I would like to know how they measured that. There are a ton of people throwing out 200W on forums but none of it verified. Some companies measure power differently as well. I would like to know if their near 250W was actually measure, reported max TDP for the chip, other? In my new thread the published info is 100-170W. Not saying that is correct but I have seen a LOT of general writes throwing out numbers (I think IBM has been pretty coy about this AND the chip is only purchased as part of a server, further limiting access/insight. Hence why I would like to know how/where realworldtech got their info). That said I don't doubt that high end Power7 chips are MASSIVE POWER SUCKERS.
- Status
I would like to know how they measured that. There are a ton of people throwing out 200W on forums but none of it verified. Some companies measure power differently as well. I would like to know if their near 250W was actually measure, reported max TDP for the chip, other? In my new thread the published info is 100-170W. Not saying that is correct but I have seen a LOT of general writes throwing out numbers (I think IBM has been pretty coy about this AND the chip is only purchased as part of a server, further limiting access/insight. Hence why I would like to know how/where realworldtech got their info). That said I don't doubt that high end Power7 chips are MASSIVE POWER SUCKERS.
Ninjaprime
Regular
The chip in the Anandtech link is 3.3ghz and goes up to 170w... there is your proof. Those are server setups, you only see high clocked ones in HPC machines usually. If you dig around there is a IBM whitepaper with Power7 8 core at 3.55 with turbo up to 3.86, IIRC it says TDP is 200w. The 4ghz to 4.14 with turbo is 240w+. David Kanters info is always solid.
I am not saying David is wrong, but I don't think what Anandtech says indicates conclusively that the 3.3GHz model is 170W. Read it again:
The Xeon E7 is a product line up (Westmere-EX). For example the E7-4870, a 6 core 1.87GHz chip, has a TDP of 95W and the 8 core 2.67GHz E7-8837 a 130W. So when Anand says the E7's have a 95-130W TDP range that fits *exactly* with the product line up (check my link).
If that is the context Anand is using in that sentence there remains ambiguity as the POWER7 lineup runs from 4 core to 8 core variants ranging from 3.0GHz all the way up to 4.14GHz (Turbo). If Anand's information is correct (?) you would think a 3.3GHz 4 core chip would fall on the lower end of the 100W-170W range.
Put that into perspective: If a 3.3GHz 4 core chip is 170W how is a 4.14GHz 8 core chip only (less than) 250W? That is some amazing scaling if doubling the cores and jacking the frequency up 25% results in a 50% bump in TDP--that is amazing even. And not likely, so pardon my reservations. I don't necessarily doubt David's 4.14GHz / 8 cores / 240W, but that would indicate the 3.0GHz / 4 core models are running at the 100W low end TDP Anand provides.
As for those whitepapers I have not seen or found them. You may be right and I have been digging. But what I do know is that based on the following slides, page 14, a Power 755 4U has 8 chips (32 total cores, 4 cores a chip) at 3.3GHz with 256GB of memory has a total peak power draw of 1650W. While that may reconcile to 170W per chip it also could reconcile to a lower end as well.
Hey NinjaPrime I have moved the TDP discussion for the POWER7 to the POWER7 thread. I hope you don't mind as it seems more appropriate there than the Xbox 3 rumors thread at this point. Thanks!
Just found this in another forum searching for fresh rumors on google
Maybe the last rumor about 16 cores or 4x4threads was generated by this?
It's plausible? 80MB L3 edram is a lot 0_0 at this moment only ibm can do it, but really? 0_0
how big can be something like this at 32nm?
can this be only an unannounced server processor?
someone is able to decode the blue text on the bottom?
Maybe the last rumor about 16 cores or 4x4threads was generated by this?
It's plausible? 80MB L3 edram is a lot 0_0 at this moment only ibm can do it, but really? 0_0
how big can be something like this at 32nm?
can this be only an unannounced server processor?
someone is able to decode the blue text on the bottom?
Just found this in another forum searching for fresh rumors on google
Maybe the last rumor about 16 cores or 4x4threads was generated by this?
It's plausible? 80MB L3 edram is a lot 0_0 at this moment only ibm can do it, but really? 0_0
how big can be something like this at 32nm?
can this be only an unannounced server processor?
someone is able to decode the blue text on the bottom?
Accessible by cores means on-die?
The rumor is compatible with this news:
http://www.advancedsubstratenews.com/2012/01/gfs-ny-fab-8-debuts-with-ibms-32nm-soi/
If it's a fake, it's a really good one.
DuckThor Evil
Legend
You'd need some CSI tech to clear that blue text 
Interesting find. Where exactly did you found this?
Interesting find. Where exactly did you found this?
Assuming it's not a fake I would say no and that's indeed related to MSFT next system.
See tI first I was wondering about the 25.6GB/s low latency interconnect as why would you need low latency interconnect if not to connect other CPUs/processors. The bandwidth was not fitting for a GPU.
Then I looked better and notince that there are three blue part on the upper left corner of the chip that connect to the GPU. That is not server related as GPU are a critical part to say the least.
So it's indeed related to the Xbox if true. As it is we can bury the dual GPU rumors.
Basically this chip is to be connected to DDR3 the GPU will have 76.8GB/s to the 80MB of EDRAM (not enough to feed a monster GPU by the way).
There could be three GPUs but we all will agreed on the likelihood of that, even two sound unnecessarily complicated.
So now the question is what kind of GPU, I would put my bet on GCN 2 at this stage. The system is design call for a GPUs that plainly support a coherent memory space between the CPU and the GPU.
MSFT might want to use the low latency communication to offload a lot of calculation to the GPU. THey have backed up from throughput oriented cores, I would not be surprised though if SIMD have been widened.
Fake. Photoshoped from older picture.
http://blogs.gartner.com/andrew_white/files/2010/10/IBM-IOD-Information-Supply-Chain.JPG
http://blogs.gartner.com/andrew_white/files/2010/10/IBM-IOD-Information-Supply-Chain.JPG
That set the deal or the same guys were in the front row?
DuckThor Evil
Legend
People who create fake information are the lowest type of vermin... Would like to crush skull.
Yep it's really bothering but by looking at the amount of rumors around and the reactions to them I would not be surprised if somebody manages to leak something accurate soon.
By this summer after E3 we will know...
You have to admire their photochopping skills on this one though. The composite is superb.
maybe he needs some practice with grammar to be perfect
Can you understand if that floorplan is a real product or a mix? maybe a gpu and those are shader cores?
Well, slides are often in note form, similar to newspaper headlines and bylines. If every slide with such orthodox errors are fake, than I dare say half the presentations given in the world are fraudsters.maybe he needs some practice with grammar to be perfect
The actual lighting and blur and noise is very convincing. Let this be a warning for future slides we see. Perhaps GDC never happened, and all that photographic evidence is wool over our eyes?Megadrive1988
Veteran
zdnet picked up the story of Xbox3 having 16 cores (not threads).
http://www.zdnet.com/blog/home-theater/next-generation-xbox-durango-could-ship-with-16-core-cpu/5792
edit: scratch that, I didn't see the Durango slide above.
http://www.zdnet.com/blog/home-theater/next-generation-xbox-durango-could-ship-with-16-core-cpu/5792
edit: scratch that, I didn't see the Durango slide above.
RancidLunchmeat
Veteran
MS had a hands down winner this generation if only not for the RRoD disaster (and even that may have been a strategic move to ship flawed hardware just to be first to market, I don't know).
Without that, they'd be swimming in profits. MS knows that if they can afford to, the strategy of releasing loss-leading hardware is still very viable. Especially now and in the next generation when even more revenue streams will appear through digital distribution.
I would be shocked if MS releases their next console and it has the types of dumbed down specs that are now being discussed from Nintendo and Sony.
Without that, they'd be swimming in profits. MS knows that if they can afford to, the strategy of releasing loss-leading hardware is still very viable. Especially now and in the next generation when even more revenue streams will appear through digital distribution.
I would be shocked if MS releases their next console and it has the types of dumbed down specs that are now being discussed from Nintendo and Sony.
Let's assume MS wants ti sell HW they break even on at $299 retail. How much more could MS fit than an A8-3850 and 7670 GPU? I am not saying I want the dumpy HW being discussed for a potential 2013-2020 console, but given the above criteria there isn't a lot of wiggle room, especially if TDP is emphasized.
One thing we hear a lot is how consoles tend to OVER engineer for past mistakes. MS tossed a ton of fillrate bandwidth into the 360 because the Xbox lacked there. The GCN had really low latency memory because the N64 had really poor latency. Could not a response to the RRoD being to be stricter on TDP (which ironically if a bump-gate like issue is mostly irrelevant to the final meltdown as it is the expansion between cooling/heating that kills it).
- Status
Similar threads
- Sticky