MacRumors

macrumors bot
Original poster
Apr 12, 2001
54,182
15,982
IBM introduced the Power 5 microprocessor at the Hot Chips conference this year.

Some early notes were posted by CBR Online before the conference last week. Subsequently, Chip-Architect provides some notes and photos from the actual presentation. The notes provide some numbers to the previously reported simultaneous multithreading -- with a claim of a 40% performance increase for multithreaded applications.

eWeek also had notes from the conference and reports the processor is on track to ship in 2004 and start with a 130-nm process and then move on to a 90-nm process.

More details are said to be coming in October at the Microprocessor Forum 2003.

The Power5 is the successor to the Power4 processor which served as the basis for the PowerPC 970 Processors used in the PowerMac G5.
 

gaomay

macrumors regular
May 28, 2002
116
0
Scotland, UK
PowerPC 980

POWER5 sounds good but what about the 980? Any news on that yet? Is that the next generation referred to at WWDC03 by the IBM guy?
 
Comment

arn

macrumors god
Staff member
Apr 9, 2001
15,968
5,126
No new news on the 980.

In fact, it's entirely a rumor-site product... IBM hasn't admitted to the project.

However... seeing as the PowerPC 970 was derived from the Power4.... it's clear that the Power5 technologies will trickle down into whatever the sucessors to the PowerPC 970 will be.

arn
 
Comment

LimeLite

macrumors 6502a
Mar 20, 2003
651
0
Los Angeles, Ca
That was quick Arn! In the time it took for me to hit the comment button, the Power5/Power4 typo was already fixed. I'm impressed.
 
Comment

MrMacMan

macrumors 604
Jul 4, 2001
7,002
11
1 Block away from NYC.
Well again we are only guessing the next Processor will derive from the Power5, it might still derive from the Power4!

Who knows?

Only really IBM and Apple.

The Power5 will be a decent server chip with much improved preformance and new features.

Lets hope the next Processor *does* come from the Power5.
 
Comment

richie

macrumors member
Jul 16, 2002
91
0
Melbourne, Australia
O/T, Sorry

Originally posted by Macrumors
The Power5 is the successor to the Power4 PowerPC processor which served as the basis for the PowerPC 970 Processors used in the PowerMac G5.

So, am I to understand these machines are... Powerful?
 
Comment

cb911

macrumors 601
Mar 12, 2002
4,125
4
BrisVegas, Australia
finally some good news for the week. it's good to hear that IBM is on track with their processor development, even if we're not 100% sure that the Power5 will turn out to be the 980. ;)

in regard to the 'next' processor - i thought that it would be a 970 made with a 90nm process, available sometime in 2004? but i guess if the 980 is here in time...
 
Comment
Re: Power5 at Hot Chips

Originally posted by Macrumors
The Power5 is the successor to the Power4 PowerPC processor which served as the basis for the PowerPC 970 Processors used in the PowerMac G5.

The POWER4 is not a PowerPC processor. It is a POWER (Performance Optimized With Enhanced RISC) processor. The PowerPC processors are derivatives of the POWER processors.

The POWER5 is the same way. Supposedly IBM is making a PowerPC derivative of the POWER5 -- which many people have dubbed the PowerPC 980.
 
Comment

AngryAngel

macrumors regular
Jul 12, 2003
119
0
Re: Re: Power5 at Hot Chips

Originally posted by shadowself
The POWER4 is not a PowerPC processor.

I thought (heard in a forum) that the POWER4 abided by both the POWER ISA _and_ the PowerPC ISA. This may be something that attracted IBM and Apple to use it as the basis for the 970.
 
Comment

Mudbug

Administrator emeritus
Jun 28, 2002
3,848
1
North Central Colorado
So at this point it safe to call this thing vaporware....

As for the 970/980 130nm/90nm question - will there be a really significant speed gain by moving to the 90nm process for the 970, or is that mostly for power consumption/heat? If it's speed, then I wonder if the smaller-process 970 might be a speed equivalent of the larger-process of the 980, but what do I know?

nothing... that's why I guess a lot. :)
 
Comment

MisterMe

macrumors G4
Jul 17, 2002
10,709
63
USA
Re: Re: Re: Power5 at Hot Chips

Originally posted by AngryAngel
I thought (heard in a forum) that the POWER4 abided by both the POWER ISA _and_ the PowerPC ISA. This may be something that attracted IBM and Apple to use it as the basis for the 970.
The original PowerPC was the POWER chip set reduced to a single chip. The Power 4 and soon to be Power 5 are single chip implementations of a more advance POWER architecture. In a sense, the already thin lines between PPC and POWER have blurred. The ISA's are the same except for a couple of instructions added to the PPC 601 (for transition purposes) and AltiVec. Apple used POWER-based IBM RS/6000 workstations to port System 7 to the PPC.
 
Comment

b_riggs

macrumors newbie
Jun 23, 2003
18
0
North Carolina
I just want to emphasize that we simply do not know what Steve was talking about when he said they would have a 3 GHz chip within a year because there are several possibilities.

The most likely seems to me to be a 90nm version of the 970 because there would be no (or minimal) design changes. The second possibility is a new PPC chip based on the Power5.

A new PPC chip based on the Power5 will very likely happen but I think it will be in a little later time frame. And I can't say that I care either, since both chips will move us forward in a nice progression.
 
Comment

Phinius

macrumors regular
Mar 15, 2003
196
0
Los Angeles
Why use the 970 on a 90-nm process?

Some of the features described for the Power5 so far would be very desirable in a 9XX PowerPC processor for Apple's use. Adding SMT to the Power5 chip increases performance by 40%, there are power saving features that turn off parts of the chip and if SMT is not utilized, the instructions per clock cycle increases (IPC). So, at a given clock speed, the Power5 derived chips would be faster than the Power4 derived chips and perhaps use the same or less power.

There is a strong possibility that a Power5 derived chip would match or beat the performance of the Xeon, Itanium, Prescott or Pentium-M chips in 2004. Right now the Power4 is behind in performance compared to Itanium-Madison, and the 2 GHz 970 about matches the performance of a 1.7 MHz Pentium-M.

So why would IBM bother to move the 970 over to a smaller 90-nm process when a Power5 derived 9XX processor could be made? Seeing the frequencies mentioned for the Power5 and the upcoming 9XX chips, it looks very likely that the next version of the 9XX processor will be derived from the Power5 core.
 
Comment

gregorypierce

macrumors regular
Jan 28, 2002
162
0
Originally posted by Mudbug
So at this point it safe to call this thing vaporware....

As for the 970/980 130nm/90nm question - will there be a really significant speed gain by moving to the 90nm process for the 970, or is that mostly for power consumption/heat? If it's speed, then I wonder if the smaller-process 970 might be a speed equivalent of the larger-process of the 980, but what do I know?

nothing... that's why I guess a lot. :)

Decreasing the mm process helps with power consumption and heat, which of course means that the same tech can run at a higher clock and yield better performance.
 
Comment

akac

macrumors 6502
Aug 17, 2003
478
73
Colorado
Re: Why use the 970 on a 90-nm process?

Originally posted by Phinius
Some of the features described for the Power5 so far would be very desirable in a 9XX PowerPC processor for Apple's use. Adding SMT to the Power5 chip increases performance by 40%, there are power saving features that turn off parts of the chip and if SMT is not utilized, the instructions per clock cycle increases (IPC). So, at a given clock speed, the Power5 derived chips would be faster than the Power4 derived chips and perhaps use the same or less power.

There is a strong possibility that a Power5 derived chip would match or beat the performance of the Xeon, Itanium, Prescott or Pentium-M chips in 2004. Right now the Power4 is behind in performance compared to Itanium-Madison, and the 2 GHz 970 about matches the performance of a 1.7 MHz Pentium-M.

So why would IBM bother to move the 970 over to a smaller 90-nm process when a Power5 derived 9XX processor could be made? Seeing the frequencies mentioned for the Power5 and the upcoming 9XX chips, it looks very likely that the next version of the 9XX processor will be derived from the Power5 core.

Well, for one thing - deriving a 9xx processor from a Power5 is going to be a lot of work. We'll probably see it at some point, but probably not for another couple years at best. You don't just run a processor design through a "make for desktop use" macro or something. It took IBM and Apple several YEARS to get the 970 from the Power4 which has been around for awhile.

Of course the Power5 is being built with the 9xx in mind, so it should be easier, but still not anything they'll introduce next summer.
 
Comment

jettredmont

macrumors 68030
Jul 25, 2002
2,731
328
Originally posted by Mudbug
So at this point it safe to call this thing vaporware....

It's way too early to call the POWER5 vaporware (it's on-schedule as far as can be told, and by all measures looks like it will debut normally). The 980? That's rumorware.


[pb]As for the 970/980 130nm/90nm question - will there be a really significant speed gain by moving to the 90nm process for the 970, or is that mostly for power consumption/heat?
[/b]

The main inhibitor of speed in modern CPUs is heat dissipation. If you can pull more heat away from a chip, it can operate significantly faster. If you can have the chip generate less heat to begin with, you have less to have to pull away for it to operate at faster speeds.

If you pay close attention you'll occaisionally hear of Intel running a P4 at 10 GHz or somesuch (in a box chilled to cryogenic temperatures). It's all about the heat. If you can take it away, then processors can run faster (until you hit other limitations, of course).


If it's speed, then I wonder if the smaller-process 970 might be a speed equivalent of the larger-process of the 980, but what do I know?

If you took a old 680040 chip and revved it up to 3 GHz you don't end up with a PPC 970. You end up with a really fast, several-generations-behind chip. The POWER5 (and, presumably, the 980) adds SMT and other significant architectural advances. These don't just affect how fast it processes instructions; they change the fundamental way it processes instructions. There is not, in other words, a single dimension for processor speed, and either one chip is faster or it isn't; there is a whole topology to consider.

No matter how fast the 970 clocks, it won't likely perform hard-core multithreaded tasks faster than a Power5-derived PPC.
 
Comment

jettredmont

macrumors 68030
Jul 25, 2002
2,731
328
Re: Why use the 970 on a 90-nm process?

Originally posted by Phinius

There is a strong possibility that a Power5 derived chip would match or beat the performance of the Xeon, Itanium, Prescott or Pentium-M chips in 2004. Right now the Power4 is behind in performance compared to Itanium-Madison, and the 2 GHz 970 about matches the performance of a 1.7 MHz Pentium-M.

Huh?

The Power4 and Itanium are close enough to be considered equals (and last I heard the edge was on the side of the Power4+, not the Itanium).

Based on IBM's divulged information, the Power5 would absolutely decimate an Itanium chip today. As to what Intel will have out next year: who knows? But the Power5 reportedly performs 4x faster on benchmarks than the Power4+. Given rough equivalence between the Power4+ and Itanium today, that means that the Power5 will perform about 4x as fast as today's Itaniums.

The 2GHz 970 is at least roughly equivalent (in SPEC-style benchmarks) and at best completely blows away (in Photoshop and Mathematica tests, according to those programs' authors) 2.8 GHz P4's. I'm a bit rusty on the rough equivalence between the P-M and P4 lines, but my understanding was that a 1.7GHz P-M was about as quick (on non-FSB-limitted tasks) as a 2.4 GHz P4 desktop (and of course consuming much less power, which is why one would choose the 1.7GHz PM instead of a 2.2GHz P4M).



So why would IBM bother to move the 970 over to a smaller 90-nm process when a Power5 derived 9XX processor could be made?

A Power5-derived processor will not be seen until at least Q4 2004. It would be nice to sell a few chips between now and then. It's not like IBM is starved for talent resources. They can work on both projects at once!

Seeing the frequencies mentioned for the Power5 and the upcoming 9XX chips, it looks very likely that the next version of the 9XX processor will be derived from the Power5 core.

Depends on what you mean vby "version". Speed-bump? Definitely 970. Process change? 970. Redesign? 980/Power5.
 
Comment

jettredmont

macrumors 68030
Jul 25, 2002
2,731
328
Originally posted by gregorypierce
Wake me when they start making microprocessors from diamond instead of silicon. That's the next big advance.

Diamonds for Microprocessors

'Cause, yeah, 300mm diamond wafers are so much less expensive ... :)

Diamond may be a possibility for a next-generation substrate, but it's far from a sure thing (there are other possibilities). And there's a whole world of advances to be made before that. If you wait for the next "big" leap you'll miss the order of magnitude increases due to smaller leaps between.
 
Comment

Daner

macrumors newbie
Aug 13, 2003
9
0
4 vs. 5, 130 vs. 90, 970 vs. ?

I'd be very surprised if we didn't see the 970 die-shrink (with the attendent lower power consumption and heat production as well as potentially greater speed potential) long before we see a consumer version of the Power 5.

Furthermore, just because the Power 5 will debut at 130 doesn't mean that a consumer version of it could not debut at 90.
 
Comment

stingerman

macrumors 6502
Jul 6, 2003
286
0
Re: Re: Power5 at Hot Chips

Originally posted by shadowself
The POWER4 is not a PowerPC processor. It is a POWER (Performance Optimized With Enhanced RISC) processor. The PowerPC processors are derivatives of the POWER processors.

The POWER5 is the same way. Supposedly IBM is making a PowerPC derivative of the POWER5 -- which many people have dubbed the PowerPC 980.

PowerPC's are not derivatives of Power but used the original Power to expedite the implementation of the PowerPC architecture. The spec for PPC Architecture was finanlized first and it was a 64-bit architecture from the start. Power and Power2 were not. When they set out to implement the 32-bit subset of the PPC spec, they used parts of Power to expedite the implementation. They for example did not use the double word precision instructions, and they added about 39 new instrustions. So the first PPC and the first Power were not compatible.

IBM's Power2 was the first of the PowerPC Power processors that were binary compatible with the PPC architecture. Around 1993 IBM did a few internal studies and determined that the PPC instruction set had a better legs as the basis of IBM's future processors. The Power3 processor was the first to implement the full 64-bit PPC architecture. In three major revisions to the Power3, IBM added Copper, SOI and Power Management and better process designs. The Power4 added the dual core, super-scalar pipelines, and higher instruction level parallelism (ILP). The Power5 adds SMT, finer grain power management, shorter internal wiring lengths, etc.

Although the internal engineering of each processor has changed radically, from Power2 forward, the Power family is an PowerPC processor with full PPC binary compatibility. The PPC spec was developed not just by IBM but by Apple and Motorola. Apple is a major contributor to the ongoing innovation found in the PPC families of processors. Two such recent innovations born by Apple are 1. The coherent processor interconnect technology in the 970, and 2. the flexible front-side bus along with the super advanced U3 ASIC controller.

Apple's partnership with IBM is also going to result in the next generation of SIMD vector unit that should appear inside the 980 (G6) processor family.
 
Comment

stingerman

macrumors 6502
Jul 6, 2003
286
0
Re: 4 vs. 5, 130 vs. 90, 970 vs. ?

Originally posted by Daner
I'd be very surprised if we didn't see the 970 die-shrink (with the attendent lower power consumption and heat production as well as potentially greater speed potential) long before we see a consumer version of the Power 5.

Furthermore, just because the Power 5 will debut at 130 doesn't mean that a consumer version of it could not debut at 90.

The 970 will go to 90nm first (before the 980 is released.) and probably before January 2004. This will significantly reduce the cost of producing the 970 as more can fit on each 300mm wafer. Too, IBM will be able to include a 1 and 2 MB L2 cache running at the processors core speed, compared to todays 512KB L2 Cache. Clocks well in excess of 3GHz should be possible. And performance increases should be quite significant. It will be interesting to see if Apple changes the Core:FSB ratio from 2:1 to 8:3 or something like that. I can't imagine a 1.5GHZ FSB, though that would be awesome. The current G5 supposedly can handle a 2.2 GHz G5 out of the box; a 1.1GHz FSB.
 
Comment
Re: Re: Re: Re: Power5 at Hot Chips

Originally posted by MisterMe
The original PowerPC was the POWER chip set reduced to a single chip. The Power 4 and soon to be Power 5 are single chip implementations of a more advance POWER architecture. In a sense, the already thin lines between PPC and POWER have blurred. The ISA's are the same except for a couple of instructions added to the PPC 601 (for transition purposes) and AltiVec. Apple used POWER-based IBM RS/6000 workstations to port System 7 to the PPC.

In reality the original POWER chip was a chip set (5 or 7 chips made up the original CPU). It was considered a very big deal when IBM got it down to a single chip. This single chip version still had the POWER label, NOT the PowerPC label. After getting it to a single chip IBM created other single chip variants including the RAD6000 (a radiation hardened, space qualified version). All of these were called POWER chips. POWER was originally an acronym: Performace Optimized With Enhanced RISC. The "opmised with enhanced RISC" was based upon optimizations and enhancements from the original IBM RT CPUs.

The PowerPC was originally envisioned as a very different chip. It was concieved by Apple, IBM and Motorola (the original AIM group). It was to be a combination of the best parts of the single chip version of the IBM POWER CPU and the best parts of Motorola's 88000 chip set (which had one through 5 chips in the CPU chip set depending upon the exact confinguration used). This combination was to be tuned to the Macintosh System (as it was called back then -- System 6 reigned at that time. Pink was a functional lab experiment. Rhapsody and OS X were not even on the horizon.)

Since the instructions sets were very similar all the compiler houses used the IBM RS/6000 workstations to experiment with their codes and compilers. This included not only Apple, but also MetroWorks (who early on had a MUCH better compiler than Apple), Green Hills, Absoft, and others. The PowerPC 601 was very similar in instruction set to the single chip version of the POWER CPU. The 601 (the long forgotten 602), the 603, 604 and "G3" did not have the AltiVec/Velocity Engine instructions.

Additionally, the various chips have had different levels of symmetric multiprocessing. The first one to have the full standard MERSI set was the 604. The 603 had only MEI (thus BeOS had to do most of their cache checking and such in software). If I recall correctly the G4 has MESI but not the full MERSI. I have not checked on what the G5 has.

Originally posted by akac
Well, for one thing - deriving a 9xx processor from a Power5 is going to be a lot of work. We'll probably see it at some point, but probably not for another couple years at best. You don't just run a processor design through a "make for desktop use" macro or something. It took IBM and Apple several YEARS to get the 970 from the Power4 which has been around for awhile.

Of course the Power5 is being built with the 9xx in mind, so it should be easier, but still not anything they'll introduce next summer.

Yes, it does take a couple years or more to design a chip -- even a derivative chip. However, the original POWER4 was not designed to have a direct derivative PowerPC. There are many, many rumors floating around suggesting the POWER5 was designed with making a PowerPC derivative in the plans from the beginning. Additionally, it has been heavily rumored that there exists an IBM team which has been working on a PowerPC derivative of the POWER5 for almost a year now. Even though these are just "rumors", I doubt it will be two years from now before there exists a PowerPC derivative of the POWER5.

Originally posted by Mudbug
So at this point it safe to call this thing vaporware....

No, the POWER5 is definitely not vaporware. There currently exist systems in test based upon the POWER5. These are server class systems. People don't buy the bleading edge of these systems. They want to know IBM has tested them for several months in various configurations before they ship systems to the enterprise. There are exceptions of course (including the DOE and other USGov organizaitons), but most large enterprises will only commit to these high end server systems after a company had tested them thoroughly in-house first. Large enterprises won't commit the resoures to install these until they can be assured the uptime is not like the typical desktop.

Originally posted by Mudbug
As for the 970/980 130nm/90nm question - will there be a really significant speed gain by moving to the 90nm process for the 970, or is that mostly for power consumption/heat? If it's speed, then I wonder if the smaller-process 970 might be a speed equivalent of the larger-process of the 980, but what do I know?

nothing... that's why I guess a lot. :) [/B]

In reality we all know nothing. The sum total of what we know divided by what we don't know would definitely cause an underflow error on any currently shipping computer system (or in the case of the old Burroughs systems -- which used BCD for computations and the only limit on precision was the amount of memory you had -- a memory overflow resulting in a null answer anyway.) Thus by current binary computations we all know nothing. :) :)

Will a die shrink from 130nm feature size to 90nm feature size give a significant performance improvement? Is 15 to 30 percent improvement significant to you? For most people, probably not. For some people who have highly compute bound processes, maybe so.

No one knows what the performance improvement will be in the POWER5 derivative versus the PowerPC 970. Even the people within the system clearly state any hard number predicted for improvement is pure conjecture. There will be improvement. Of that you can be assured, but exactly how much is pure conjecture -- and probably will be for at least 9 or more months from now.

My apologies for this being sooooo long winded.
 
Comment

macrumors12345

Suspended
Mar 1, 2003
410
0
Re: Re: Why use the 970 on a 90-nm process?

Originally posted by jettredmont
The Power4 and Itanium are close enough to be considered equals (and last I heard the edge was on the side of the Power4+, not the Itanium).


Itanic has a slight FP advantage right now (10-15%) when running a single thread, but of course Power4+ has two cores, and these chips are not intended to just run a single thread. So yes, I would say that Power4+ has the advantage right now.

But the Power5 reportedly performs 4x faster on benchmarks than the Power4+.

Actually, that is 4x faster than the original (1.3 Ghz) Power4, but it's still quite impressive. The Pentium4 and Opteron have yet to beat the FP performance of a single 1.3 Ghz Power4 core (of course, in part that is because they have less memory bandwidth).

Given rough equivalence between the Power4+ and Itanium today, that means that the Power5 will perform about 4x as fast as today's Itaniums.

Well, not quite (see above). And we have to see what Intel rolls out, though so far the only thing I have heard is bigger caches for Itanic. That won't be enough. At the very least, it is probably safe to assume that IBM won't be losing ground with Power5!

I'm a bit rusty on the rough equivalence between the P-M and P4 lines, but my understanding was that a 1.7GHz P-M was about as quick (on non-FSB-limitted tasks) as a 2.4 GHz P4 desktop (and of course consuming much less power, which is why one would choose the 1.7GHz PM instead of a 2.2GHz P4M).

Yeah, that sounds about right, although it should be noted that the 1.7 Ghz Pentium-M is not supposed to ship until next year (Q1). It should be using the 90 nm process...I guess Intel can't scale Pentium M any higher on the 130 nm process. See http://www.digitimes.com/NewsShow/Article2.asp?datePublish=2003/08/18&pages=01&seq=1
 
Comment
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.