Whoops: XB360 at 2.8GHz

[Carl B]

NDA??!? I thought everyone saw those docs....

Well, just because everyone has seen them, doesn't mean they weren't under NDA. An NDA is only as effective as it's weakest 'link' of course, so to speak. And leaks happen.

Happily I received it when it was still 'secret,' and I can assure you that I didn't feel like being one to leak it. I'm sure that was the case for 99.9% of those that had it. But then someone does, and well there ya go. In fact I don't even remember it being leaked per se; I do remember a lot of the same info being transmitted out of that Spanish conference MS did though on developing for 360. Maybe I was out of town when the doc itself finally got transmitted around, or possibly I just didn't care since I already had it.

The point is though - 3.2 GHz, *not* 3.0 GHz, unless there's something else out there...

 

[Mefisutoferesu]

I haven't seen them, mind posting them for the curious??

 

[Carl B]

Well, that's going to have to be someone else. As far as I'm concerned, I'm not supposed to have them, and I'm not going to violate that in such an overt way as to post them. If they really did flood the web, I'm sure someone else can provide the link. For my part, I'll just post the title of the doc and the stated 360 CPU operational frequency when I return to my abode.

 

[standing ovation]

I hope you're not serious here.

And what if I am?

The multiprocessor design chain is so iterative (not to mention complicated) that it is hard to believe an elementary oversight such as this could go unnoticed by armies of software and hardware architects from both companies.

Therefore, giving developers a less powerful kit had to be a premeditated act.

 

[Carl B]

And what if I am?

The multiprocessor design chain is so iterative (not to mention complicated) that it is hard to believe an elementary oversight such as this could go unnoticed by armies of software and hardware architects from both companies.

Therefore, giving developers a less powerful kit had to be a premeditated act.

LOL, well I still can't really tell if you're joking around or not, so I'll treat the post both ways.

For those that think just because a project is technical and complex, human error will be weeded out via oversight, let's not forget certain recent lamentable satellite missions. This article provides a good summary of all the ones I would have mentioned: Satellite screw ups

As for this particular instance, if ShootMyMonkey's word is accurate, then I really don't see it as anything other than some 2.8 GHz chips from the Beta kits getting mixed in with the chips to be put in to the final kits. Software and hardware engineers would have had nothing to do with it, and technically I need to emphasize that the chips are the same - it's just the 'final' ones are set at a higher clock multiplier.

 

[Jawed]

I think it's also worth pointing out that when you have a dev or debug machine in front of you there's no way to find out what clockspeed it's running at.

Special software is needed. Software that isn't part of the normal dev environment.

Jawed

 

[Fafalada]

I hardly think reading a CPU performance counter constitutes as "special software".

Maybe some people can't be bothered to write those 5 lines of code, but I know getting a new devkit that's one of the first things I'd take a look at.

 

[Carl B]

***UPDATE***

Well, Fouad - I guess you were right. Indeed in Microsoft's NDA'd developer notes, they indicate the final tri-core design at 3 GHz rather than 3.2 GHz. Now I'm not sure what the significance of this is; honestly I still think 3.2 would be what to expect. But indeed you are vindicated in having preached what you were preaching.

Taken from the document: 'Xbox 360 Alpha vs. Final Hardware Performance'

Keep in mind within the doc itself it states that specs could change substantially prior to release, and that the doc itself is an early release. So again I'm still going with the 3.2 GHZ Microsoft stated, but indeed the doc we were both quoting *does* say 3 GHz.

Going back to a question I asked earlier in this discussion, does anyone know the FSB frequency for the XeCPU chipset/CPU? I'm wondering what the clock multiplier would even have to be to run at 3 GHz. For some reason I had been thinking 400 MHz clock multiplier increments before.

 

[Andy]

Would dropping the processor speed down from 3.2Ghz to have 3.0Ghz have any reprocussions of the RAM speeds as well? eg Cell is at 3.2Ghz, XDR RAM is also at 3.2 GHz

 

[Carl B]

Speaking in the hypothetical of a 3 GHz XeCPU, I don't think it would effect the RAM speed.

The speed of Cell and the XDR are linked due to the FlexIO. In the case of the 360 the memory controller's on the GPU anyway, and it doesn't have a problem with asynchronous operation.

 

[Lysander]

Before e3 times, when I looked at cell and X2cpu it was like: cell@4ghz, X2cpu@3ghz. Then at e3 cell and X2cpu got even @ 3.2ghz and that how it stayed until now. Perhaps you are all reading old papers. Even in papers for cell today there is still 4ghz note.

 

[SenatorMonkey]

xbdestroya - that doc you're referring to is from last December. Everything recently has explicitly stated 3.2GHz, so this shouldn't be a concern.

 

[Carl B]

Is it? I have no idea - mine's not dated in any appreciable way. I've been a believer in the 3.2 GHz thing anyway so that's fine by me. December though? That seems a little early to have had a document describing the changes from Alpha to Beta floating around, but I'll readily accept it as fact. I only got involved in this aspect of the thread in the first place because I *thought* I was proving that the docs stated 3.2 GHz.

 

[doob]

And the impact of having some early Devkits with 2.8 or 3Ghz instead of 3.2Ghz like its meant to be on final consumer hardware ... is???

 

[AlNom]

Is it? I have no idea - mine's not dated in any appreciable way.

Nothing in the file creation date?

 

[standing ovation]

For those that think just because a project is technical and complex, human error will be weeded out via oversight, let's not forget certain recent lamentable satellite missions.

We need only point to the whole Xbox Protection Cord fiasco to validate Murphy's Law. (Personally, I think this lawsuit-in-the-making is what hastened the console's demise and its successor's introduction to market.)

Software and hardware engineers would have had nothing to do with it, and technically I need to emphasize that the chips are the same - it's just the 'final' ones are set at a higher clock multiplier.

Curious executives aside, who else could be mulling around clean rooms but über-geeks? I mean it's not as if the chips were in mass production at the time. And speaking of which, who else would be tinkering with this stuff -- assembling and testing dev kits -- but company nerds?

 

[Carl B]

Nothing in the file creation date?

Well, the file I have was created in June - but I don't know if that really means anything as I don't know if that's reflective of the original creation date or not; that's just when I got a hold of it.

Anyway if people have questions about this doc go ahead and PM me. Though ironically this thread already had it's primary question addressed a couple of pages ago (that the reason some final kits have 2.8 GHz chips is due to a mix-up), I feel like there's better things to discuss here than the listed clockspeed in that doc.

I for one would like to know the clock multiplier of the XeCPU, simply to undertand better how many clock steppings off from the final the beta chips that went out were. (just for personal curiosity - not because it matters)

 

[Li Mu Bai]

Is a possible 200mhz downgrade only bothering insecure XBX ******s? From any angle, the PS3 trumps it power-wise comprehensively. But quality software does not reside within the hw, I thought you all would've learned this by now?

 

[fouad]

Well, the file I have was created in June - but I don't know if that really means anything as I don't know if that's reflective of the original creation date or not; that's just when I got a hold of it.

Anyway if people have questions about this doc go ahead and PM me. Though ironically this thread already had it's primary question addressed a couple of pages ago (that the reason some final kits have 2.8 GHz chips is due to a mix-up), I feel like there's better things to discuss here than the listed clockspeed in that doc.

I for one would like to know the clock multiplier of the XeCPU, simply to undertand better how many clock steppings off from the final the beta chips that went out were. (just for personal curiosity - not because it matters)

I think the file was a recent one, at least after E3 2005 for sure, and it talk about final xbox360 hardware...


But its weird why no one noticed the 3 GHZ frequency ?!!!! was it a typing error, or a marketing secret strategy ?

 

[Jawed]

I hardly think reading a CPU performance counter constitutes as "special software".

Maybe some people can't be bothered to write those 5 lines of code, but I know getting a new devkit that's one of the first things I'd take a look at.

Well M$ sent out software to help debug this, as far as I'm aware.

Jawed