator board because it goes half-way to this
>by at least making my older 68000 CPU available for compatbility reasons.
>True, it can't be accesses as a Co-processor or Parallell Processor while
>the 68030 is being used, but I have the use of my old CPU still. The next
>step would be to make the CPU useable WHILE the 68030 is being used, in
>whatever manner it can be used.
>As you say, "I'm sorry, but this cannot be that hard to do."!!
>Other than Dave Haynie (I believe it was) saying this was next to impossible,
>I have heard no other argument against its viability. And for some reason I
>feel that Dave (it it was even him) may have misunderstood, or something.
>I think he was saying that a TOTAL re-work of the Amiga-OS, etc would have to
>be done to have a Multi-CPU Mother Board, etc, which is years away, if at all.
>But I, and the previous person, are refering to add-on cards that could have
>the necessary logic onboard to work with the OS as it currently stands. For
>the 2nd CPU to be transparent to the Amigas OS. It can be done.
>How about it developers. Who will be the first?
> Rob Bamford
> (wfb...@netcom.com)
Hmm I`m afraid that I agree with Dave on this one. I`m just finishing
a course on advanced computer architecture, the exam is on Monday!!!
Multi-processor systems are indeed the future, but unfortunatly not
that easy to make.
Picture this, you have your normal Amiga, e.g. one processor, fast
mem, graphics chips etc. You then go and strap in another processor.
Firstly how do you connect it? In the same way as the other one? Ok
then, so they both are now connected to the same memory and address
busses. Hmm they are now going to have to share the busses so they
can each have half the bandwidth, Ok not so bad just give each of
them large second level caches.
Right, now what do we have, hmm a problem, when one procesor writes to
a cache, the value in memory remains the same, now if we have one
processor it can deal with this, with two things become rather more
complex. Cos what happens when the other reads this value, all it
finds is the out of date copy. I could go into the details of this
but unfortunatly it would take rather a long time and a lot of
bandwidth.
So ok next problem, software. What do we do, run one program on each
processor??? Ok then here we go, how do we decide which process goes
on which processor, hmmm we need some pretty fancy scheduling system.
In other words a complete re-write of the exec.
Multi processing might arrive on the amiga but dont hold your breath.
It is a long way, read 2-3 years, probably more, remember Be-Box has
been in development for 4/5 years now!!
It might sound like a good Idea, but so does running cars on water.
Chris "Big-Kahuna" Rampson
>I just had a absolutely STUNNING jump of logic, reading this
>thread. Us consumers pay a lot of money (700 bucks?) for
>a decent chip one year, only to see it have a 3rd of that
>value the next year. Why can't we have a OS that is able
>cannot be that hard to do. I think the lack of multi-CPU
>OS'es and multi CPU system boards is directly related to
>some evil scheme by Intel...
>It HAS to be true. :)
>-StrangeBrew
This is a great idea from the hardware side of things, the problem is with the
software (OS) that has to deal with multiple CPUs. It is probably the MOST
difficult software "problem" that one could ever find - writing an OS to use
multiple CPUs. I have seen this work well with the latest SUN Solaris OS (8
CPUs on a SPARC 1000). SUN went through some growing pains to bring Solaris up
to speed though.
First off, you MUST have a multi-threaded OS to make this work. This is what
SUN did when they created Solaris. Of course there are many more things to do
to make this work, but the multi-thread is a must. Can the next AmigaDOS be
written in this way? Unknown. It would be cool, but it would also be a bitch to
write!
--
Owner and wielder of Occam's razor! (Committed to regula philosophandi)
_/_/_/ _/ _/ Chris "Big Kahuna" Rampson M.S.
_/ _/_/_/ _/ _/_/ _/_/_/ cram...@ford.com
_/ _/ _/ _/_/ _/ _/_/ NO cutdowns, flames or insults
_/ _/ _/ _/ _/ _/_/ NO such thing as a stupid question!
_/_/_/ _/ _/ _/ _/ _/_/_/ "NOT an OFFICIAL FORD spokesperson"
Nathanael J Henderson
: I just had a absolutely STUNNING jump of logic, reading this
: thread. Us consumers pay a lot of money (700 bucks?) for
: a decent chip one year, only to see it have a 3rd of that
: value the next year. Why can't we have a OS that is able
: to use banks of old CPUs in parallel? I'm sorry, but this
Problem A: Getting software to run on it. Rewrites would be almost
inevitablty needed.
Problem B: Cost of the additional circuitry needed to use multiple
CPUs. Would it be justified by savings on processor chips?
Problem C: Diminishing returns. 2 chips may be 1.8 times as fast
as one. 3 chips may be 2.5 times as fast as 1. 4 chips may be 3.1 times
as fast as one. For each processor added, the efficiency tends to drop.
Practically, the cost and difficulty of supporting multi-cpus isn't
worth the trouble for a consumer-level machine.
Joern Plewka
(09.01.1996)> : lol...@hsr.no (Lasse Olsen) writes:
>
> : > To clearify (hmm joyful reading ;)) the issue I droped by
> : > Motorolas price-book today. (http://motserv.indirect.com/cgi-bin/pg)
> : > Interresting to note is the 33Mhz version of the 020.
>
> : Most interesting list :)
>
> : The main point is that a 40MHz 68EC030 is cheaper by a factor of 3,
> : mainly because the 40MHz 68030 just exists in PGA package.
>
> Not by a factor three though. The plastic 030/40 is $48
> in Qts of 36 - the EC040/20 is only $76 in Qts of 24.
>
> : A 33MHz
> : 68030 (in PLCC package) is just only slightly more expensive. So you
> : could argue wether you want a 40MHz CPU without MMU or a 33MHz one
> : with MMU. But it is a bit difficult to argue why a slower machine
> : has to cost more.
>
> Yep. :)
>
> : Another interesting fact is that a 25MHz 040 is only slightly more
> : expensive than a 50MHz 030 and a 33MHz LC040 is even cheaper.
> : This basically rules out the 50MHz 030 for everything and the choice
> : is between 33MHz 68030, 40MHz 68EC030 (about same cost) or a more
> : expensive 040 version.
>
> This probably has something to do with ATs choice of the 40Mhz.
A very "inteligent" idea;-(( no memoryprotection with OS4.x and much more
important: NO VMEM ... only expensive ram...
And a slow base for later CHRP-Software. That`s much more expensive than
takeing a computer-type cpu for a "telephon"-type;-((.
Yes, VMM is running on A3000 with 4MB and it`s possible to run ADPRO and
other nice programms...an A4000ec with 8MB isn`t -(((((((.
What about 68040PGA? There should be 3.3V parts I heard? It would me
a better choice to take it, IMHO. It`s available in LC, EC and FULL...
Prices are dropping and it won`t be a problem to use the
the MC68040RC-40MHz version on demand or in later versions without
a new design.
They are doing a new 060-board for A4000T...why not use the same
chip for an 040 in A1400?!
I don`t think a customer would buy an ec030 if he could get an ec040
for additional $60 with more than twice that speed.
It`s one thing I don`t understand...it`s a pure advertizing problem...
Pentium is getting minimum now...compared to it the MC68ec030-40 is
obsolete. Wake Up, there are NO Apple machines left with 030....
and 040 are getting off.
> DSP56300CLASB DPZZ TCCX N 1 300.020 - DSP-MAC (80MHz, soon 100Mhz)
What`s its speed compared to C80-66 or the new IBM?
An A1400 would be a little overpowered with ..300 but it would be
useful for A4000T class. There doesn`t have to be a new mainboard
because they are doing a new CPU module...
What about a f*cking ET4000w32 for $30...or S3-868 (not 64V*) for viewing
MPEG with A4000T? A4000T for Multimedia? Yesterday...
I don`t thing this stuff is a dream (except 56300)...it`s needed till
the PPCs are comming mid `97!
/ no / J_Pl...@People-S.people.de / AMIGA /
/ MS+ / J_Pl...@Amtrash.comlink.de / is /
/ gAIDS / Plew...@Informatik.FH-Hamburg.de / back /
/ +-----------------------------------+ /
/ http://users.informatik.fh-hamburg.de/~plewka_j /
Joern Plewka
(12.01.1996)
> There is no reason as to why both couldn't be implemented -
> except a minute cost-increase.
> : Point two - the memory subsystem of an 040 (standard, lowcost,
> : economy, or cold fire) is a LOT more complicated than that of an 030.
It`s not low cost...it`s very expensive due design...AFAIK
and the 040 is used for emulation of the 030 bus. That`s no good for
performance...as you can see.
> : Case in point - look at the 3640 (Commodore's 040 card for the
> : A3k/A4k). It has a memory bus from hell, and it was STILL expensive
> : to make. How do you expect to make a low end computer when you have
> : to deal with that? You can't.
>
> Might be a point. But, the 3640 was designed my a team
> having less-than-satisfactory experience with 040s -
> more than 4 years ago.
> This is 1996.
That`s TRUE but it was difficult to implement in the A3000 design...
implementation in a new machine design would be the real way.
> : Point three - the A1200 MB is designed to handle an 030
>
> Sidenote: So is the A4000 MB.
No, that`s not all...its designed to handle an 68020...;-(((
Joern Plewka
(11.01.1996)
> Point one - the Amiga line has some real tightwads out there, who
> would bitch if you put in a bbu clock, let alone an FPU. Lets get an
> MMU standard before we get an FPU in there...
Agreed...
Yep, but an MC68030FE33C is $63 and MC68EC030FE40C is $47...that`s the
problem;-(((
An 040 starts at $90 with ec25 up to real 040 at $380 maybe a range for
more versions and upgradable to 060 without changes.
That`s what I would take if I had to do a new CPU board for the A4000
the VLSI would be benefit for A1200, too...with Jedec72 memory.
A feature connector would be good for dockingstation but should be
compatible (adapter) with A1200. That`s the right place for a board
maybe with DSP and S3-868 for acceptable GFX and SOUND.
That`s a difference of max. $150...and really upgradeable for a time.
Ian Parkinson
Some chaps from Acorn came and demonstrated it to us - they claim that it
is able to utilise multiple processors.
Even more impressively, they can use different types of chip - although at
the moment cards are only available for the ARM processor and for 486's.
This, I admit, made me consider (only for a second) dumping my Amiga... (Note:
I've always been impressed by Acorn, so this wasn't too unnatural!)
However, when I asked them how use of the multiple CPUs is implemented from a
software level, they said that it is essentially up to the individual tasks
to secure an extra processor and use it. I got the impression that multitasking
is performed just on one chip, but the programs can 'get' an extra chip and
use it exclusively. If this fails (i.e. because there aren't enough chips)
then they would have to fall back to normal multitasking.
This, IMO, is not good. Multitasking should be implemented on each CPU, i.e.
the tasks should be shared around the available processors. The idea is that
when any one CPU's current task 'runs out of time' it takes over any task (of
high enough priority) currently waiting.
I don't pretend to know anything about the hardware side of this sort of
implementation - it must be possible because Acorn have done it - but I don't
see any enormous difficulties on the OS side - if anybody would care to
contradict me I'd be happy to hear it!
For the record, the chappies from Acorn half-hinted to me that they were
planning this sort of system for RiscOS, but I got the impression that it
may have been sales pitch, and they didn't really know what they were talking
about...
If AT can produce something along these lines the world would rejoice!
Ian
Bernd Bernie Meyer
>On a different reply, someone seemed to indicate that at 33MHz, 30ns RAM is
>really 'needed'. In strict theory, without a 'real world' situation, this
>is of course true. However, when you have to factor in the workings of your
>computer, layers, intuition, BIOS, and whatever else interacts and sets
>between your CPU and RAM, this goes down to the 70ns/60ns arena.
Even in a "real world" situation, this is absolutely 100% true. Why do you
think a 486/33 takes 4 clock cycles per burst for the first access to main
memory? Certainly not because of "intuition, BIOS and whatever else". No,
simply because the RAM doesn't give back the values any earlier.
A further point for this argument is that the cache _is_ in the 15-25ns
area (which is 30ns minus "workings of the computer, layers" [which make
the time that is left for the memory shorter, not longer]). And cache
typically gets accessed in a 1-1-1-1 burst.
On a Pentium with 66MHz external clock, the first access to the cache is
taking two clock cycles. Subsequent burst accesses are anticipated and served
in a single clock cycle. This, of course, only if you have pipelined burst
cache....
Mark Melton
Well having spent a number of hours revising this very suject, I`ll be
happy to contradict you.
I have a friend who got a RISC PC a few weeks ago. What would appear
to be happening is this. The RISC PC software run`s on the ARM and
the PC software run`s on the 486. This is technically speaking a
multi processor machine, but if we are prepared to call this multi
processing then the amiga already is a multi processor machine. The
copper in the amiga is a processor, it has its own instruction set and
executes a list of instructions. It does this independantly from the
680x0 processor. Also if a bridgeboard is fitted to an amiga this is
also a multi processor system. The RISC PC is nothing new in this
sense, still you know what marketing people are like!!!
("We`ve just had this great Idea, its called multi media!" Microsoft)
True multiprocessing, in the modern sense or what is correctly called
a "centralised shared-memory architecture", consists of a number of
identical processors, using a shared memory space. The real problem
is what does the programmer see when they write software for the
machine. Do they see individual processor which run totally separate
tasks? or does the user see one virtual machine? The latter is
obviously the best model, as further processors can be added without
the need to re write programs to take advantage of multiple
processors.
Anyway I could go on all day but I must get some revision done!!
Michael van Elst
>A very "inteligent" idea;-(( no memoryprotection with OS4.x and much more
>important: NO VMEM ... only expensive ram...
You assume that OS4 will run on old Amigas.
And you also assume that VMEM can replace real RAM.
>And a slow base for later CHRP-Software.
It won't be able to run CHRP-Software anyway... this is no PPC, right ?
Do you want a PPC emulator ?
>What about 68040PGA? There should be 3.3V parts I heard? It would me
>a better choice to take it, IMHO. It`s available in LC, EC and FULL...
The 3.3V parts are just EC.
PGA parts are a bad choice for low price machines.
In any case you should have read the Motorola price list to get an
idea what extra cost a 040 would involve.
>Prices are dropping
Have a look at current prices.
>and it won`t be a problem to use the
>the MC68040RC-40MHz version on demand or in later versions without
>a new design.
Obviously there would be.
>I don`t think a customer would buy an ec030 if he could get an ec040
>for additional $60 with more than twice that speed.
If that $60 had anything to do with reality... maybe. But didn't you
complain about the lack of an MMU ? Wouldn't people rather buy another
2MB RAM instead for those $60 ?
>Pentium is getting minimum now...compared to it the MC68ec030-40 is
>obsolete. Wake Up, there are NO Apple machines left with 030....
>and 040 are getting off.
Admit it :) You are DOOMED.
Bernd Bernie Meyer
>The next
>step would be to make the CPU useable WHILE the 68030 is being used, in
>whatever manner it can be used.
>As you say, "I'm sorry, but this cannot be that hard to do."!!
If you support that quote, you have obviously never tried to construct a
multiprocessor machine :-)
Even hooking up 4 Z80s to a common memory needs a lot of glue logic and
careful planning (and a fast memory, plus local memory for each processor).
>Other than Dave Haynie (I believe it was) saying this was next to impossible,
I think that is quite enough of an argument against it. I may not always agree
with Dave on his views about Software, but he definitely is an engineer
who knows what he is talking about!
>How about it developers. Who will be the first?
Not me.... Maybe once linux/68k incorporates the SMP functionality. But
that still woudln't solve the hardware nightmare....
Stefano Agostinelli
> An 040 starts at $90 with ec25 up to real 040 at $380 maybe a range for
> more versions and upgradable to 060 without changes.
What is the price for the 040LC ?
Cheerio
Stefano
+-----------------------------+-----------------------------------------+
| Stefano Agostinelli | Now developing for You ARM: |
| Genoa - Italy | the State of Art AmigaRoleMaster System |
+-----------------------------+-----------------------------------------+
| IRC: arm | 'Where are you going, Master?' |
| E-mail: ago...@pn.itnet.it | 'To the Havens, Sam' |
+-----------------------------+-----------------------------------------+
Ben Hutchings
[...]
> Yep, but an MC68030FE33C is $63 and MC68EC030FE40C is $47...that`s the
> problem;-(((
How can Motorola possibly get away with charging this much, is what I
want to know? You can get an i486DX-33 for about this much (about $50 as
I recall), and that would be about 50% faster and with an FPU included.
Who exactly is buying them? Laws of supply and demand surely dictate that
they should be real cheap now that Apple has stopped buying... but then,
Intel has cloners and Motorola doesn't. <sigh>
> An 040 starts at $90 with ec25 up to real 040 at $380 maybe a range for
> more versions and upgradable to 060 without changes.
$380 is an absurd price. You can get a Pentium-90 for about that, maybe
less, these days. That's maybe 3 times as fast.
> That`s what I would take if I had to do a new CPU board for the A4000
> the VLSI would be benefit for A1200, too...with Jedec72 memory.
> A feature connector would be good for dockingstation but should be
> compatible (adapter) with A1200. That`s the right place for a board
> maybe with DSP and S3-868 for acceptable GFX and SOUND.
>
> That`s a difference of max. $150...and really upgradeable for a time.
Someone, please clone Motorola CPUs!!!
--
Ben Hutchings, student. Finger me on worc...@sable.ox.ac.uk for various info.
email: benjamin....@worc.ox.ac.uk WWW: http://sable.ox.ac.uk/~worc0223/
Users of the world unite! You have nothing to lose but your Micro$oft software
James McArthur
: On 9 Jan 1996, Michael van Elst wrote:
: > That's what I think too. The EC030/40 is the best _cheap_ CPU
: > for an A1200 class machine.
: >
: > > I sure wouldn't mind spending a little bit more on a A1200/040.
: > > Of course, I'm not sure of exactly how much extra glue logic
: > > it would take to acchive this. And you'll loose the ability
: > > to upgrade with a FPU. Might even turn out to be too hot.
: The best choice would be to have the A1200+ equipped with an 40Mhz EC030
: chip and a Processor Direct Slot like in the A3000/A4000 models. Then you
: can install a 040 or a PPC if you like, when you like. The PDS should be
: located near the CPU and the old trapdoor or similar slot should still be
: there aswell for other cards.
Remember those press releases AT had about the PowerPC? Didnt they say
that they were going to have a PPC upgrade for the March A1200?
So maybe they are going to have a PDS thingie..
: /Thomas
John Bayko
Ian Parkinson <sjoh...@sable.ox.ac.uk> wrote:
>With regard to having lots of processors, has anybody seen Acorn's RiscPC?
>Some chaps from Acorn came and demonstrated it to us - they claim that it
>is able to utilise multiple processors.
>
>Even more impressively, they can use different types of chip - although at
>the moment cards are only available for the ARM processor and for 486's.
>
>This, I admit, made me consider (only for a second) dumping my Amiga... (Note:
>I've always been impressed by Acorn, so this wasn't too unnatural!)
>
>However, when I asked them how use of the multiple CPUs is implemented from a
>software level, they said that it is essentially up to the individual tasks
>to secure an extra processor and use it. I got the impression that multitasking
>is performed just on one chip, but the programs can 'get' an extra chip and
>use it exclusively. If this fails (i.e. because there aren't enough chips)
>then they would have to fall back to normal multitasking.
Acorn's RiscOS doesn't (or didn't last I heard) do 'normal'
multitasking to begin with - it's co-operative, like MS-Windows and
the Macintosh, not pre-emptive, like the Amiga, so a program normally
'gets' the main CPU, then releases it when it feels like it (for a
system call or event poll).
That's the reason I decided to go with an Amiga rather than an
Acorn (I like some other aspects of the Acorn OS better as well (RTG
for example), but not enough to choose it). In particular, co-operative
multitasking means slow serial communications on the Archimedes
series, just like under MS-Windows.
--
John Bayko (Tau).
ba...@cs.uregina.ca
http://www.cs.uregina.ca/~bayko
Joern Plewka
(21.01.1996)
> >A very "inteligent" idea;-(( no memoryprotection with OS4.x and much more
> >important: NO VMEM ... only expensive ram...
>
> You assume that OS4 will run on old Amigas.
> And you also assume that VMEM can replace real RAM.
Yep, my second A3000 is running with 4MB and 40MB VM.
OK, the zips will be replaced to a 8MB-Jedec72 next week...
it`s wasn`t fast but it worked well for me and my father (its my PC;-().
Second problem is Cybergfx. It`s useless if there`s no VMM...the heap is
trashed extremly fast...
> >And a slow base for later CHRP-Software.
>
> It won't be able to run CHRP-Software anyway... this is no PPC, right ?
> Do you want a PPC emulator ?
That`s not what I wanted to say...on one side there will be 603e/604 and
on the other side ec030. Remember when A3000 got on market...
> >What about 68040PGA? There should be 3.3V parts I heard? It would me
> >a better choice to take it, IMHO. It`s available in LC, EC and FULL...
>
> The 3.3V parts are just EC.
> PGA parts are a bad choice for low price machines.
What`s the difference of $67 and $100?...and what about later, when 060
is really shipping...
> In any case you should have read the Motorola price list to get an
> idea what extra cost a 040 would involve.
I wrote this text because I had a look into it...and got surprised
(ec020 for <$20 now)
> >and it won`t be a problem to use the
> >the MC68040RC-40MHz version on demand or in later versions without
> >a new design.
>
> Obviously there would be.
Synchronus to asynchronus?
>
> >I don`t think a customer would buy an ec030 if he could get an ec040
> >for additional $60 with more than twice that speed.
>
> If that $60 had anything to do with reality... maybe. But didn't you
> complain about the lack of an MMU ? Wouldn't people rather buy another
> 2MB RAM instead for those $60 ?
Yep, sure...but if they need the mmu, they can get a real 060 or 040,
or LC-Types if they like...nothing than changing the CPU...and adding ram.
If I have a look at Cyberstorm...more than $400 (3Q95) for 68060...
> >Pentium is getting minimum now...compared to it the MC68ec030-40 is
> >obsolete. Wake Up, there are NO Apple machines left with 030....
> >and 040 are getting off.
>
> Admit it :) You are DOOMED.
;-) never really played one of these games...exept testing Marathon
on ShapeS....no, really not! I`m playing 4-player-Lotus and Pinball,
if there is time...don`t need PPC for Games...
Pauli Porkka
>: >Michael van Elst (mle...@serpens.rhein.de) wrote:
>: >: This is the evaluation kit. DSP + 32k words RAM + 16bit stereo codec.
>: >: Nice thingie and I might have some Amiga software for it in the future.
>: > By all means, please do.
>: > I heard talk of some DSP boards comming for the Amiga some
>: > time ago - are they still vapour-ware?
>: AFAIK you could buy the Delfina board for some time.
> Ah, that's right.
> The Finish one with the 56002 at approx $800 (?)
> Cheers...
More info available at www.tit.fi/~icpopa/delfina.html
Proud owner of THE FIRST SOLD Delfina DSP Audio card
--
Pauli Porkka/Digital Audio Systems Designs
E-Mail: ppo...@walli.uwasa.fi
HomePage: http://www.tit.fi/~icpopa/Welcome.html
Amiga CDROM HomePage: http://www.tit.fi/~icpopa/acdrom.html