Cell article: 40x faster than EE, Cell arch to last 10 yrs
- Thread starter Megadrive1988
- Start date
- Status
If Sony hadn't said PS3 will be 1TFlop, Microsoft wouldn't have announced that xbox360 will be 1TFlop 
That proves Sony said PS3 will be 1 Teraflop.
That proves Sony said PS3 will be 1 Teraflop.
rabidrabbit said:If Sony hadn't said PS3 will be 1TFlop, Microsoft wouldn't have announced that xbox360 will be 1TFlop
That proves Sony said PS3 will be 1 Teraflop.
I don't know about that - sometimes specs just kind of create themselves, and become embedded in the public's psyche. Microsodft might have said the 1 TFlop thing just due to the fact that the PS3 = 1 TFlop thing had become so embedded. Sony actually saying something after all isn't too different from everyone thinking they said something, as far as PR is concerned.
I remeber KK talking 1 Teraflop. I'd have thought passing "Ken Kutaragi" "1 Teraflop" "Cell" would have come up with the goods, but not so.jvd said:
Certainly the Sony contigent have been associated with 1 teraflop for a while, so someone somewhere said it, or hinted at it. Just where's that first source?...
KK did talk about 1Tflop Cell chips long long long ago, but I cannot remember if he meant specifically for PS3's CPU or not. The "ideal implementations" described in the early Cell patents further fueled the 1Tflop "dream".
Either way, it was so long ago that anyone with half a brain would know he was stating goals rather than actual claims for existing hardware (like those claims made at the ISSCC, which are bankable)! What matters now is what they claim at E3.
Either way, it was so long ago that anyone with half a brain would know he was stating goals rather than actual claims for existing hardware (like those claims made at the ISSCC, which are bankable)! What matters now is what they claim at E3.
KK did hint at the Cell iniative to reach teraflop of power performance (like Titanio stated). However, Sony NEVER claimed 1 teraflop for PS3. I have done the search myself as well as others and I extend a challenge to anyone that can find an article where KK stated that 1 teraflop will go in the PS3. [IMO, over the years the two got mixed in with each other]
BOOMEXPLODE
Regular
Jawed said:
It's always true that first generation games on a console only scratch the surface of what's possible. Anything else would be unheard of! The only thing Sony would have to "worry" about is that in 2006 X360 will be starting to get second-gen games while the new PS3 will be getting first-gen games. That might help X360 remain competitive if PS3 is technically superior to a significant degree.
marconelly!
Veteran
It's funny, but Allard and Bach themselves have said that the total power of next Xbox will be 1TF. No ifs and buts were mentioned, and it doesn't get any more official than that. Considering that everything that is known about CPU and GPU doesn't add up to anywhere near that, how they got to that number is a complete mystery.
On the other hand such explicit comment was never made about PS3, it was always talk along the lines of cell chip where it was never clear how many of those chips would have to be used or more importantly what iteration of them they are talking about (it's not impossible to imagine a cell chip five years in the future that will actually have 1TF)
I'm really looking forward to hearing how is that 1TF in Xbox 360 justified, and also to what kind of spin Sony will put on their specs. Because let's face it, we all know they will.
Yeah, you are completely correct. 1TF in PS3 so far has absolutely never been mentioned to my knowledge at least. I fully expect however, that we'll get some crazily spun numbers come E3.
Sony's only worry (imo) is that they will have to show in a visible way, how PS3 is better looking then xbox 360. For cross platform third parties it's gonna come down to art direction. For first parties it's about maximzing the features of the hardware and the art direction.
I think considering the graphics companies invovled, there won't be any significant visual difference between what you can see. The difference will come from the games available, backwards compatability, etc...
I think considering the graphics companies invovled, there won't be any significant visual difference between what you can see. The difference will come from the games available, backwards compatability, etc...
I've just wasted half an hour looking, and can't find a specific quote of PS3 having 1 TFlop. Cell was tooted as being scalable to TFlop, and there's lots of 4 Cells giving PS3 Tflop performance, but no claims as such. The was also talk of PS3 being a supercomputer, which requires >800 GFlop if going by the Top500 list.marconelly! said:
No official statements on this though. Only reports from journalists on what KK/IBM said. It looks like TFlop was a target for 4 Cell BBE and never a claim for PS3. As a reult it looks like MS pulled a TFlop figure from their magical hat for XB360, and now Sony will have to do something similar unless they want the uneducated masses reading this stuff (linked to earlier) to think XB360 is 4x power of PS3. If Nintendo go with be more conservative with their power claims they'll look like they're in the Dark Ages
Shifty Geezer said:No official statements on this though. Only reports from journalists on what KK/IBM said. It looks like TFlop was a target for 4 Cell BBE and never a claim for PS3. As a reult it looks like MS pulled a TFlop figure from their magical hat for XB360, and now Sony will have to do something similar unless they want the uneducated masses reading this stuff (linked to earlier) to think XB360 is 4x power of PS3. If Nintendo go with be more conservative with their power claims they'll look like they're in the Dark Ages
Nah, I think Sony will be fine in that regard, especially considering that the unwashed masses already think that it's getting 1TFlop anyway.
jvd said:
You know, you can go on and on about that 1 tflop figure but in the end, it was clear even 30 pages ago that the 1 tflop figure was always centered around a cell implementation, aka "broadband engine". True, it was very popular belief among many on this very forum that the patent of that cell specific implementation (implementation because CELL is scalable, in other words an architecture) would be the one that would be targeted for PS3.
Kutaragi back in the days when CELL development started already made hints that this would likely be the CPU for the next PlayStation, but it was never officially confirmed right up until long after the patent of the broadband engine was found and throroughly discussed. In the end, it just doesn't hold weight.
So people, lets just recap some facts that have been clear for years now:
1.)
CELL is an architecture, not a specific CPU implementation
2.)
CELL is scalable, so that it can target a broad range of applications (from 3d workstations to tv sets etc)
3.)
a CELL implementation (*) has already achieved 1.6 tflops of performance, yet it obviously isn't an implementation that will go into PS3...
* As far as I'm concerned, CELL isn't just the specific implementation of 1 PPE + 8 SPEs but that also an implementation of multiple "CELLs" (i.e. 4 PPEs, each having 8 SPEs) is "CELL", even if it's a 4 chip solution, though that CELL implemetation would have a name like i.e. "broadband engine".
The only claims that are meaningless and were stated by Kutaragi was the 1000 the power of PS2 claim back about a year into PS2's life. "1000 times the power" is quite a meaningless number that could mean anything. In the end it's pure nonsense for techheads trying to base some kind of meaning out of it - it just doesn't work. It's certainly much different than Allard's comment on Xbox2 having a total of a trillian calculations per second. (Btw, 1 trillian calculations/sec != floating point performance!).
BTW; Isn't ironic that Sony only publicized the number of 66 million polygons/sec perspective transformation number of 1 VU (VU1) when in fact they could have publicised some meaningless number including both VUs? But then again, they probably would have if they knew Microsoft would market their Xbox with meaningless numbers based on an efective fillrate it really never had (micropolygons, based around an inflated fillrate). So much for marketing, ey... too bad, we'll probably see Sony resort to the same level since they clearly have to demonstrate that their later console will be at least equal in performance. :?
Megadrive1988
Veteran
you know its funny, when I first heard that PS3 would be,
or might be,. 1 TFLOP, I was kinda disappointed. to ME, that was a major *downgrade* since, whether or not Sony actually said 1 TFLOP for PS3 specifically, they had *definitally* previously said that PS3 would be 1000 times as powerful as PS2, which of course, at face value, would require a lot more than 1TFLOP. obviously in raw performance you'd need 6.2 TFLOPs of course that is without looking deeper at the architecture and realizing that, even with only ~100 times the raw power and much increased efficiency, rendering features, rendering quality, etc, you could reach "1000 times greater performance than PS2"
nevermind my rambling
or might be,. 1 TFLOP, I was kinda disappointed. to ME, that was a major *downgrade* since, whether or not Sony actually said 1 TFLOP for PS3 specifically, they had *definitally* previously said that PS3 would be 1000 times as powerful as PS2, which of course, at face value, would require a lot more than 1TFLOP. obviously in raw performance you'd need 6.2 TFLOPs
of course that is without looking deeper at the architecture and realizing that, even with only ~100 times the raw power and much increased efficiency, rendering features, rendering quality, etc, you could reach "1000 times greater performance than PS2"nevermind my rambling
considering how one of the VU's was totally crippled, that wouldn't suprise me aif they did include it in the performance figures. PS2 never came close to to the 66 million polygon figure.
Honestly i think Xbox could nearly double the amount of geometry the hardware could transform. You'll need ERP to give you a more accurate idea, but it was certainly no slouch. BOTH PS2 and Xbox published theoretical numbers that weren;'t achiveable in a game situation. I find it really disturbing to see that you would argue something like this, after 5 years of previous arguments about the same thing across the net.
- Status
Similar threads
- Locked