Hironobu Sakaguchi's opinion of cell and PS3

[The_legend_of_drtre]

When quoting evidence to back up a statement, you need to include a source, as the source adds or detracts credibility. eg. If what you are quoting is an IBM document on Cell, it has a lot more weight than if what you are quoting is an analysis at ArsTech, or a forum post at GAF, or even something you just made up! We need to be able to check sources to know information provided is trustworthy.

So they just made all this up?

http://en.wikipedia.org/wiki/Cell_microprocessor

 

[The_legend_of_drtre]

Wow is that SOOOOOOOO wrong...

Oh really?

http://www.zdnetindia.com/news/hardware/stories/134099.html

"Each Cell chip has a straightforward PowerPC 970 processing core, called the Power processor element (PPE), that can run Linux and other readily available software. What makes Cell unusual--and hard to program--are the eight accompanying synergistic processing elements (SPEs). Each of these special-purpose engines can run carefully crafted programs and has its own connection to memory and to the other cores on the Cell chip. "


Looks like i was SOOOOOOOOOO right...

The PPE is based off the 970.

You were saying?

Add to the fact that it was even said on this forum:

http://forum.beyond3d.com/archive/index.php/t-31059.html

"Beginning 2000 (Emotion Engine 2)
-Named PowerCore.
-Various Cores with MIPS architecture
-VLIW architecture
-Perhaphs 667Mhz of clock speed.

2001:
-IBM enters in the design
-VLIW cores evolves to SPE (SIMD with Local Store)
-Cell architecture is defined
-Power4 Core (Power4 has 2 cores) at 1.4 Ghz.

2003:
-Power4 1.4 Ghz changed to PowerPC 970 2Ghz.
-IBM promises to Microsoft, Apple, Sony and Toshiba the creation of a 970 at 3Ghz.

2004:
-IBM cannot put the 970 at 3Ghz and they create the PPE. PPE is 970 but without OOO (reducing the number of instructions from 5 to 2) but adding Multithreading and Virtualization.-Cell Prototype is finished

2005:
-IBM adds VMX-128 in the PPE.
-Cell is released to the market in the form of Blade Servers for the US Army and Workstation for Mercury (made for medical application)."

 

[nAo]

Oh really?

Looks like i was SOOOOOOOOOO right...

The PPE is based off the 970.

You were saying?

You (and that article) are wrong, it's a well known fact (not to you obviously) that PPE is derived from an internal research project at IBM, the PPC 970 derivative story is an old urban legend, this info comes straight from IBM engineers mouth at ISSCC 2005:
http://www.realworldtech.com/page.cfm?ArticleID=RWT021005084318&p=10

Neither microarchitectural details nor the performance characteristics of the POWERPC Processing Element were disclosed by IBM during ISSCC 2005. However, what is known is that the PPE processor core is a new core that is fully compliant with the POWERPC instruction set, the VMX instruction set extension inclusive. Additionally, the PPE core is described as a two issue, in-order, 64 bit processor that supports 2 way SMT. The L1 cache sizes of the PPE is reported to be 32KB each, and the unified L2 cache is 512 KB in size. Furthermore, the lineage of the PPE can be traced to a research project commissioned by IBM to examine high speed processor design with aggressive circuit implementations. The results of this research project were published by IBM first in the Journal of Solid State Circuits (JSSC) in 1998, then again in ISSCC 2000.
The paper published in JSSC in 1998 described a processor implementation that supported a subset of the POWERPC instruction set, and the paper published in ISSCC 2000 described a processor that supported the complete POWERPC instruction set and operated at 1 GHz on a 0.25µm process technology. The microarchitecture of the research processor was disclosed in some detail in the ISSCC 2000 paper. However, that processor was a single issue processor whose design goal was to reach high operating frequency by limiting pipestage delay to 13 FO4, and power consumption limitations were not considered. For the PPE, several major changes in the design goal dictated changes in the microarchitecture from the research processor disclosed at ISSCC in 2000. Firstly, to further increase frequency, the per stage circuit delay design target was lowered from 13 FO4 to 11 FO4. Secondly, limiting power consumption and minimize leakage current were added as high priority design goals for the PPE. Collectively, these changes limited the per stage logic depth, and the pipeline was lengthened as a result. The addition of SMT and the two issue design goal completed the metamorphosis of the research processor to the PPE. The result is a processing core that operates at a high frequency with relatively low power consumption, and perhaps relatively poorer scalar performance compared to the beefy POWER5 processor core.

 

[The_legend_of_drtre]

It's a "new core" just like the 750cl is a "new core" It is still based off the 970 just like the 750cl is based off the 750 cxe.

http://www.embedded.com/columns/showArticle.jhtml?articleID=188103194

"Note that although the central processor is based on IBM's PowerPC architecture, it's a new design, not an existing PowerPC core. Cell's central processor is similar to the current PowerPC 970 chip, although it's not an exact match."

It's a new design but it is still very similar to the 970. Which is what... exactly what I said?

The PPE is a 970 derivative with less instructions and a higher clock speed.

 

[nAo]

It's a new design but it is still very similar to the 970. Which is what... exactly what I said?

The PPE is a 970 derivative.

Stop quoting random crap from the internet and stop making up things, IBM engineers at CELL presentation confirmed that CELL PPC core IS NOT derived from PPC 970, period, case closed.
Now stop this or go to troll on some other forum.

 

[The_legend_of_drtre]

Stop quoting random crap from the internet and stop making up things, IBM engineers at CELL presentation confirmed that CELL PPC core IS NOT derived from PPC 970, period, case closed.
Now stop this or go to troll on some other forum.

How am I trolling? You asked for proof and I gave it to you. You just didn't like the outcome. I see nothing in your link that says the PPE is not derived from the 970.

I didn't make up anything. Stop living in fairy land and read the damn articles I posted.

I'm sure you think these guys don't know what they are talking abouth either right?

http://www.avsforum.com/avs-vb/showthread.php?t=758390

 

[nAo]

How am I trolling? You asked for proof and I gave it to you. You just didn't like the outcome. I see nothing in your link that says the PPE is not derived from the 970.

If you see nothing buy a new pair of glasses. In that articles IBM engineers said PPE is based on a new core that was an internal IBM project, this project IS NOT PPC970 and it's not a derivative of it, in fact you can check the original work in the bibliography you can find at the end of that article.

I didn't make up anything. Stop living in fairy land and read the damn articles I posted.

The articles you quoted are simply wrong. Who do you trust? some wikipedia article or the guys who have designed that damn thing?
The PPC970 story/rumour ran for months and months, articles popped everywhere on the net, too bad it was wrong.
BTW..still waiting for you mirable definition of fully featured PPC core, just to have a laugh or two.

 

[The_legend_of_drtre]

Like I said. The PPE is based on the 970, your link does not state otherwise. It just says it's a new core. Ok.. Broadway is new too but it's still based off of Gekko. Find me an article on IBM's website for the PPE that does not mention the 970. Everywhere the PPE is talkined about the 970 is brought up well... why? because the PPE is a variant of the 970.

Good day.

 

[nAo]

Like I said. The PPE is based on the 970, your link does not state otherwise.

IT does, read the damn bibliography at the end of it, it has NOT originated from the 970, do your homeworks before telling other ppl they're wrong.
970 is a completely different beast, it has OOO Execution, vastly different execution units and overall organization compared to the PPE which is an in order processor.
They are so different that saying that one is derived from there other one is madness.

Find me an article on IBM's website for the PPE that does not mention the 970.

Can you give me a proof that santa claus does not exist?

Everywhere the PPE is talkined about the 970 is brought up well... why? because the PPE is a variant of the 970.

With a logic like that you can't fail, that's guaranteed.
Anyway, already wasted too much time on this, please go on without me.

 

[archie4oz]

The PPE is a 970 derivative with less instructions and a higher clock speed.

This isn't even true, because the PPE implements more of, optional PowerPC Book II and III instructions than the 970 does. Plus the PPE no-ops dst* instructions instead of cracking them and serializing them like the 970 does (which sucks), which is the proper way to handle them if you're going to rely on dcb* instructions to hint the hardware prefetchers (an annoying gripe I have with the 970).

Nor is the PPE even close to a 970 derivative. The pipeline structures are *TOTALLY different (which is a *BIG* hint to you that it's a different core than the Power4/970/Giga core is). The instruction fetch and dispatch are about as different from each other as you can possibly be while still being from the same ISA: hell in that regard it has more in common with the 7xx/74xx than it does with the 970.

You should start by reading this (warning: PDF), then move on to some processor manuals...

 

[Fafalada]

I'm sure you think these guys don't know what they are talking abouth either right?

Hi Deadmeat.

(why?!?)

PS3 PPE is PPC compliant, 360 PPE is not. What the exact reasons were to maintain compliancy you should ask IBM Sony and Toshiba engineers involved.
SPEs ISA is very different from VMX, so there's really nothing much to compare. The most you can talk about is that it was designed following the same basic principles as VMX - some of which I adamantly disagree with.

 

[crystalcube]

Everywhere the PPE is talkined about the 970 is brought up well... why? because the PPE is a variant of the 970.

thats brilliant. Now we know whenever RSX is talked about Xenos is brought up and vice versa. Now I know its because one is variant of other.

You just cleared all the confusion

 

[-tkf-]

Like I said. The PPE is based on the 970, your link does not state otherwise. It just says it's a new core. Ok.. Broadway is new too but it's still based off of Gekko. Find me an article on IBM's website for the PPE that does not mention the 970. Everywhere the PPE is talkined about the 970 is brought up well... why? because the PPE is a variant of the 970.

Good day.

Here is a suggestion, when you start questioning actual IBM developers, you know the guys that made the chip your trolling about, please backup your claims with some incredible hard evidence. In this case the only thing you can do is to buy a Cell CPU and the 970, send them off to a company that can rip the top off and take a few pictures, now if the PPE and 970 looks the same (cache, logic etc at the same places) then you may have a case.

Actually i think someone can dig up those things if they bother, maybe you should have a go.. but remember... no photoshop!

Hi Deadmeat.

hehehe

 

[StefanS]

Point. Set. Game.

I think the issue has been cleared up and settled. Thus locked.