From a CB10 run a while back--ignore voltage (it's actually 1.225V):
http://www.pcrpg.org/pics/computer/o...b10x64-001.png
Printable View
From a CB10 run a while back--ignore voltage (it's actually 1.225V):
http://www.pcrpg.org/pics/computer/o...b10x64-001.png
yeah but that dosen't say anything about what the cpu consume. Let alone what a theoretical 4core westmere cpu would. :p:
Can't compare that, since they used cinebench 32bit. 64bit gains quite a few hundret point more then the 32 bit version.
CanardPC says 128.2W for 3.2 GHz, a good source ;)
I doubt that since, we all know the first thing i7 does is "turbo throttleing" when the TDP limit is reached. If the stock 3.2 ghz version would use 128.2W (the 0.2 W really cracks me up :rofl:), it would never use turbo mode. But a 965 or even better a 975 uses turbo on all cores even on the stock cooler.
They even mention that in there articel and they still don't question there own numbers they have come up, while knowing that....
What's the threshold for throttling on the current i7s anyway?
What's the TDP threshold?
The temperature threshold is about 80C I think.
Nobody but intel knows for certain.
We know the family TDP is 130W so its a good indicator for what the TDP ceiling is, we can also guess the thermal throttel point, due to observation.
Im not sure if its 80°C, i guess its 100°C but with 80°C for the PCU to disables turbo mode and after that he starts with various energy conservation techniques.
The 920 starts to throttle on my DFI JR when it hits around 85C and current/TDP limits are not disabled in bios. Not sure about the TDP barrier tho.. I'd guess around 150W.
Really looking forward to replacing all my single and dual socket Nehalems with Gulftowns, 50% more crunching-power :up:
It's not a matter of trust, it just don't adds up with the observations and data we have here in the forum. If Ci7 would nearly run near the brink of its TDP at stock, there would be no turbo mode.... but the fact is, it runs turbo even on stock cooling.
The whole turbo throttling issue for some boards shaded some light on this. Depending on the cooling people got different states where the Ci7 began to use turbo throttling, but that was way above stock frequency (3.8-4ghz area) and then it was often enough to reduce the voltage to nearly stop the turbo throttling.
Later on the issue was fixed with new bios from various board makers.
Oh heh, a new benchie - I'll def. check it out sometime :)
That is quite a dissapointment :( A case of software catch up again? Yawn.
Makes me happy I haven't moved from 65nm. Never thought I'd say that 2 years later. Waiting for the next q6600 to rear its head. I guess 8 core 22nm?
I'm not at all surprised about real world results. i7 didn't even buy that much for most people and this isn't that much of a real improvement for 99% of people.
I hope Intel can get the clock vs clock speeds up a LOT, at least compared to 775 socket chips to give people a legit reason to upgrade to "next gen."
You will never see such big jumps in performance per clock 20-50% over a broad range, more like the stady increasment of 0-10% each new cpu generation with a declining tendency.
The thing to blame for that is x86, its already hard to keep a 3 issue architecture filled (amd) and got even harder when intel introduced C2 (4 issue).
Thats why you saw the return of HT, to keep the execution engies field.
X86 needs to die if you want to see huge gains in IPC or performance per clock again.
TDP does not determine the threshold, the power output of actual consumption and the efficiency of the cooling solution determines load operating temperature and TM1 and TM2 are programmed to throttle if that temperature threshold is exceeded.
Not sure about the core iX series, but 105 deg was Tjmax for some of the prior products.
Bulldozer isn't mentioned in AMDs 2011 roadmap anymore. It was expectable since AMD had to adopt Intel AVX extension. BTW Q4-10/Q1-11 is a time of Sandy Bridge intro which was taped out 2 months ago, probably because of longer debug/test cycle of new gen + graphics.