Remember this post by Tommy McClain about XBox2?
- Thread starter bbot
- Start date
You lost me MFA. Do you mean inter-thread vs. intra-thread parallism ?
If you do: I don't see any reason for M$ not to use a CMP style system. Multiple simple cores that use SIMD to exploit data-level parallism (which is what CELL does, but not with the same memory model) is certainly a way to go.
Cheers
Gubbi
If you do: I don't see any reason for M$ not to use a CMP style system. Multiple simple cores that use SIMD to exploit data-level parallism (which is what CELL does, but not with the same memory model) is certainly a way to go.
Cheers
Gubbi
Am I correct that PS3 will use the 1 tflop of computing power to do REYES rendering? Since Xbox2 will use a different method of rendering (mentioned earlier), then having 1 tflop would be wasteful. And therefore a lower performance general-purpose cpu, like Pentium M, would be totally adequate for the job.
bbot said:
You can do anything with it (to an extent), including shader programs. REYES was just an example we discussed due to it's unconvential nature [for consumers].
First off, it's really not different - just more flexible. So, answer the second part yourself.
As per what Dave said earlier, I think he's a bit off on this one. Perhaps if he'd think back to Microsoft's desire for a more specilized CPU (What was it, vector constructs in the MPU?) and then think about what Mfa been talking about - he'd change his mind.
PS. As for your other question from earlier about what 65nm can do, a fellow from Intel has stated that at 65nm they'll have a Billion+ transistor chip constructed of a 4-core Itanium and 16MB of shared cache - by 2007.
Gubbi said:
I can't believe Microsoft is going to use a Pentium-M. Hardly anything has been revealed about what ATI got from the Intel liscence deal. What if ATI got a license for Intels upcomming Arbiter bus technology? The odds of a multi-core CPU would certainly be greatly increased for XBox 2. The secreacy is probably to keep Sony guessing what is going to happen on the CPU front.
From that Design Chain article:
" As of this writing, Kahle's team is busily building a set of complete schematics for the masks that the factories will use to manufacture the Cell chip. The manufacturing design will utilize 90nm geometry and silicon-on-insulator (SOI) technology, which increases transistor switching speed by reducing capacitance (build-up of electrical charges in the transistor's elements), and thus reducing the discharge time and power requirements."
90nm.
" As of this writing, Kahle's team is busily building a set of complete schematics for the masks that the factories will use to manufacture the Cell chip. The manufacturing design will utilize 90nm geometry and silicon-on-insulator (SOI) technology, which increases transistor switching speed by reducing capacitance (build-up of electrical charges in the transistor's elements), and thus reducing the discharge time and power requirements."
90nm.
bbot said:
been explained already.
in laymans terms, at 90nm the Cell chip would be faaar too big and hot, that is IF it would work at all....
not that i have something against big and hot things...
We've been through this sooo many fricking times bbot.
SCEI is mass producing Cell on 65 nm for use in their product, which we all know as PS3.
That 200 billion yen investement was to build 65 nm lines for their version of Cell which will be used in PS3, Toshiba will also mass produce Cell on 65 nm for use in SCEI's products in 2004.
london-boy:
Sorry, but the article clearly states that they're making the mask for cell for the 90nm process. And as for heat, all they have to do is run it at a low clock speed.
Paul,
First of all:
PLEASE STOP THE PROFANITY. Thank you.
Second:
On second thought, it doesn't matter if Sony doesn't use the 90nm process. All that matters is that someone (IBM) uses the 90nm process. That might explain why on that slide it has 1 processor core having the performance of 1 gflop rather than the expected 32 gflops. Perhaps they're going to start with a mature process (90nm), with a relatively low clock speed. Then later they'll improve the performance of cell chips.
Sorry, but the article clearly states that they're making the mask for cell for the 90nm process. And as for heat, all they have to do is run it at a low clock speed.
Paul,
First of all:
PLEASE STOP THE PROFANITY. Thank you.
Second:
On second thought, it doesn't matter if Sony doesn't use the 90nm process. All that matters is that someone (IBM) uses the 90nm process. That might explain why on that slide it has 1 processor core having the performance of 1 gflop rather than the expected 32 gflops. Perhaps they're going to start with a mature process (90nm), with a relatively low clock speed. Then later they'll improve the performance of cell chips.
bbot said:
yeah, then Sony wont be able to say "look we have a 1Tflop+ system here!!!". which we know they somehow will one way or another, even if that means cutting on other things. like, the cover,.... they could just sell the internal bits still not assembled and leave it to the customers to put it all together... to cut costs!!
just kiddingit might be fun now that i think about it....
a PS3-KIT, PS3 DIY.... how cool is that... errrrrrrrr...
zidane1strife
Banned
As stated by IBM Cell is being designed to deliver teraflops of computing performance.
IBM isn't going to use Cell in a small device such as a PDA or Playstation; but rather huge servers. In servers, heat doesn't really matter as you can have HUGE cooling solutions you cannot have in a game console. So in essence, IBM can get away with having it at 90 nm, whereas SCEI which is making a 1TFLOPS class microprocessor for a console can't have a huge coolilng solution in a ps3. This is why they need cell on 65 nm.
There was none, but nice scapegoat though; to slightly mask the fact that your grasping at straws but don't worry, everyone see's through it.
You don't GET IT. That slide with 1GFLOPS was from blue gene; Cell = not blue gene, it's an entirely new architecture.
Cell and blue gene are TOTALLY DIFFERENT THINGS, does this just go over your head? Do you listen? I'm curious.
Every sony exec has stated that Cell = 1tflops or multi teraflops of computing power.
Joe DeFuria
Legend
Is there some source, somewhere, that has any real indication of the Cell configuration, specific to the PS3?
I mean, "Cell Technology" is destined for several different applications, and it seems to me that "cell specs" are bandied about without much reverence to the specific application that it's targeted for.
I mean, "Cell Technology" is destined for several different applications, and it seems to me that "cell specs" are bandied about without much reverence to the specific application that it's targeted for.
Joe DeFuria said:
The staff at the enquirer don't have much faith in the PS 3 "Cell Technology" for 2005.
http://inquirerinside.com/default.aspx?article=9053
Strong words coming from the enquirer.
Joe DeFuria
Legend
Inquirer said:
Brimstone said:
If that is true (standard "Inquirer is the analyst" disclaimer applies), then of course nVidia is missing out on a huge opportunity with the lost X-box contract. (The more bad news for Sony, the better the news for Sony's competitors...)
Similar threads
