DeanoC said:
This is a non-issue for the time being.
So, 1 PE after-all or is this just for GDC 2005 ?
- Thread starter Panajev2001a
- Start date
This is a non-issue for the time being.
Brimstone said:
I can understand why he emphasized that this was a rumour only.
The "what IBM is doing is moving around parts on the CPU to make them look different but the technology is the same!!! Also he said they are hoping that Sony and Microsoft don't figure it out..." part doesn't sound too credible, to say the least.
I think we have to accept that the details are still firmly under NDA. They won't be forever.
Entropy, thanks for that background.
You're absolutely correct, I was thinking of the developer comments I'd read here. I can't actually recall reading the IBM PR quoted. I find it odd that given the reasoned points that respected developers on this board have made, someone would think I was creatively interpreting vauge PR statements from the past when talking about Xenon's processor design.
DaveBaumann said:
You're absolutely correct, I was thinking of the developer comments I'd read here. I can't actually recall reading the IBM PR quoted. I find it odd that given the reasoned points that respected developers on this board have made, someone would think I was creatively interpreting vauge PR statements from the past when talking about Xenon's processor design.
I'm repeatedly asking DeanoC about the definition of "clean sheet design", even citing the vague IBM PR from the past, but he's been very quiet so far.function said:
Nevertheless I respect his words, for when all business logic and time constraint suggest that it's not a clean-sheet design, only he as a developer affirms that it's a clean sheet design, then there's something wonderful in the Xenon CPU which made him think like that I imagine.
For example, they can add a custom feature to discourage those who attempt to run a Linux kernel on it.
A clean sheet design can be one that is not similar to any of IBM's current offerings, and should a CPU such as desicribed by the MS patent be the case then that would indeed constitute a clean sheet design.
As for "all business logic and time constraint suggest that it's not a clean-sheet design" - IMO you are presupposing MS's busness logic, the targets they have and the goals they are setting without adequate knowledge of it as yet; it would seem that many people are hanging a lot of opinion on the one "cost" issue, when we don't really know how much of an issue that it is for what they are projecting for future projects that directly and indirectly relate to Xenon and the use of the technology from it.
As I said previously, I think your "time constraint" though could well turn out to be incaccurate - I know when MS tendered to the graphics companies and I know they had certain things in mind for the configuration of the graphics because they already had a lot of thoughts on the types of CPU capabilities - the resultant configuration of the graphics processor is still congruent with those initial ideas. Given this I believe they were already working with IBM on the CPU and the timescales are very much inline with what we'd expect for a new processor development.
As for "all business logic and time constraint suggest that it's not a clean-sheet design" - IMO you are presupposing MS's busness logic, the targets they have and the goals they are setting without adequate knowledge of it as yet; it would seem that many people are hanging a lot of opinion on the one "cost" issue, when we don't really know how much of an issue that it is for what they are projecting for future projects that directly and indirectly relate to Xenon and the use of the technology from it.
As I said previously, I think your "time constraint" though could well turn out to be incaccurate - I know when MS tendered to the graphics companies and I know they had certain things in mind for the configuration of the graphics because they already had a lot of thoughts on the types of CPU capabilities - the resultant configuration of the graphics processor is still congruent with those initial ideas. Given this I believe they were already working with IBM on the CPU and the timescales are very much inline with what we'd expect for a new processor development.
function said:
And of course I could be wrong, its not like I know exactly how MS and IBM built this thing...
I think I regret saying clean sheet, its been quoted to many times...
What I know of its origins is that its not an existing processor design you can buy or read any public information on, that MS asked and got some specific hard to make features (that IBM had problems with) and its quite powerful.
Yes, there are no 3 cores offering from IBM at the moment, so its a unique chip that IBM is selling to MS.
Even, the current XCPU would fit the clean sheet design as well.
Beside real clean sheet design doesn't always turn out to be good either. So why are people here making a big deal out of clean sheet design is beyond me.
Thanks for the respect BUT your logic and time arguments are based on the idea that IBM doesn't speculate. You assuming that IBM didn't start early...one said:
What if IBM started building the processor design before they knew the customer... They have managed to corner the console processor market, they didn't do it by being shy, I think its reasonable to assume they were VERY aggresive in the selling and I strongly suspect they started early so that it was almost madness not to use there designs...
V3 said:
We don't know how many cores there will be on a single CPU module and I think there is a good likelyhood of that changing over the course of the consoles lifetime.
What do u mean it could be changed over the course of the console? You mean from here to launch day? Cause it's unlikely that the architecture will be changed once the platform is released. Actually, make that impossible.
london-boy said:
I'm refering to the old "silicon shrink cost reductions".
The patent design is one that is flexible in terms of both CPU cores per module (chip) and number of chips. Lets suppose that the "6 threads" figure heard long ago was the case - initially these could be achieved with 3 modules each housing 2 cores - later in the lifetime, when process costs allow it, they may move to just two CPU's with 3 cores each, and then perhaps even later, a single CPU with 6 cores.
DeanoC said:
Oh, that's a very interesting argument. What you suggest is, an exact description of off-the-shelf technology solution, isn't it?
Off the shelf of IBM Engineering & Technology Services, I mean. It's known that Power is one of the most flexible architectures in the world, and IBM is known by its IT service these days. They customize their Power processors to needs of customers, from laser printers to Mac G5. Their customization menu has many items and items increase day by day, adding what IBM speculates possible customers would like to see.But from 2004 things are progressing further, as IBM started the Power Everywhere campaign and now allowing other parties to modify their Power architecture. The Xenon CPU may be one of the initiative processors in the campaign, but still it's different from what eetimes called as "clean sheet design".
That doesn't necessarly mean clean sheet design though, it could be just derived from other technologies that IBM has in their stockpile.
But like I said before this clean sheet design you commented is rather silly. Only people at IBM know for a fact its clean sheet or not, and that wouldn't get public anyway.
You should have told us, the difference between this chip and the current existing ones. That would make a more interesting discussion.
But I know you're under NDA, and we here respect that, so for the time being we just have to discuss these silly things. Before more info goes public.
It'll be a single chip from the start. At the moment its 3 cores capable of dual threads each. That may change when the actual specs is out ie cram in more core, larger die, but I'll be suprised if its 3 seperate chips.
With all the talk about inexpensive design, and only to go with 3 seperate chips, that's just silly.
Based on what?
Where is this talk coming from?