The Faceless Master
Member
since nobody has said it yet...
sell your xbox2's and gamecube2's, it's all over!
sell your xbox2's and gamecube2's, it's all over!
CELL PROCESSOR AND PS3 details
- Thread starter Shapingo
- Start date
If the battle is about graphics...Nintendo shouldn't care. I don't think they are going to be affected because the PS3 outrageous performance. They never claimed they were going to push awesome graphics...The people who wants to buy Nintendo games will get the Nintendo system even if it isn't the most powerful one.
Another thing is the XBOX. The entire marketing has been based on the fact that the system was the most powerful...This could damage a lot their userbase.
Well, there is nothing confirmed but if this is true next gen will be very unbalanced...We still have to hear Nintendo's take on the affair.
Another thing is the XBOX. The entire marketing has been based on the fact that the system was the most powerful...This could damage a lot their userbase.
Well, there is nothing confirmed but if this is true next gen will be very unbalanced...We still have to hear Nintendo's take on the affair.
Shrike_Priest
Member
translation from the japanese page:
"Then the interesting topic how many baskets will be introduced. First,
those where most it is observed are announcement of the 1st generation
Cell processor which is by IBM, SONY and Toshiba. Total 5 the
dissertation announcement which is planned in regard to Cell.
According to summary, this Cell processor is the multi core processor
which is produced with 90nm SOI process, it has become SoC where eight
streaming processor cores are loaded onto 1 tip/chip. It operates at
power supply voltage, 1.3V as for operating speed as for 4.60GHz and
operating temperature reaches to 85 degrees, as for inter- connecting
between the tip/chip it says that it has the transmission rate of
6.4Gb/s. According to the authorized personnel, with ISSCC it does not
pass the dissertation which the operational sample does not do, so,
therefore this Cell seems already the sample is operating."
Not much new, although it does seem as though the clock-frequency on the CPUs is also 4.6ghz, and it would seem that they have the chip up and running.
"Then the interesting topic how many baskets will be introduced. First,
those where most it is observed are announcement of the 1st generation
Cell processor which is by IBM, SONY and Toshiba. Total 5 the
dissertation announcement which is planned in regard to Cell.
According to summary, this Cell processor is the multi core processor
which is produced with 90nm SOI process, it has become SoC where eight
streaming processor cores are loaded onto 1 tip/chip. It operates at
power supply voltage, 1.3V as for operating speed as for 4.60GHz and
operating temperature reaches to 85 degrees, as for inter- connecting
between the tip/chip it says that it has the transmission rate of
6.4Gb/s. According to the authorized personnel, with ISSCC it does not
pass the dissertation which the operational sample does not do, so,
therefore this Cell seems already the sample is operating."
Not much new, although it does seem as though the clock-frequency on the CPUs is also 4.6ghz, and it would seem that they have the chip up and running.
Panajev2001a
GAF's Pleasant Genius
aaaaa0 said:
Cut the ALUs clock in half and the Pentium 4 will slow down to a crawl... that is how the design was thought out... a different design would have performed much better with regular faster but not double-pumped ALUs.
That bus (from the APU core to its LS), if we are talking about the same bus, is 256 bits wide: this would mean 153.6 GB/s of internal bandwidth.
The buss that conencts APUs and the PU is 1,024 bits wide, but it would not run at full speed.
MightyHedgehog
Member
ourumov said:
I disagree with this. While their marketing in the first year or so was about touting the console's comparative ability, their main push was always the features and software...Live, games like Fable, Halo 2, Blinx, etc. The last two years are where MS has sold the most of their system, I believe...based on having the games that they've pushed as the reasons to own a system. If we're going to use your reasoning, PS2 did the same when it launched, with much more hyperbole and general all-around bullshit than any of the others this gen, IMO. It always seems to start this way anymore...tech-dazzle, then games...NEC and Sega started this tech-shit with the 16-bit systems.
In the end, the games and, to a lesser extent, the features exclusive to each system are what will sell the systems.
Anyway...*puts thread back on track*...incredible numbers!!
To the hardcore gamer, yes. To the BIG mass...well, it deppends. And yes, Sony began the game. Anyways delete my post from your mind...
After reading B3D I am seeing that things aren't so easy as they seemed regarding the famous "1 TFP". Seems those figures are for just a PE....(85º)
After reading B3D I am seeing that things aren't so easy as they seemed regarding the famous "1 TFP". Seems those figures are for just a PE....(85º)
It's a bit too vague to say...
"with heatsink" is that supposed to mean with no fan? If that's the case I'd say that's not particularly hot at all.
Then there's the issue of what "operating temp" really means
I remember back in the day when my K6-2 300 would run ~30degrees while I would be say, typing text in notepad - but then running Unreal would cook it up into 80ies as well... Yet chip was "operating" during both of these
And obviously there's questions like whether this is the same config that will be used in PS3? Bigger? Smaller?
[quote="mrklaw]If 1PE-CPU could match the whole of Xenon, then maybe Sony won't have a GPU - maybe it'll all be done in software?[/quote]
They'll have a GPU, I'm quite sure of that by now. Absolutely no idea what it may be like though.
Panajev2001a
GAF's Pleasant Genius
Fafalada said:
Well, squeeze more balls and get an answer
.Panajev2001a
GAF's Pleasant Genius
Fafalada said:Believe me I've been trying
You just admitted you have not been trying hard enough
.Come on, come on: GOOOOOO FAF!!!!!!!! UNLEASH THE POWER OF THE MUUUUUMUUUUUs!.
After-all, why do you think you have kept Alex on top of his game if not for situations like this
Panajev2001a
GAF's Pleasant Genius
MrSingh said:
YOU SO SEXY!
Doom_Bringer
Banned
any gpu details yet?
deadhorse32
Bad Art
MrSingh said:
That man is willing to TALK. GO PANA SQUEEEEEEEEEEEEEEEEEEZZZZZE !!!!
Panajev2001a
GAF's Pleasant Genius
gofreak said:Pana, we need your analysis/insight/explanations etc..
Let me digest...
This adds some new info... but I am afaid of jumping the gun trying to speculate what each of the new info actually entails... this is meant to be a tease presentation...
Start a bit of the hype before CES 2005 and the Xbox 2/Xenon presentation and then go full force some weeks after the CES 2005 Xbox 2 presentation with PlayStation 3 and CELL news
.Brief analysis: they beat my expectations in terms of clock-rate and of a better PU ( main CPU) than I expected.
90 nm tech and 4.6 GHz ? If they can solve all the kinks with their 65 nm CMOS6 SOI manufacturing process this could mean a few things.
1) lower power consumption at the same clock-speed which should allow a good fan and a good heatsink to keep the CPU cooled at 4-4.6 GHz in PlayStation 3.
2) more e-DRAM (or e-DRAM at all if the 90 nm version did not use any) and more cache present in the system.
3) 1) and 2) compromised with importance given to higher yelds for high volume manufacturing (which means lower production cost).
I expect Sony to experiment more with 65 nm and with two solutions: 1 PE and 2 PEs based CPUs.
I think that with the frequency and temperature head-room they have that the final 65 nm solution uses 1 PE (I am making the hazardous assumption that this announced CELL chip uses a single PE).
This would mean a peak of about 256-294.4 GFLOPS for the CPU at a clock-speed betweeen 4 and 4.6 GHz.
I know that size would drop quite a bit going from 90 nm to 65 nm, but we do not know the cost of each chip and what is the problem they want to address the most: Intel can push out in 130 nm technology a 400+ mm^2 Itanium 2 CPU at about $140 per chip (internal production cost)... let's say that the 90 nm CELL CPU they are presenting costs about the same amount of money to produce in high-volumes, maybe SCE wants to spend less than that on CPU amnufacturing and that is why they will be pusihing for 65 nm.
SolidSnakex
Member
mumu said:
Panajev2001a
GAF's Pleasant Genius
gofreak said:Cheers Pana, if anything more strikes you, do let us know
Also, a question: how do we know an APU can do 8 flops per clock? (sorry, this is probably backtracking on old territory for you)
Because that is the peak of its SIMD FP capabilities (it can do scalar and vector processing): a 4-way parallel FP MADD implies 2 FP ops * 4 and it is executed in 1 cycle (pipelined).
Register (128 bits): X Y Z W (each field is 32 bits for single precision SIMD computation).
We have 4 FMAC (FP multipliers/adders/subtractors) and each can do a FP MADD (fused multiply-add) per clock-cycle (pipelined).
X1 = (X2 * X3) + X4
Y1 = (Y2 * Y3) + Y4
Z1 = (Z2 * Z3) + Z4
W1 = (W2 * W3) + W4
This is an example of a 4-way parallel MADD (without broad-cast IIRC).
Panajev2001a
GAF's Pleasant Genius
Fafalada said:
HAHAAAA... Alex-unmasked
.Panajev2001a
GAF's Pleasant Genius
border said:
:lol, why would the LS's clock-speed be synchronized with the XDR signalling-rate when you still have to go through the PE bus, the DMAC and the XDR Memory Controller (XMC) to access the XDR RAM ?
Panajev2001a
GAF's Pleasant Genius
teiresias said:
XDR specs sheets.
That's what I get for not knowing my RAM specs. It just seems like an odd practice to spec something by throughput on a pin rather than overall. I'm just out of the loop on memory and everything graphics. Damn my computer security oriented graduate thesis. :lol
EDIT: Ummm, yeah, it would have also helped if I'd noticed that was Gb with a little 'b' :lol
EDIT: Ummm, yeah, it would have also helped if I'd noticed that was Gb with a little 'b' :lol
Read this over at B3D just before I went to bed. Looks like that thread exploded. I've got some fun lunchtime reading now. 
Anyway, this is what I alluded to in that Xbox2 thread. The theoretical specs thrown around in that thread seem off what Sony's been going after. And like I said, CELL is all about bandwidth. The eDRAM is the real clincher, b/c with enough of it, they'll free up bottlenecks. But the design seems ideal for graphics since it seems to address the issues of bandwidth first, which is like the problem in every design.
4.8GHz is fast, I wasn't aware they were shooting for a system clock this high. Then again, I don't remember the patents specifying. Clearly Faf, Pana and others have been following this closer than I have. I'm gonna try and catch up now, it's about time for the next gen hardware hype wars, right? Anyway, so far so good. Looks like the numbers are falling into place. Some people thought Kutaragi was off his rocker, but the guy is ambitious. The PS3 will be outperformed in a year of its release, but CELL is probably ahead of its time.
That said, has anyone updated the block diagram for CELL? I know it was speculation, but with this updated information, the picture has become a little clearer. PEACE.
Anyway, this is what I alluded to in that Xbox2 thread. The theoretical specs thrown around in that thread seem off what Sony's been going after. And like I said, CELL is all about bandwidth. The eDRAM is the real clincher, b/c with enough of it, they'll free up bottlenecks. But the design seems ideal for graphics since it seems to address the issues of bandwidth first, which is like the problem in every design.4.8GHz is fast, I wasn't aware they were shooting for a system clock this high. Then again, I don't remember the patents specifying. Clearly Faf, Pana and others have been following this closer than I have. I'm gonna try and catch up now, it's about time for the next gen hardware hype wars, right?
Anyway, so far so good. Looks like the numbers are falling into place. Some people thought Kutaragi was off his rocker, but the guy is ambitious. The PS3 will be outperformed in a year of its release, but CELL is probably ahead of its time.That said, has anyone updated the block diagram for CELL? I know it was speculation, but with this updated information, the picture has become a little clearer. PEACE.
sonycowboy
Member
gofreak said:
:lol
Deadmeat has been debunked, ridiculed, and laughed out of every forum he posts in: Beyond3d, GAF, & even Opa. (and you gotta really suck to get embarressed there)
:lol
Deadmeat history:
The PS2 will cost $700
The PSP will cost $500
The PSP's battery lasts for 90 minutes
Cell = 1.15GHz and doubles as a toaster
Doom_Bringer
Banned
sonycowboy said:
shit, that is hilarious :lol
Panajev2001a
GAF's Pleasant Genius
gofreak said:
I think that figure comes from deep inside his butt.
Supposely quite a while ago someone visited SCEA offices and managed to sneak-a-peek at some CELL WorkStation HArdware and it read 1.15 GHz in the BIOS screen.
If you ask around though you will hear laughter coming from the industry sector and people with ties to EA and SCE/SCEJ... which does not help if you know Deadmeat.
sonycowboy
Member
Panajev2001a said:
Can we get some layman's interpretation of what we know of Cell vs Xenon architecture? Who's winning the power battle?
Panajev2001a
GAF's Pleasant Genius
sonycowboy said:Can we get some layman's interpretation of what we know of Cell vs Xenon architecture? Who's winning the power battle?
It is too early to tell, but if PlayStation 3 ships at least 3 months after Xenon there is at least the theoretical chance of an over-all power advantage, but at the end each unit will be able to do some things better and some things worse I think like it was the case for PlayStation 2, Xbox and GameCube.
Similar threads
- Poll