Whoops: XB360 at 2.8GHz

[Carl B]

Well, normally the chip's clock multiplier is set by the manufacturer; for instance my AMD 3400+ has a clock multiplier of '12.' This then gets multiplied by the frontside bus frequency to determine the chip's operational frequency.

In my case, the 200 MHz associated with HyperTransport gives me a 2.4 GHz chip. I've overclocked it ever so subtley via the FSB (200--->208) to give me ~2500 MHz.

Anyway, so this is the reason I think that this is so - well, surprising. I don't remember what the 360's root frontside bus frequency is, but it seems like the difference between someone at MS (IBM, wherever) providing chips with the clock multiplier set at 7 vs 8; NOT something that just happens on it's own.

Now, the only reason I could imagine this is if legitimately someone screwed up; maybe indeed it was a bunch of lower clocked non-final chips that made their way into the final kits.

Either that or due to yield issues a number of chips weren't reaching 3.2 GHz at the designated voltage, got clocked down, and pressed into service anyway to meet demand.

 

[Gholbine]

Ouch...

We got a source on this? If they were forced to drop the clock speed of the final Xenon, that would be disasterous...

 

[Sis]

1) the title of this thread is misleading to the point of being a troll.
2) Where's this info coming from? And more importantly, how far spread is the problem, if in fact the source is accurate?
3) Is there really a concern with the clock being off by roughly 15%? Can't you just assume things in the real version will be 15% faster? It would seem having the actual hardware is more imporant than losing a month waiting for hardware that seems to be different only in clockspeed.

.Sis

 

[Li Mu Bai]

Why all the shock & non-sensical hypothetical possibilities? (dev. kit chips sneaking into finalized hw?) If this is indeed valid, it was a clock downgrade plain & simple. It happened this gen to both the GC & the XBX iirc.

Either that or due to yield issues a number of chips weren't reaching 3.2 GHz at the designated voltage, got clocked down, and pressed into service anyway to meet demand.

This seems to be the most logical answer, if what is rumored is true. November is right round the corner & demand appears to be exceedingly high.

 

[MrWibble]

3) Is there really a concern with the clock being off by roughly 15%? Can't you just assume things in the real version will be 15% faster? It would seem having the actual hardware is more imporant than losing a month waiting for hardware that seems to be different only in clockspeed.
.Sis

In answer to this question - yes, as a developer I'd be concerned if my "final" hardware was not at the correct clock speed. It is very hard to assume any kind of fixed improvement between two differently clocked machines where the clock actually only affects one particular component (or set of them). Other parts of the machine are almost certainly on seperate clocks, and it's the interactions between these components where a certain amount of stalling will occur. Without the final clocks it's very hard to know how that will change with any specific alteration.

For example, doubling CPU clock while leaving the memory clock alone would double performance of tightly written code but have almost no affect on something memory bound.

My general approach is that I target whatever bit of hardware I've got on my desk regardless of promises being made - that way I'm unlikely to have a last minute scramble for more performance if things don't pan out the way the hardware folks would like. There's usually stuff I've had to take out for speed or scale back - so finding things to put in if the machine does suddenly get faster is rarely a problem.

Usually the clock speed on a modern machine would be alterable in software - either BIOS/firmware or even a system call. I could maybe understand/explain this issue if MS accidentally flashed a bunch of kits to the wrong version, maybe while testing them. I find it harder to believe they *accidentally* put the wrong bit of hardware in the kit.

However if it's not easily fixable (software would be a reflash at worst, surely?) then that would suggest something more serious, which intrigues me...

 

[MoeStooge]

Why all the shock & non-sensical hypothetical possibilities? (dev. kit chips sneaking into finalized hw?) If this is indeed valid, it was a clock downgrade plain & simple. It happened this gen to both the GC & the XBX iirc.



This seems to be the most logical answer, if what is rumored is true. November is right round the corner & demand appears to be exceedingly high.

Can everyone re-read the original post?

IT SAYS NOTHING ABOUT FINAL CONSUMER XBOX 360 HARDWARE.

Not only that, but it says nothing about the number of development kits affected. And it specifically states that they are going back.

Let's all grab our reading glasses and put away our Jump-to-conclusions mats.

 

[Li Mu Bai]

Can everyone re-read the original post?

IT SAYS NOTHING ABOUT FINAL CONSUMER XBOX 360 HARDWARE.

Not only that, but it says nothing about the number of development kits affected. And it specifically states that they are going back.

Let's all grab our reading glasses and put away our Jump-to-conclusions mats.

Understandable MS, but if finalized dev-kits aren't in the hands of studios by now........

 

[Shifty Geezer]

The OP mentions SOME of the final kits were downclocked. Not all. And these SOME kits are being returned and replaced with 3.2 GHz kits. There's absolutely no reason to see this as a downclocking of final hardware (which the thread title implies )

 

[Jawed]

My apologies for the "misleading" thread title. Not sure how to edit it. Just editing the first message's title doesn't work.

Jawed

 

[ShootMyMonkey]

It's basically an error in some of the kits. Most all the parts themselves are actually binned at 3.2 GHz, but they're still locked at 2.8 due to some settings not being changed. Apparently this happened for several reasons in several different cases, but among those cases include the fact that a lot of beta kits were returned because of defects, so some of them are now having final hardware installed and the black dunce cap replaced with a gray one to mark them as final (i.e. multiplier settings left alone or something to that effect)

MS still says the final hardware SHOULD be at 3.2, even though a fair fraction of them went out at 2.8. As long as the same mistake isn't made with the retail units, which I really don't expect (the retail units probably go through a separate production line).

 

[Carl B]

Well, then I guess it's the 'Beta Kit as Final' theory that is the correct one here. By the way, since the knowledge is lost to me, what's the clock of these chips to begin with? Or I guess just as useful, what's the FSB frequency for the chips? Is it a 7 vs 8 / 400 MHz clocking situation, or a 3.5 vs 4 / 800 MHz situation - that sort of thing.

 

[Jawed]

Glad to see an independent source sign up to this

Jawed

 

[standing ovation]

Maybe the dev kits are slower on purpose? Just think. If a publisher can get its projects running smoothly here, then everybody can rest assured it will be as smooth as silk on the end-user platform.

Besides, there are WAY too many über-geeks at IBM and Microsoft for this kind of "mistake" to have gotten past both tech giants.

 

[Carl B]

Maybe the dev kits are slower on purpose? Just think. If a publisher can get its projects running smoothly here, then everybody can rest assured it will be as smooth as silk on the end-user platform.

Besides, there are WAY too many über-geeks at IBM and Microsoft for this kind of "mistake" to have gotten past both tech giants.

I hope you're not serious here.

 

[fouad]

The CPU of XBOX360 will be clocked at 3.00 GHZ, I read this on a paper under NDA for xbox360 game developers only given by a friend here on the forum ( thanks for you ), this paper explained the differences between first DEV KITS and final DEV KITS and hardware of xbox360. PERIOD

 

[scificube]

3.0GHz? I don't believe it. MS would be nuts to lower the clocks willingly and it's hard to believe they would be forced into this situation with fab problems. IBM is fabbing now right? So why can they get Cell up to 3.2GHz and not MS's chip?

If this is true I'll need constant treatment for my eyes-rolling-back-in-my-head-syndrome.

 

[fouad]

3.0GHz? I don't believe it. MS would be nuts to lower the clocks willingly and it's hard to believe they would be forced into this situation with fab problems. IBM is fabbing now right? So why can they get Cell up to 3.2GHz and not MS's chip?

If this is true I'll need constant treatment for my eyes-rolling-back-in-my-head-syndrome.

Believe it or not, I have said this for some months now, and no one believed me, but everyone will know the truth when xbox360 will ship. the paper under NDA say everything...some parts of the GPU also were under powered, some other were improved, same thing for the CPU...
PS : maybe for marketing reasons MS dont want to tell gamers that their CPU is 3 GHZ and the CELL ps3 cpu is 3.2 GHZ...maybe...imagine the impact on gamers ?

 

[Titanio]

It this the same PM/doc that was floating around earlier regarding the differences between the G5s in the alpha kits and the Power cores in the Betas? Although I didn't receive that PM/doc, no one ever mentioned at the time anything about CPU clocks, and I figure that would have been the bigger news to mention..

 

[hadareud]

Believe it or not, I have said this for some months now, and no one believed me, but everyone will know the truth when xbox360 will ship. the paper under NDA say everything...some parts of the GPU also were under powered, some other were improved, same thing for the CPU...

It does sound hard to believe, especially since it is getting very close to launch indeed. But we'll see I guess

 

[scificube]

I'm contacting a specialist ahead of time since I know I am prone to rolling eyes attacks...

Not at you of course but at MS if they weakened the CPU from this point of view without some good improvement elsewhere.

To be honest what you claim about Xenos is both scarier and possibly more excitng...stop playing with my head you! Just because you can doesn't make it the right thing to do.

...rolls-eyes...oops...that one got away from me...