PDA

View Full Version : More 90nm PowerPC Notes


MacRumors
Jan 8, 2004, 04:01 AM
A few more notes and confirmation of the 90nm PowerPC 970s...

This PowerPC G5 Whitepaper (PDF) (http://www.apple.com/server/pdfs/L301298A_PowerPCG5_WP.pdf) from Apple has been updated to provide information of the New 90nm PowerPC.

According to the paper (page 15), the new 90nm PowerPC 970 contains 58 million transitors, uses a 90-nanometer, silicon-on-insulator (SOI) process [not SSOI (http://www.macrumors.com/pages/2003/11/20031121150200.shtml)] and has a die size of 66 square millimeters. The 130-nm 970 PowerPC currently used in the PowerMac G5s have a die size of 121 square millimeters (http://www.arstechnica.com/cpu/02q2/ppc970/ppc970-1.html).

More information of the 90nm PowerPCs should be presented by IBM in February (http://www.macrumors.com/pages/2003/11/20031113110834.shtml). The new PowerPCs would logically be used in upcoming PowerMac revisions.

krassy
Jan 8, 2004, 04:12 AM
Originally posted by Macrumors
More information of the 90nm PowerPCs should be presented by IBM in February (http://www.macrumors.com/pages/2003/11/20031113110834.shtml). The new PowerPCs would logically be used in upcoming PowerMac revisions.

more information about the 90nm process will be presented (http://www-306.ibm.com/chips/events/semico04/) by ibm today in japan.

Mac-Xpert
Jan 8, 2004, 04:19 AM
Well, lets hope we will see them in the Powermac soon. Reading Rob Galbraiths comparison of Mac-Pc performance on raw digital image processing
(http://www.robgalbraith.com/bins/multi_page.asp?cid=7-6451-6410) shows we can still use some extra power.

I'm hoping for 2.6 ghz duals in February :)

Henriok
Jan 8, 2004, 04:39 AM
I'm interessted in what the 90 nm 970 wil be called. 970FX seems to be the most likely candidate but an official word from IBM would be nice.

krassy
Jan 8, 2004, 04:48 AM
Originally posted by Henriok


hi henriok (from appleinsider forums?) ... funny, how small the world is...

krassy
Jan 8, 2004, 05:24 AM
Originally posted by Mac-Xpert
Well, lets hope we will see them in the Powermac soon. Reading Rob Galbraiths comparison of Mac-Pc performance on raw digital image processing
(http://www.robgalbraith.com/bins/multi_page.asp?cid=7-6451-6410) shows we can still use some extra power.

I'm hoping for 2.6 ghz duals in February :)
these benchmarks are not worth it. try to use a dual G5 beside a dual xeon and run a few more apps simultanously and you'll know what i mean... the dual G5 is superior ...

xid
Jan 8, 2004, 05:55 AM
Originally posted by Mac-Xpert
Well, lets hope we will see them in the Powermac soon. Reading Rob Galbraiths comparison of Mac-Pc performance on raw digital image processing
(http://www.robgalbraith.com/bins/multi_page.asp?cid=7-6451-6410) shows we can still use some extra power.

I'm hoping for 2.6 ghz duals in February :)

Of course faster Macs would help, but a far bigger speedup would result from optimizing the applications being used. I've had some code which ran about as fast on a 2GHz G5 than it did on a 1 GHz Athlon. I didn't even bother comparing to a 3GHz Intel. After optimizing, the G5 was about 6 times faster than before. The code wasn't bad, but in this case I had a very tight loop where I multiplied lots of ints with lots of floats and the G5 pays a *huge* penalty for the conversion; Intel/AMD and even the G4 don't. I took that out, inserted some AltiVec on the way... I had to put in some code to check the results because after the initial results I couldn't believe the speed I got and had to make sure I actually did some calculating!

I don't know what the Benchmarks do, but it seems to rely on two things: bandwidth and serial processing of data streams. Basically this is what a G5 is designed for. If a G5 is not a lot faster for this than any Intel processor, it is a problem with the code and not with the G5.

A 2.6 GHz G5 would be at most 30% faster and would still lose at a lot of these benchmarks. Better code and the 2GHz would probably be dominating already.

Cheers

Henriok
Jan 8, 2004, 06:26 AM
Originally posted by krassy
hi henriok (from appleinsider forums?) ... funny, how small the world is... Semi-quote from Halo/Marathon: We're everywhare!! :)
The Mac world isn't that large after all. Thank god!

x86isslow
Jan 8, 2004, 06:39 AM
before the thread is hijacked by those clamoring for a g5 powerbook,
a few questions:
does the white paper say anything about heat dissipation?
what does the change in die-size mean?
are the new xserves shipping now?

gaomay
Jan 8, 2004, 06:40 AM
AFAIK a large part of the Opteron's advatage over the G5 comes from its 1024Kb L2 cache. Does the new 90nm 970 have a 1024Kb L2 cache or have they stuck with the 512?

Henriok
Jan 8, 2004, 07:06 AM
Since the 90 nm 970 is made up by the same ammount of transistors (58 Million) we can rule out any sigifiant redesign of the chip. Larger cach, integrated memory controller, andvanced power savings and such is out of the question.

I'd say that the smaller size will be responsible for the chip being cheaper. The processor size is about half of the former, so twice the number of processors can be produced on one wafer, thus reducing cost per processor dramaticly.

The new Xserves will ship in February.

The white paper only says that the die shrink results in less heat dissipation but not by how much.

gaomay
Jan 8, 2004, 07:24 AM
Thanks for the clarification - I should have realised that myself!
So we will have to wait for the mythical 980 for a larger L2 cache?

When is the 980 coming - I'm planning on buying a tower this time next year, after the start of yer refresh - will they be 980s by then?

Mac-Xpert
Jan 8, 2004, 07:42 AM
Originally posted by xid
Of course faster Macs would help, but a far bigger speedup would result from optimizing the applications being used. I've had some code which ran about as fast on a 2GHz G5 than it did on a 1 GHz Athlon. I didn't even bother comparing to a 3GHz Intel. After optimizing, the G5 was about 6 times faster than before.

I totally agree. The numbers on Rob Galbraiths site do also show that the 2 ghz G5 was faster in MacBibble than the 3.06 Xeon machine in the windows version. Unfortunately most software vendors don't seem to care much about the speed of their apps and won't put much effort in optimizing them for the g5 processor. I hope this might change one day, but I don't think it will be anytime soon.

Henriok
Jan 8, 2004, 08:41 AM
Originally posted by gaomay
Thanks for the clarification - I should have realised that myself!
So we will have to wait for the mythical 980 for a larger L2 cache?
It seems like it. I'm not to surprised. 970 was the first step in the new 900-family and i think it might be somewhat of a working design study; the first step in a long series of processors. Much will be learnt from 970 design, things that will go into future designs.. 980 (or whatever it will be called, my best guess is still 975) will be the first.
Originally posted by gaomay
When is the 980 coming - I'm planning on buying a tower this time next year, after the start of yer refresh - will they be 980s by then? Steve Jobs have promised us G5 processors @ 3 GHz sometime this summer. The 970 is a very unlikey candidate for that feat, so that leaves 980 to fulfill Steves promise.

I guess that we'll se 980 delivered this fall. If you buy a new tower next spring, you might be into the very nice prospect of buying a dual 3.5 GHz machine :)

We don't have ANY confirmed information about 980, not even its name. We don't even know it exists at all, that's just something that's very likely. If someone claims to have solid info about 980, you should take it with a handful of salt. At least until IBM and/or Apple says something.

Mr. Anderson
Jan 8, 2004, 09:01 AM
I'll be really interested in seeing what IBM presents, the Apple white paper is more marketing oriented.....

But it looks good that they've got the 90 nm process going for the XServe. I wonder what the odds are that the top end will be 2.6 GHz - any chances that might be higher?

D

gaomay
Jan 8, 2004, 09:06 AM
Dual 3.5Ghz would be sweeeeet.

Imagine UT2004 running on that.

Am I right in saying that a doubling of L2 cache would provide a significant performance boost? It all comes from when the P4 went from 256 to 512K cache which seemed to boost performance by >10% clock for clock. So maybe a 15% boost going from 512K tro 1024K?

This may be a large part of Prescott's increased performance also. i believe it has 1024k L2 cache.

Mac-Xpert
Jan 8, 2004, 09:21 AM
Originally posted by gaomay
Am I right in saying that a doubling of L2 cache would provide a significant performance boost? It all comes from when the P4 went from 256 to 512K cache which seemed to boost performance by >10% clock for clock. So maybe a 15% boost going from 512K tro 1024K?

This may be a large part of Prescott's increased performance also. i believe it has 1024k L2 cache.

I have the feeling that increasing the cache on the 970 might be less effective than on the P4 because the bus speed on the G5 is much higher in comparison to the processor speed.

Photorun
Jan 8, 2004, 09:23 AM
Someone pinch me, reading the post and the white paper, is the 90nm 970 actually only HALF the size of the last chip will all the speed? I feel week in the knees.

Put it in the Powerbook? Heck, put one in an iPod! (kidding).

Henriok
Jan 8, 2004, 09:48 AM
The increase of L2 cache on previous PPCs show similar performance boosts as the Pentium. The main thing is that cache is expensive. Just look att the Pentium4 EE with 2 MB cache. It is REALLY expensive compared to its little brothere with just 1 MB. Itanium 2-processors with huge caches (6-9 MB) will cost more a piece than a complete dual-G5. I think it's hard to justify verry large caches, but processors will undoubtfully get higher performance with them. 1 MB L2 cache in 980 seems reasonable though. The 750GX got 1 MB so why doesn't 970? It's really a fair question.

The extremely high bandwidth the 970 and its successors got might be one of the reasons for the farily small cache size and the lack of L3 cache, but it got a terrible overhead so even if it transfers data fast it takes time to initiate a transfer. precious time.

Concerning the shrinkage of die size. Just do the math:
130x130 = 16900
90x90 = 8100
== half.
If everything else is the same, which seems to be the case.

sergeantmudd
Jan 8, 2004, 10:02 AM
I love how people assume that even with a new die process, the 970 can't even get bumped in speed once. The Pentium4 has gone from like 1.4 ghz to 3.6ghz, and with Prescott will get pushed even further. Do you honestly belive that the 970 has run out of room to grow, even with a new a die process?

dxp4acu
Jan 8, 2004, 10:16 AM
who assumes that??

Of course it will.

wizard
Jan 8, 2004, 10:23 AM
Since there is alot of head room in the current processor I suspect the 90nano unit will clock well over 3GHz. It appears that it is more of a question about how hot Steve wants to run the processors.

Now how well that processor will perform at the higher clock rates is an open question. The lack of a cache size increase does cause concern. One will have to waite for IBM do release all the details. Hopefully they have made some improvements to the processor, it owuld be rather sad if the only thing we got out of this rev is a process shrink and the corresponding faster clock rate.

Thanks
Dave


Originally posted by Mr. Anderson
I'll be really interested in seeing what IBM presents, the Apple white paper is more marketing oriented.....

But it looks good that they've got the 90 nm process going for the XServe. I wonder what the odds are that the top end will be 2.6 GHz - any chances that might be higher?

D

wizard
Jan 8, 2004, 10:33 AM
cost really isn't part of the equation, with the process shrink they could have kept the die size and used the extra space for the extended cache. Sure it would be a problem if you enlarged the die size but this should not happen with the process shrink.

As you point out IBM alread delivers a low cost processor with a large cache. Granted there are probally difference in implementation but still it is only a cost consideration if the die becomes to large.

The bandwidth capability of the G5 will mean nothing as the ratio between clock rate and memory pefromance widens. It will widen because even if Apple implements faster RAM it still will not keep up with the 970 as it passes 3GHz.

Now this may sound like terrible news, but there is the possibility that IBM has attacked the issue in other ways. It is not impossible to improve the performance of the cache. The release of the new documentation from IBM will be very interesting to say the least.

Originally posted by Henriok
The increase of L2 cache on previous PPCs show similar performance boosts as the Pentium. The main thing is that cache is expensive. Just look att the Pentium4 EE with 2 MB cache. It is REALLY expensive compared to its little brothere with just 1 MB. Itanium 2-processors with huge caches (6-9 MB) will cost more a piece than a complete dual-G5. I think it's hard to justify verry large caches, but processors will undoubtfully get higher performance with them. 1 MB L2 cache in 980 seems reasonable though. The 750GX got 1 MB so why doesn't 970? It's really a fair question.

The extremely high bandwidth the 970 and its successors got might be one of the reasons for the farily small cache size and the lack of L3 cache, but it got a terrible overhead so even if it transfers data fast it takes time to initiate a transfer. precious time.

Concerning the shrinkage of die size. Just do the math:
130x130 = 16900
90x90 = 8100
== half.
If everything else is the same, which seems to be the case.

mjtomlin
Jan 8, 2004, 11:48 AM
It was believed that the 130nm 970's would be able to hit 2.6, 2.8 at the highest and to get to 3GHz or more we would have to wait for 90nm 980, which is supposedly due out this fall. Well now that the 90nm 970 is out, who knows maybe we'll see new PowerMacs this month or next pushing 3GHz.

Think about it... they got a 90nm 970 crammed into a 1U rack. It may only be running at 2GHz because of heat issues. The towers have tons of space for heat sinks and fans allowing them to crank up the speed of the 970.

Code101
Jan 8, 2004, 11:52 AM
I don't know why there not using SSOI! With all the heat the PPC 970 produces, .90 alone won't fix that problem. The Pentium 4 (Prescott) that will be out next month is .90, Intel had to use SSOI in order to deal with the 100+ watt heat problem. The PPC 970 is having the same problem. I would think IBM will have to do this in order for Apple to use the G5 in a notebook.

I believe SSOI is Intel's patent. IBM may have to get a license to use it.

arn
Jan 8, 2004, 12:20 PM
Originally posted by Code101
I don't know why there not using SSOI! With all the heat the PPC 970 produces, .90 alone won't fix that problem. The Pentium 4 (Prescott) that will be out next month is .90, Intel had to use SSOI in order to deal with the 100+ watt heat problem. The PPC 970 is having the same problem. I would think IBM will have to do this in order for Apple to use the G5 in a notebook.

I believe SSOI is Intel's patent. IBM may have to get a license to use it.

Intel's not using SSOI.

See the SSOI link in the article

arn

ddtlm
Jan 8, 2004, 12:43 PM
Henriok:

Steve Jobs have promised us G5 processors @ 3 GHz sometime this summer. The 970 is a very unlikey candidate for that feat, so that leaves 980 to fulfill Steves promise.
Why would you think this? The 130nm G5 wasn't maxed out at 2.0ghz; 2.0ghz was pretty much it's launch speed. 2.4ghz seems like a very attainable speed at 130nm, and I don't think asking 25% higher for the 90nm chip is unreasonable. If you ask me, people are so set on the idea of a "980" that they are underestimating the scalability of the 970. I am not expecting any "980" till after the summer/fall tower refresh (although I must say that IBM's choice of 512k L2 seems to leave the high-end wide open for a "980").

Mac-Xpert:

I have the feeling that increasing the cache on the 970 might be less effective than on the P4 because the bus speed on the G5 is much higher in comparison to the processor speed.
Heh, the G5 has not one but two things higher about its FSB vs the P4's. One is the bandwidth, and the other is latency. (Unfortunately high latency is bad.) Further, you should not compare processor clock speed to FSB speed and draw conclusions, what you want compare is processor performance to memory speed. What matters is how much data the processor can process, not how fast it is clocked. By that measure the G5 does not have a FSB advantage over the P4. Yes, a larger L2 would help.

mjtomlin:

Think about it... they got a 90nm 970 crammed into a 1U rack. It may only be running at 2GHz because of heat issues.
Consider AMD's 130nm 2.2ghz Opteron that works fine 2 per 1U machine. One of three things is possible:
a) The 90nm G5 throws off more heat than a 130nm Opteron, or
b) The 90nm G5 is not currently available above 2.0ghz, or
c) Apple decided to make Xserves less compeditive than they could be.

tortoise
Jan 8, 2004, 12:52 PM
Originally posted by gaomay
AFAIK a large part of the Opteron's advatage over the G5 comes from its 1024Kb L2 cache. Does the new 90nm 970 have a 1024Kb L2 cache or have they stuck with the 512?


Opteron's biggest advantage is that it has the fastest memory subsystem out there bar none for a normal CPU. As in, the G5 has TWICE the latency of the Opteron. If you are running an app that doesn't fit into the L2 cache and regularly has to access the main core, the superior latency specs of the Opteron's memory subsystem will dominate the performance. This is due in significant part to the fact that Opteron has an on-chip memory controller. Combine the faster memory access and larger L2 cache, and you have a chip that is much better at keeping the execution units busy in the core.

A lesser advantage Opteron has is that the execution units in the core are more general purpose in scope than the ones in the G5, and therefore can be used more efficiently in some cases. But I personally think that this is largely unimportant compared to the substantial differences in the memory subsystems.

If the G5 had a memory subsystem that had super-low latency like the Opteron, the performance gap in the general case would be closed.

ddtlm
Jan 8, 2004, 01:06 PM
tortoise:

If the G5 had a memory subsystem that had super-low latency like the Opteron, the performance gap in the general case would be closed.
Hot dang, if the G5 core had been equipted with AMD's 1 meg L2 and on-die 128-bit memory controller then there would be no competition. Well actually that's not true... but it'd be fast, anyway.

Code101
Jan 8, 2004, 01:12 PM
Originally posted by arn
Intel's not using SSOI.

See the SSOI link in the article

arn

Maybe were both thinking of SSOI meaning another thing? I'm referring to "strained silicon." This is Intel's baby. It looks like a very promising technology across the board. IBM would do well to use this in the 970/980 chip.

Here are two articles of reference.

http://www.eetimes.com/semi/news/OEG20020813S0012

and

http://www.eetimes.com/story/OEG20020816S0048

mjtomlin
Jan 8, 2004, 01:15 PM
ddtlm:
Consider AMD's 130nm 2.2ghz Opteron that works fine 2 per 1U machine. One of three things is possible:
a) The 90nm G5 throws off more heat than a 130nm Opteron, or
b) The 90nm G5 is not currently available above 2.0ghz, or
c) Apple decided to make Xserves less compeditive than they could be.

Well it seems they each sacrificed one thing for something else...

Apple: 3 drive bays
AMD: 2

Apple: 2 full length PCI-X slots
AMD: 1 full, 1 half

Apple: 8 RAM slots
AMD: 16

...etc

Apple's system may use a smaller cooling system and they needed to wait for the smaller chip (less heat dissipation)

The 2.0GHz? Well Apple may have wanted to keep the price down, there may not be a ton of them available yet, or they didn't want to overstep the towers, which is their bread and butter so to speak.

Code101
Jan 8, 2004, 01:16 PM
FYI.

The above articles I posted speek of IBM's intrest in SS technology.

Code101
Jan 8, 2004, 01:20 PM
To date, IBM Corp. has been the main proponent of strained silicon, with a plan to add its version of strained silicon to IBM's 65-nm process node, which is expected to move to first manufacturing in 2005. IBM plans to combine its strained silicon expertise with its silicon-on-insulator capabilities. However, IBM researchers have indicated at least a 10 percent cost adder for strained silicon, and said much work remains to balance the mobility enhancement in the NMOS and PMOS portions of a CMOS device. Dan Hutcheson, president of VLSI Research Corp. (San Jose, Calif.), said "it is pretty clear that Intel has made a major breakthrough here. It is amazing that they would use strained silicon at the 90-nm node, and if the cost adder is only 2 percent then the process additions would need to be pretty trivial."

Taken from EE Times,
http://www.eetimes.com/semi/news/OEG20020813S0012

daveL
Jan 8, 2004, 01:39 PM
Originally posted by Code101
Maybe were both thinking of SSOI meaning another thing? I'm referring to "strained silicon." This is Intel's baby. It looks like a very promising technology across the board. IBM would do well to use this in the 970/980 chip.

Here are two articles of reference.

http://www.eetimes.com/semi/news/OEG20020813S0012

and

http://www.eetimes.com/story/OEG20020816S0048
Actually, there are several different implementations of SSOI; Intel is just one of them. Your own IBM link seems to bear that out. If I recall, AMD is also working on a different (vs Intel) approach to SS.

daveL
Jan 8, 2004, 01:46 PM
Originally posted by mjtomlin
It was believed that the 130nm 970's would be able to hit 2.6, 2.8 at the highest and to get to 3GHz or more we would have to wait for 90nm 980, which is supposedly due out this fall. Well now that the 90nm 970 is out, who knows maybe we'll see new PowerMacs this month or next pushing 3GHz.
I don't recall anyone stating a 2.6-2.8 GHz top end for the 130nm 970. A search of the archives should show that the early info on the 90nm 970 indicated a top end of 2.8 GHz. Although, as you say, these limits may be subject to a thermal target Apple has set, rather than the maximum frequency the process can produce.

Eric_Z
Jan 8, 2004, 01:56 PM
To clarify, SSOI is a combination of the manufacturing technologies Strained silicon and Silicon On Insulator, thus the name SSOI. As it is now Intel uses strained silicon while IBM uses SOI, it's rumored that IBM will begin producing SSOI chips sometime late Q2. It's estimated that IBM will reduce power consumption on there chips by ~30% by using SSOI.

Snowy_River
Jan 8, 2004, 02:15 PM
Originally posted by gaomay
Dual 3.5Ghz would be sweeeeet.

Imagine UT2004 running on that.

Hey, why limit yourself? Why not UT2005?

Snowy_River
Jan 8, 2004, 02:17 PM
Originally posted by mjtomlin
The 2.0GHz? Well Apple may have wanted to keep the price down, there may not be a ton of them available yet, or they didn't want to overstep the towers, which is their bread and butter so to speak.

Or, maybe they're reserving the faster chips for the new Power Macs that will be announced in a couple of weeks? ;)

army_guy
Jan 8, 2004, 02:40 PM
Alot of people seem to be unaware that while we are reaching 90nm technology you are assuming the chips will run cooler, quite the oposite is true.

Taking a 130nm design and reducing it to 90nm will lower the power dissipation to a degree but all is not so jolly the leakage current are much greater meaning higher static power dissipation the P4 prescot 3Ghz=104W under max load, rougly 80-90W when idle so ive been told, though 104W seems suspicious to me considering the 3.4Ghz eats around 140-150W. Cooling for the prescotts is an Intel custom flower HSF design with fan @ 45db noise level with ideal temps around 70 Degrees C, err thats quite hot and loud dont you think what are the load temps?

The die shrink makes cooling the chip more difficult as there is less contact area and with speed boosts you actually end up with a hotter chip than you started with (albiet at a highter speed and also much more noisy fans etc..). There is nothing wrong with this unfortunetly air cooling for me went obsolite more than a 2 years ago and has currently hit a brickwall in terms of performance. Anyone thinking a 3GHz G5 is gona be a cool, quiet, and less power hungry chip is misstaken.

manitoubalck
Jan 8, 2004, 03:16 PM
Originally posted by Macrumors
According to the paper (page 15), the new 90nm PowerPC 970 contains 58 million transitors, uses a 90-nanometer, silicon-on-insulator (SOI)

I'm suprised the transistor cout is so low. The Athlon XP has around 50million and the Opteron has just ove 100, with this intel cores being above that.

manitoubalck
Jan 8, 2004, 03:22 PM
Originally posted by ddtlm
Henriok:Consider AMD's 130nm 2.2ghz Opteron that works fine 2 per 1U machine. One of three things is possible:
a) The 90nm G5 throws off more heat than a 130nm Opteron, or
b) The 90nm G5 is not currently available above 2.0ghz, or
c) Apple decided to make Xserves less compeditive than they could be.

Consider this, maybe the 90nm core gives great chip yields @ only 2GHz.

army_guy
Jan 8, 2004, 03:27 PM
Originally posted by Henriok
The increase of L2 cache on previous PPCs show similar performance boosts as the Pentium. The main thing is that cache is expensive. Just look att the Pentium4 EE with 2 MB cache. It is REALLY expensive compared to its little brothere with just 1 MB. Itanium 2-processors with huge caches (6-9 MB) will cost more a piece than a complete dual-G5. I think it's hard to justify verry large caches, but processors will undoubtfully get higher performance with them. 1 MB L2 cache in 980 seems reasonable though. The 750GX got 1 MB so why doesn't 970? It's really a fair question.

The extremely high bandwidth the 970 and its successors got might be one of the reasons for the farily small cache size and the lack of L3 cache, but it got a terrible overhead so even if it transfers data fast it takes time to initiate a transfer. precious time.

Concerning the shrinkage of die size. Just do the math:
130x130 = 16900
90x90 = 8100
== half.
If everything else is the same, which seems to be the case.

Yes cache is expensive because large caches mean more silicon area hence a higher failier rate, you think throwing away a 9MB cache Itanium 2 is cheap because of a few dodgy transistors in the caches.

Why are you mentioning the Itanium 2 and saying its expensive, its a workstation ment for EDA tools, the cost of these babies is just a rounding error compared to the software which they run.

army_guy
Jan 8, 2004, 03:34 PM
Originally posted by manitoubalck
Consider this, maybe the 90nm core gives great chip yields @ only 2GHz.

I find it hard to believe this, in my opinion as I said before the 90nm version is very likely running hotter than the 130nm version due to the process shrink and also the 1u heatsinks, bear in mind any 1u rack server runs hotter than thier bigger cousins.

mim
Jan 8, 2004, 03:56 PM
Originally posted by manitoubalck
Consider this, maybe the 90nm core gives great chip yields @ only 2GHz.

It may actually be a bit more like option 'c'.

Consider how Apple got burnt by demand for the original 2gHz machine - I would bet they're playing safe so not to repeat the delivery problems - especially for servers which could hurt their reputation a lot more than for desktop machines.

The move to the 90nm process undoubtedly raised clock speeds. Probably as high as the 2.6 mark that everyone has been speculating about. And with the interest in G5 clusters (and with the release of the server cluster node) I'd guess that Apple is expecting demand for these babies to outstrip supply of the higher clock procs. Or maybe they're just playing it safe.

So possibly they're either stockpiling higher clocked chips, or they're underclocking them. I'd expect a speed bump in a month or so when demand for the servers has been established. And I think then it will be to 2.3gHz or so, leaving another step before 3.0gHz at the end of the year. The last thing Apple need as they're trying to break into the enterprise market is a reputation for late delivery (which they already have to a certain extent).

army_guy
Jan 8, 2004, 04:20 PM
How is Apple gona enter the enterprise market by using ATA drives?

Do you see servers using ATA drives?
OK SATA has some SCSI features but mechanically it is still built to be a desktop drive.

Who puts a drive that is meant for desktop use (home/games/multimedia/Digital content creation etc....) into a server?

Where is SCSI which is designed to run 24/7, has outstanding reliability, performance especially under heavy/constant/multiple user load. Iam afriad ATA dont cut it for servers period even for lowend ones, people would probably use WD Raptors.

Yes SCSI is expensive but how valuable is your data?
Can you put a price on it?

Common Philosophy of a SCSI drive in my opinion:

Reliability 1st
Performance 2nd
Total Cost of ownership 3rd

I can only see Apple entering the low-end server market nothing more.

army_guy
Jan 8, 2004, 04:39 PM
Originally posted by mim
It may actually be a bit more like option 'c'.


So possibly they're either stockpiling higher clocked chips, or they're underclocking them. I'd expect a speed bump in a month or so when demand for the servers has been established. And I think then it will be to 2.3gHz or so, leaving another step before 3.0gHz at the end of the year).

I bet the problem for apple is keeping these chips cool, i think that even 2.6GHz with the 970 would be possible just by moving to liquid cooling, of coarse this would be no easy feat to adapt this kind of cooling technique for home use although it has gone mainstream via complete kits.

Henriok
Jan 8, 2004, 04:46 PM
Originally posted by sergeantmudd
The Pentium4 has gone from like 1.4 ghz to 3.6ghz, and with Prescott will get pushed even further.That's like saying that the G3 design pushed from 200 MHz to 1.1 GHz. It did but 750 and 750GX are quite different chips produced at completely different fabs. IBM introduced copper, SOI, larger and on die L2 cache and faster/wider memory buses on its road to 750GX.

There are many cores and development stages in a Pentium 4-design just as there are in the G3 or G4-designs.

The Willamette/Northwood/Prescott Pentium4s going from 1.5 to 4 GHz (+266%) on 130 nm and 90 nm is directly comparable to the the development Motorola did from 7450@533 MHz to 7457@1.42 GHz (+266%).
Look! Motorola did just as good as Intel! And.. Intel haven't got to 4 GHz just yet.
Letting the 970 go from 1.4 to 2.8 in just one step is not to shabby at all. Doing it in under a year is fenomenal.

If you look at the individual 750 models, they didn't scale at all. They are introduced at one frequenzy range and then they didn't climb. 970 was supposted to range from 1.4 to 1.8 GHz, Apple pushed it to 2 GHz. I have never seen any official data on the complete 970-family ranging higher than 2.5 GHz, and that'd be the 970FX I presume.

AppleInsider's sources clamied that 970 will only be updated once, to 90 nm, before 980 will take its place. 980 will, according to AI, be updated two times after the first model. In this matter I have no objection to what AI's sources claims.

I guess that 970 and 970FX will scale past its initial range when the fab mature and they eventually get enough individual cores that can be clocked past 3 GHz to start selling them. But I'm certain that, in volume, 980 will get there first and that it's that chip that Steve was referring to.

And.. I'm pretty certain that Intel have introduced their processors at an intentionally low frequency just so they can "speed bump" them a few months later. Intel's processors have almost always have an insane potential for overcloking. This is not the case with IBM and Motorola who have been under quite a lot of pressure just to keep up.

Originally posted by mjtomlin
It was believed that the 130nm 970's would be able to hit 2.6, 2.8 at the highest and to get to 3GHz or more we would have to wait for 90nm 980, which is supposedly due out this fall. Well now that the 90nm 970 is out, who knows maybe we'll see new PowerMacs this month or next pushing 3GHz.
The "970" that was supposted to go to 2.8 GHz was the 90 nm variant, ie 970FX. The fastest 130 nm 970 I've seen even suggested previously is 2.4 GHz! That would be one hot processor. Not likely!

A die shrink in the past have always resulted in speedbump in the ammount of 30-50% if all other factors stay the same. That would be 970 @ 1.8-2.0 GHz shrunk to 970FX @ 2.7-3.0 GHz while beeing an optimist. 970FX @ 2.6 wouldnt surprise anyone, but pushing up to 3 GHz anytime soon would be VERY surprising. At least to me.

Originally posted by Code101
I don't know why there not using SSOI!IBM are planning for a SSOI deplyment on their 65 nm-process. I guess they might quicken their pace since Intel is getting there first. Intel on the other hand has mountains of problems with power consumptions, IBM has not.

Originally posted by ddtlm

The 130nm G5 wasn't maxed out at 2.0ghz; 2.0ghz was pretty much it's launch speed. 2.4ghz seems like a very attainable speed at 130nm, and I don't think asking 25% higher for the 90nm chip is unreasonable. If you ask me, people are so set on the idea of a "980" that they are underestimating the scalability of the 970.With all due respect, we know almost nothing about the scalability of 970. AFAIK, it has already surpassed its original top speed at 1.8 GHz by 11% to 2 GHz. Going to 3 GHz from 1.8 GHz is a 66% increase and that's too much to ask I think. Why do you think 2.4 GHz is a very attainable speed for the 970 design? Care to elaborate?

If 970 goes to 2.4 GHz (33% above original spec) and 970FX gets a 50% increase due to die shrink to 90 nm, and it then too get a 33% overclock speed bump just for the presumed exceptional scalability, we are pushing for 4.7 GHz.. Not likely!
Even increasing it a mere 25% due to die shrink and then boost it 33% due to the supposed greatness would put 970FX at 4 GHz. Not likely either.

I have seen no evidence at all for supporting any claims for exceptional scalability of the 970 design or IBM's fab. Heck.. even going up to 2 GHz seems to be pushing it with heat sinks the size of bricks and the necessity of 9 fans.

Intel would **** their pants though :) If tye arn't already.

Originally posted by army_guy
The die shrink makes cooling the chip more difficult as there is less contact area and with speed boosts you actually end up with a hotter chip than you started with
This seems reasonable. Question: Might using dual core chips or insanely large caches be way to enlarge the die area for cooling purposes? Not the only purpose of course.

Originally posted by army_guy
Why are you mentioning the Itanium 2 and saying its expensive
BEcause it IS expensive. And Intel initially intended it to take over the world and then some.. Guess they have had some time to rethink that strategy.

army_guy
Jan 8, 2004, 05:04 PM
They intended on taking over the world what with $5000 machine i doubt it.

The Itanium is expensive for the performance you get compared to say a SUN machine but if the software was actually thier it would sell in the intended market however iam talking 8-10 years ago, roughly around 1992 EDA companies started compiling all thier EDA tools to 64-bit SPARQ machines and hence SUN still remains number 1 in terms of software tools and support of which not even HP or IBM can match. EDA companies cannot just sudenly rewrite, recompile and recertify thier tools, it takes money lots of money and its the reason why so few tools are available on the Itanium.

And Hspice does not count in my opinion as Intel is sysnopsis's biggest customer they were litrally forced to port Hspice from SPARQ to Itanium. Itanium is a good product but it needs far too much effort for the performance which you get in the end.

Itanium pricing starting from $5000

where does it end Ill tell you id say around $34000 for a fully loaded system with 16GB of RAM and TOP spec DUAL 1.5GHz CPUS.

Expensive for a machine that only has 1 major EDA application what a joke. You can get 2 TOP SPEC SUN BLADE 2500's and run them using the distributed networking.

nek
Jan 8, 2004, 05:05 PM
The reason for the Xserve to be limited to 2GHz at moment is because speed is more important in the Powermac. I expect faster Powermacs to be announced within a couple of weeks. Possibly at 2.2, 2.4, and 2.6GHz.

army_guy
Jan 8, 2004, 05:09 PM
Using DUAL cores could be a way out in terms of thermals but would require complicated power regulation (5/6 phase) on the MB depending of the power requirements but that wouldnt be a problem id worry about the yield thats a 120M transistor chip @ 90nm running at more than 2.2-3GHz thats gona be no small feat and how much $$$

I would say the 970 should be able to reach 2.6GHz at 130nm (I struggled to get my Opteron to 2.4Ghz on air cooling and thats a 100M transistor chip so I would assume the G5 could clock higher as its only 59M )considering the size of the HSF's in the G5 although Iam not certain of the noise???

mjtomlin
Jan 8, 2004, 05:57 PM
Please people, this is a rumors site. Keep your facts to yourself. This is all about speculation. :p

Some guy at Apple (at the show) said the rumor sites were way off on what's coming up. So that obviously means 2.8-3GHz PowerMacs. Do I have to draw you a picture or something? ;)

And then of course in the Fall, we'll see the PowerPC 980 @ 65nm running at 3.5-4GHz.

It's all about under-promise, over-deliver. This is IBM, not Motorola! Geesh!

ddtlm
Jan 8, 2004, 06:01 PM
Henriok:

The "970" that was supposted to go to 2.8 GHz was the 90 nm variant, ie 970FX. The fastest 130 nm 970 I've seen even suggested previously is 2.4 GHz! That would be one hot processor. Not likely!
Since when has any of this been more than rumors and vauge comments by IBM people?

With all due respect, we know almost nothing about the scalability of 970. AFAIK, it has already surpassed its original top speed at 1.8 GHz by 11% to 2 GHz. Going to 3 GHz from 1.8 GHz is a 66% increase and that's too much to ask I think. Why do you think 2.4 GHz is a very attainable speed for the 970 design? Care to elaborate?
Yields on chips improve over time as the maker figures out how to improve the process in small ways and as they correct little problems. I'm sure you know this, Intel and AMD have pushed all of their chips a lot farther by retirement than they would go at launch. For this reason I expect that 2.0ghz is not any sort of ceiling for the 130nm G5. 2.2ghz is pretty much gaurenteed since 2.0ghz was apparently yielding great early on in the game, and I'm guessing 2.4ghz would be pretty doable. 2.6ghz sounds impractical to me.

Intel would **** their pants though If tye arn't already.
I think AMD's almost has them pooping already. :)

Snowy_River
Jan 8, 2004, 07:51 PM
Originally posted by army_guy
I find it hard to believe this, in my opinion as I said before the 90nm version is very likely running hotter than the 130nm version due to the process shrink and also the 1u heatsinks, bear in mind any 1u rack server runs hotter than thier bigger cousins.

There's one problem in your logic. The primary cause of heat production has to do with the resistance within the chip. If you shrink the die, then you can get the transistors closer together -> less resistance -> less heat. Yes, it is true that there are some issues that come up around conducting the heat away from the chip, but historically those are dwarfed by the benefits of a smaller die. It is this reduction in heat that allows the processors to be sped up without over heating them. Thus you get a faster processor at the same temperature as you had before by using a smaller die.

ddtlm
Jan 8, 2004, 08:26 PM
Snowy_River:

The primary cause of heat production has to do with the resistance within the chip. If you shrink the die, then you can get the transistors closer together -> less resistance -> less heat.
Leakage is becoming a major problem at 90nm, which is making Intel's chips generate a lot more heat than they "should". IBM is probably effected by the same thing, to some extent. I'd believe the 90nm G5 produces less heat than a 130nm chip at the same clock, but its probably not as large of a savings as previous process shrinks would suggest.

ClimbingTheLog
Jan 8, 2004, 09:02 PM
Originally posted by army_guy
...
Who puts a drive that is meant for desktop use (home/games/multimedia/Digital content creation etc....) into a server?
...
Yes SCSI is expensive but how valuable is your data?
Can you put a price on it?
....


RAID.

Most servers will run just fine with a nice RAID-50 setup. The caches are appropriately large and they run one IDE controller per drive so a bum drive doesn't hose the bus.

You see, the problem is SCSI drives are nearly 10x as expensive as IDE drives. So, unless you need massive IO (database server) where SCSI's extra heads help, it's hard to cost justify SCSI drives. Fiber Channel makes it even more irrelevant.

Of course, SCSI drives are almost always more reliable. An IDE drive will last you two years, a SCSI drive, probably 5.

But they're both obsolete in two years anyhow.

If the machine is in a remote location or labor is otherwise prohibitavely expensive (changing drives every so often) and performance isn't paramount, Western Digital offers Server-class IDE drives for about a third of the cost of SCSI.

It's a shame - I like SCSI as a protocol, but the economics aren't there anymore for general server work.

Telomar
Jan 9, 2004, 01:17 AM
Originally posted by Snowy_River
There's one problem in your logic. The primary cause of heat production has to do with the resistance within the chip. If you shrink the die, then you can get the transistors closer together -> less resistance -> less heat. Only problem is as you move transistors closer together and shorten gate lengths the electrical leakage through the transistor goes up meaning more heat is produced. This actually goes up with every process shrink and just so happens to go up quite significantly on 90 nm processes.

army_guy
Jan 9, 2004, 01:40 PM
Yes MORE heat is produced, not only that you to to dissipate it over a much smaller contact area (die size) hence cooling techniques such as liquid cooling/TEC are needed just to reduce the temperature and noise which would be generated with an air cooler. I dont know about everyone else but aircooling went obsolite at 130nm if not before. There is no solution to the heat problem at the chip level, 65nm will be worse and so on. 1st we had no cooling, then passive cooling (heatsink), then active cooling (FAN + Heatsink) now manufacturers have to move to the next step.

army_guy
Jan 9, 2004, 01:41 PM
Originally posted by Telomar
Only problem is as you move transistors closer together and shorten gate lengths the electrical leakage through the transistor goes up meaning more heat is produced. This actually goes up with every process shrink and just so happens to go up quite significantly on 90 nm processes.

This guy knows what hes talking about!

army_guy
Jan 9, 2004, 02:20 PM
Pentium Tejas 2.8GHz 90nm sample 775 pin presumably the pentium 5 i think, not really sure but it eats 150W.

Either its a fake or its a multicore (unlikely with only 775 pins?) for it to chew 150W or it could have one MASSIVE CACHE (2Mb @ 2.8Ghz? ) Intel tell me the transistor count!.


No way your gona cool this with an HSF and at the same time expect a noise level less than 45-55db's ,id say minimum water cooling preferably with a TEC.

Not only that either 5 or 6 phase power regulation. Assuming a 1.55V core if not much higher this thing is gona rip with 100 Amps of current requirments.

http://www.anandtech.com/cpu/showdoc.html?i=1943

Henriok
Jan 9, 2004, 03:36 PM
Originally posted by army_guy
Pentium Tejas 2.8GHz 90nm sample 775 pin presumably the pentium 5 i think, not really sure but it eats 150W. OMG! Even if it is a prototype, how on earth are they going to get that down to managable levels? And.. Isn't Tejas supposed to be introduced at ~4 GHz? I can't even imgine how much power that little sucker will draw at 4 GHz if it consumes 150 W at 2.8.

wms121
Jan 9, 2004, 04:05 PM
..i.e.,

to run "only" 64 bit code..do you need a BIOS switch and/or does the new 90 nm version have one ( from 64/32 mode to "only 64 bit" mode).

Some compilers have internal switches, and/or have to reformat drives..or recode Solid State Disk Drives. 64bit Java can do this in software provided the 'virtual servers' have at least one genuine 64bit capable hard drive.

Any hints on what big Steve is going to announce later this month?

<---embedded slave in motoland

mim
Jan 9, 2004, 10:13 PM
Originally posted by army_guy
How is Apple gona enter the enterprise market by using ATA drives?

Who puts a drive that is meant for desktop use (home/games/multimedia/Digital content creation etc....) into a server?

I can only see Apple entering the low-end server market nothing more.

You're completely right about the reliability of scsi drives. However, in a raid (5 say) this is really made redundant, as you're considering the reliability of the entire system, which then comes down to how quickly can you get the replacement part, and how quickly you can install it. And on top of that, the G5 servers have pretty damn good (disk) performance for a really good price.

Google have made the case for using cheap, easily replaceable parts to maximise reliability. I think Apple will set a similar trend (yet again) for storage. And yes, I think that lots of enterprise level clients will think that, now the performance is sorted out, xraids will be a really attractive storage solution.

army_guy
Jan 10, 2004, 09:01 AM
Originally posted by Henriok
OMG! Even if it is a prototype, how on earth are they going to get that down to managable levels? And.. Isn't Tejas supposed to be introduced at ~4 GHz? I can't even imgine how much power that little sucker will draw at 4 GHz if it consumes 150 W at 2.8.

Forget 4GHz, as clock vs power is not linear, for 4GHz maybe 250-350W, even with a 200-250W vapochill wouldnt be able to cool that!!!

army_guy
Jan 10, 2004, 09:07 AM
Originally posted by mim
You're completely right about the reliability of scsi drives. However, in a raid (5 say) this is really made redundant, as you're considering the reliability of the entire system, which then comes down to how quickly can you get the replacement part, and how quickly you can install it. And on top of that, the G5 servers have pretty damn good (disk) performance for a really good price.

Google have made the case for using cheap, easily replaceable parts to maximise reliability. I think Apple will set a similar trend (yet again) for storage. And yes, I think that lots of enterprise level clients will think that, now the performance is sorted out, xraids will be a really attractive storage solution.

I agree but then again iam thinking of the mechanical issues of an ATA drive under SCSI based conditions. It will come down to how long the ATA drive will last until it fails and how many times that drive will have to be replaced. Still maybe for the lowend it should be OK but anything other than that it wont. Mid/Highend servers are expensive anyway so theres nothing to gain in using ATA drives, thats not the matket apple is trying to penetrate I assume.