Update:

>> Probably Consoles as well.

> "01net also claims to know some of the technical specifications of the new console
> (translation from Develop): 'CPU is custom IBM PowerPC with three cores [...]'"
> http://en.wikipedia.org/wiki/Project_Caf%C3%A9#April_2011

More rumours, this time regarding another console:

"IBM sources claim that the new multi core PowerPC processor, which Big Blue, has spent several years developing is now part of a joint development project with the Japanese company. [...] The new 32nm Cell processor is tipped to be capable of up to 16 SPEs which is twice as fast as the current Cell processor according to IBM leaks."
http://www.smarthouse.com.au/Gaming/Industry/F5C6F8A6

(I remain very sceptical regarding this specific article as it contains some severe factual flaws.)

I've heard the 3 core IBM Wii2 a few times now. Sounds perfectly plausible.


Sony recently said they're not spending as much on the PS4 development so if anything that points to a new version of Cell. Anything else means they'll have to rewrite everything and that'll be rather expensive.

Adding a second Cell, upping the RAM and updating the GPU is the easiest option technology wise and gives them backwards compatibility. At 32nm they can probably up the clock quite a bit as well. IBM already had better Cell tech in the works so it could be a better version of Cell. IBM didn't stop Cell development (they publicly canned one chip and everyone thought Cell was dead). They had been working on a 32SPE + 4 PPE chip with the PPEs based on a POWER7 core. Don't know what happened to it but they said they were merging Cell with something else, presumably POWER.

> Adding a second Cell

As I read the article linked to above the PS4 is rumoured to be powered by a single (new) Cell processor with 16 SPEs. Maybe this new Cell will have two PPEs (like the cancelled PowerXCell 32ii was supposed to have, only with half the amount of SPEs) but I doubt they will put two Cell chips there as I think it wouldn't make sense if they're going to have a new Cell chip developed anyway.

> IBM didn't stop Cell development

Indeed.

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=6768&start=35
https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=6993&start=80

> they publicly canned one chip

Actually, it was two chips.

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=6993&start=70

> everyone thought Cell was dead

Yes, which was surprising as the declaration of death was qualified almost immediately:

http://www.hardwareheaven.com/news.php?newsid=344

> They had been working on a 32SPE + 4 PPE chip

...and one with 2 PPEs.

> with the PPEs based on a POWER7 core.

Interesting, I hadn't read that anywhere before. I'd rather have suspected a POWER6 offshoot. No IBM roadmap document on PowerXCell 32ii and 32iv I ever came across mentioned the base of their enhanced PPE called PPE' therein. A quick googling revealed this as the source for the POWER7 claim (translated):

http://translate.google.com/translate?sl=ja&tl=en&u=http%3A%2F%2Fd.hatena.ne.jp%2Fpetacell%2F20080621

Thanks for this bit of info.

> they said they were merging Cell with something else, presumably POWER.

Yes, that's what I understood as well (see first two links).

SPEs servicing a much more powerful out of order Power processor core?
Sounds great. Should be much more powerful then any proposed Cell derivative.
The only negative to this is the attendant increased complication creating optimized code for any CPU coupled with SPEs.

> Should be much more powerful then any proposed Cell derivative.

That depends on the definition of "Cell derivative". I suspect most people would have called 32 SPEs servicing two or four trimmed POWER7 cores, all packed together in a chip named PowerXCell 32ii or PowerXCell 32iv, a "Cell derivative". I'd say the inclusion of the SPEs is what makes a Cell a Cell. Thus any number of SPEs servicing one or more trimmed POWER8 cores, all packed together in one chip, would be called a "Cell derivative" by most people as well, I guess. That said, it's interesting to think about whether non-trimmed POWER8 + SPEs would be percieved as POWER with added SPEs or rather as Cell with POWER PPE ;-)

> The only negative to this is the attendant increased complication
> creating optimized code for any CPU coupled with SPEs.

At least compared to current Cell chips there would be no such increase. And regarding the PPE an out-of-order core requires less optimization than an in-order core ;-)

>I'd say the inclusion of the SPEs is what makes a Cell a Cell.

THAT'S an interesting definition Andreas.
From my examination of the two Cell processor that were made and the immediate potential successors, I thought the shift to a Power 6 successor (as the core CPU) was really significant.
Of course, the SPEs have also evolved. CellBE's SPE were much less powerful then its sucessor's SPEs.
So both the Core CPU (PPE) and its attendant SPEs continue to evolve.
Power7's core processor should be far beyond any previous PPE core. As we don't yet know how much improvement has been made in the SPE's perhaps my biased focus is unjustified.
Is it a Cell derivative? Yes, I guess it is. But as a Power6 successor Its also so much more.

> Power7's core processor should be far beyond any previous PPE core.

Yes, that's a given, even if stripped down. The same applies to POWER8 to an even greater extent. If the PS4's Cell should have a POWER PPE I'm inclined to believe it will be derived from POWER8, not from POWER7 like was planned for the PowerXCell 32ii and 32iv.

> we don't yet know how much improvement has been made in the SPE's

True. According to IBM roadmaps the PowerXCell 32ii's and 32iv's eSPE was not supposed to improve in SP FP or DP FP performance over the PowerXCell 8i's eDP SPE. Some suspected cache enlargement and latency reduction though.

http://www.google.com/search?q=powerxcell+edp+espe

> Is it a Cell derivative? Yes, I guess it is.

So we have the same point of view. Nice.

> But as a Power6 successor Its also so much more.

That's called product line evolution ;-)

>That's called product line evolution

Product evolution I first heard mention of here, thanks to your tireless investigations.
At the time you first mentioned this, I thought no ******* way.
I've got to stop being surprised when you correctly intuit something. You don't speculate or predicate. You make some very accurate analysis.

I wonder how this will affect Power's use in consumer oriented products.
Do you think we might see a return to the desktop or use in devices like the PS4?

> I wonder how this will affect Power's use in consumer oriented products.

Probably not that much, assuming you mean IBM POWER here. Non-trimmed POWER will never be in any consumer product, no matter if with or without SPEs. The rumour article I quoted says that Sony wants to use the PS4's Cell also in notebooks and in TVs. So if this new Cell is going to have a trimmed POWER core and will be in notebooks and TVs then that would mean (some) usage in consumer oriented products. But as you might have noticed, these are no less then three ifs: 1. Will the PS4 really use a Cell? 2. If it'll use a Cell, will its PPE really be derived from POWER? 3. If it's a Cell with a POWER(-derived) PPE, will it really be used in notebooks and TVs?

> Do you think we might see a return to the desktop

No, not in a scale that would generally be understood as a return (to what we had with the POWER4-derived PPC970).

> or use in devices like the PS4?

Yes, I think a new Cell in the PS4 might have a trimmed POWER PPE. But that's really just speculation.

>The rumour article I quoted says that Sony wants to use the PS4's Cell also in notebooks and in TVs. So if this new Cell is going to have a trimmed POWER core and will be in notebooks and TVs then that would mean (some) usage in consumer oriented products. But as you might have noticed, these are no less then three ifs: 1. Will the PS4 really use a Cell? 2. If it'll use a Cell, will its PPE really be derived from POWER? 3. If it's a Cell with a POWER(-derived) PPE, will it really be used in notebooks and TVs?

Sony has made statements that might be interpreted as indicating that the design would be more evolutionary (rather then revolutionary). As such a Cell seems likely. But a Power derived Cell is a complete unknown. Therefore, whatever else they use it in could but may not be Power related.


>> Do you think we might see a return to the desktop

>No, not in a scale that would generally be understood as a return (to what we had with the POWER4-derived PPC970).

But Sony's potential use in computing devices would be, at least, a small return.

>> or use in devices like the PS4?

>Yes, I think a new Cell in the PS4 might have a trimmed POWER PPE. But that's really just speculation.

I'm beginning to get that impression as well. Considering the timel ine mentioned for introduction, the next Cell processor used in the PS4 could easily benefit from IBM's developments.

> a Power derived Cell is a complete unknown.

True, assuming you refer to real products and assuming you mean POWER here, not Power ISA in general. As we established before in this thread, both the PowerXCell 32ii and PowerXCell 32iv were suspected to come with PPEs derived from POWER7.

> Sony's potential use in computing devices would be, at least, a small return.

Only if such new Cell with POWER-derived PPE is going to act as the CPU of (at least some of) Sony's future notebooks. As you may remember, when Toshiba was said to have Cell technology in their notebooks all they eventually released were some models in their Qosmio line with the SpursEngine (i.e. only SPEs, no PPE) acting as a video coprocessor chip for the x86 CPU. We'd have to see if Sony's ambitions in that regard would be more serious or if they would replicate Toshiba's way regarding notebooks containing technology based on a new Cell.

> the next Cell processor used in the PS4 could easily benefit from IBM's developments.

That's a given, no matter if the new Cell's PPE will be derived from IBM POWER or be any other Power Architecture compliant core developed by IBM. One of those two options would be realized for a new Cell for sure.

Just to be different, I'll say something on-topic (-:

It was computex this week and a load of ARMs were launched, confirmed or shown off.

TI announced a new OMAP with dual core 1.8GHz Cortex-A9s
Nvidia were showing off their Kal-el quad core A9 chip.
Qualcomm are sampling their new chip soon, it's another dual core but it goes up to 2.5 GHz. It's also a new microarchitecture - There's very little details but given their claims it sounds like it'll be something similar to the Cortex-A15.

Currently the A9 is in PPC G4 territory performance wise.
The 1.8GHz A9s should put them up against the G5s and it looks like ARM beating all the G5s could be maybe only a year away.

Then there's Nvidia's project Denver at the end of next year which sounds like it'll be more in the range of current desktops.

It's gonna get very interesting in the next year or so...

> Qualcomm are sampling their new chip soon, it's another dual core
> but it goes up to 2.5 GHz. It's also a new microarchitecture

Nice. Three days before that announcement I had just been talking about Krait having been announced for sampling for Q2/2011 right here in this thread.

> it sounds like it'll be something similar to the Cortex-A15.

At least regarding Dhrystone performance it's estimated to be a bit below (estimated) Cortex-A15 performance.

> Currently the A9 is in PPC G4 territory performance wise.

G4 exists between 350 MHz and 1.7 GHz (2.0 GHz overclocked), which means a factor of about 5 to 6 from slowest to fastest, so I think you should specify your statement a bit more. So which clock rate in G4 do you think a 1.2 GHz (which I think is the current max) Cortex-A9 roughly resembles performance-wise? After all, beating a 350 MHz G4 would be a no-brainer even for Cortex-A8 ;-)

> The 1.8GHz A9s should put them up against the G5s

That's a bold theory. I think it should be hard for a 1.8 GHz Cortex-A9 to beat even a 1.6 GHz G5 (the slowest of the bunch) let alone a faster one performance-wise.

> it looks like ARM beating all the G5s could be maybe only a year away.

Beating a 2.5 or even 2.7 GHz G5 with a 2.0 GHz Cortex-A9 is definately not possible. First Cortex-A15 based chips are minimum 1.5 years away.

So we've got a couple of years before ARM reaches the processing power of CPUs we are currently targeting?
A 1.8 Ghz A9 still sounds interesting. With built in GPUs, these SoCs are usually pretty low cost.

> Power's use in consumer oriented products

Not about a new Cell, but this started circulating 5 days ago:

http://img12.imageshack.us/img12/1228/prcspec13iu.jpg

Andreas_Wolf,

Multi-core Power6? Wow! How accurate do you think that document is?
Could it really be a Nintendo internal document or just an elaborate fake?

> Multi-core Power6?

I think that would be more like "POWER6-derived", in a "trimmed down" (except for the number of cores as POWER6 is dual-core only) sense.

> How accurate do you think that document is? Could it really
> be a Nintendo internal document or just an elaborate fake?

I really don't know. This spec sheet has been heavily discussed since it emerged and it's far from being undisputed. We'll know for sure in a few days I guess.

Quote:

If it was POWER6 based it would not really be a POWER6 in any meaningful way. POWER6 added all sorts of on-chip stuff that is completely useless for a console.

I think it's more likely to be based on the A2 used in the wire-speed chip, the A2 appears to be based on the PPE from Cell / Xenon.

POWER6 is better on general purpose stuff than the PPE but that's not very important in games.

The FLOPS numbers are weird, it works out at 12.7 instructions per cycle. Rather more likely is a slightly different clock speed. At 3.66GHz it works out to be 12 instructions per cycle - the same as the PPE.

The double precision FLOPS are the same but why on earth would you want such DP power in a console?

Rest of it sounds fairly plausible.

Quote:


I believe they're quite similar clock for clock. However it'll very much depend on the implementation. A9s tend to have bigger caches and faster memory than the G4s ever had.

Quote:

I meant the lower range of the G5s but remember the G4 was faster per clock than the G5.

Quote:

Sure, what I meant was the Qualcomm chip at 2.5GHz. It's sampling now so should be a year (or less) away. It could get close or beat the G5s.

OTOH It looks like Nvidia's "Project Denver" should blow any G5 clean into next week.

BTW most (or all?) ARMs are mode on low power (LP) or general purpose (GP) silicon processes. Nobody makes them on a High Performance process. If they did, clock rates would soar. With all the talk of servers I wonder will anyone actually do that.

> If it was POWER6 based it would not really be a POWER6 in
> any meaningful way. POWER6 added all sorts of on-chip stuff
> that is completely useless for a console.

The spec listing might refer to a custom-built chip with cores derived from the cores of the POWER6 chip so that POWER6's all sorts of on-chip stuff wouldn't play any role anyway.

> the A2 appears to be based on the PPE from Cell / Xenon.

Btw, does the A2 have AltiVec/VMX?

> why on earth would you want such DP power in a console?

Probably for the same reasons Cell's PPE and Xenon's core also have this DP power per clock and core ;-)

> I believe they're quite similar clock for clock.

Okay, so accordingly G4 above 1.2 GHz should still be faster than current Cortex-A9.

> A9s tend to have bigger caches and faster memory than the G4s ever had.

Faster memory bus yes, but bigger caches than G4s ever had? While the core is specified by ARM for up to 4 MiB L2 cache I couldn't find any actually existing Cortex-A9 with more than 1 MiB L2 cache. MPC7448 has 1 MiB L2 as well. MPC745x has less L2 but can have L3, e.g. MPC7457 has 512 KiB L2 and up to 2 MiB L3.

> I meant the lower range of the G5s

That's what I referred to with "1.6 GHz G5 (the slowest of the bunch)" as that is the very bottom of the G5 range. I still doubt that a 1.8 GHz Cortex-A9 would beat it.

> remember the G4 was faster per clock than the G5.

I don't think that's generally true except for AltiVec/VMX code.

> what I meant was the Qualcomm chip at 2.5GHz.

Ah okay, so that was a misunderstanding from my side, sorry. A 2.5 GHz Krait could possibly beat a 2.5 to 2.7 GHz G5, yes.

> It's sampling now so should be a year (or less) away.

I'm not sure Krait based chips are really sampling at 2.5 GHz already. Some sources say the MSM8960 will only clock up to 1.2 GHz and it will take until the APQ8064 to actually reach 2.5 GHz:

http://www.electronista.com/articles/11/06/02/qualcomm.dual..and.quad.core.processors.inbound/

> It looks like Nvidia's "Project Denver" should blow any G5 clean into next week.

Sure but that's even further away than Cortex-A15.

Quote:

Don't know but they could always add it.

Quote:

IIRC the PPE could do 2 Double Precision instructions per cycle. According to these numbers this core can do 7! The only PPC I know of that can do that is POWER7 (it can do 8 actually).

IIRC some of the PPCs used by Nintendo are standard parts with enhanced vector units.
I wonder could this be something like a set of A2s with a POWER7 style vector units bolted on the side. That makes a lot more sense.

Update:

>>> this started circulating 5 days ago:
>>> http://img12.imageshack.us/img12/1228/prcspec13iu.jpg

>> How accurate do you think that document is? Could it really
>> be a Nintendo internal document or just an elaborate fake?

> We'll know for sure in a few days I guess.

We now know that this sheet must be hoax as it talks about a "32nm process" whereas IBM has to say:

"IBM plans to produce millions of chips for Nintendo featuring IBM Silicon on Insulator (SOI) technology at 45 nanometers (45 billionths of a meter)."
http://www-03.ibm.com/press/us/en/pressrelease/34683.wss

No information on core count (just "multi-core") or frequency, unfortunately. And Nintendo doesn't reveal any better info either:

"CPU: IBM Power®-based multi-core microprocessor."
http://e3.nintendo.com/hw/#about

So let's wait and see if the older triple-core rumour turns out true.

>> does the A2 have AltiVec/VMX?

> Don't know but they could always add it.

Yes, of course they could, but what I would like to know is whether IBM's stock A2 core (as used in the PowerEN and Power BQC chips) has it. I ask because I've read conflicting third party statements on that matter (and no statement by IBM itself).

> IIRC some of the PPCs used by Nintendo are standard parts with
> enhanced vector units.

Yes, Gekko is PPC750CXe plus custom vector unit, and Broadway is PPC750CL (which already includes the same vector unit).

> I wonder could this be something like a set of A2s with a
> POWER7 style vector units bolted on the side.

With "POWER7 style vector units" you mean VSX? Thing is Nintendo claims the Wii U to be backwards compatible to Wii. Assuming there's no real emulation involved this means the vector unit of the new CPU must be either identical to or a superset of Broadway's vector unit. This would rule out VSX (as well as any other standard vector unit like AltiVec/VMX). On the other hand, there could always be more than one, but I doubt it ;-)

@minator

ARM for the future!