Intel Nehalem Absolutely Insane -> 975 ^^

[saaya]

0_o I'm pretty sure Intel contracted extra-terrestrials to design this chip. 4ghz on Air is impressive, 5Ghz on air is revolutionary.

Perkam

you still dont got it right perkam
4ghz is pretty much normal, i havent seen a chip that cant run 4g on air, even old c0 chips 4.5 on water seems to be normal as well jcool, i havent seen anybody reaching less than that on water, some even reach it on good air, the newer c0 and the d0 chips need less voltages for it however.

starting at 4.6 there seems to be some weird barrier
that many chips cant surpass stably even with ln2 and huge vcore, must be some transistor limitation or a fundamental design limitation...

Guys since this is a ES 975 chip does this mean when the retail version comes out it will be the same or better?

worse or the same is more likely, intel has been binning press samples very very well for the past years

pcrinimal, yes, there is something very weird about cpu multipliers, especially with intel cpus... just ask KP and hipro5 about their 780ES chips... the diferences between the multipliers were insane, some clocked ridiculously low compared to others. im pretty sure it has something to do with internal timings and interface buffers that dont work well with certain multipliers. in the end, multipliers are pretty similar to memory dividers, and we all know what vast diferences memory dividers on chipsets had. multipliers seem to be a lot less weird in that regard but it still happens... i dont know if intel can adjust some timings or settings to improve this, but i think they can... but they obviously spend most if not all their time tweaking everything for the stock multipliers, and not higher multipliers...

[jcool]

you still dont got it right perkam
4ghz is pretty much normal, i havent seen a chip that cant run 4g on air, even old c0 chips 4.5 on water seems to be normal as well jcool, i havent seen anybody reaching less than that on water, some even reach it on good air, the newer c0 and the d0 chips need less voltages for it however.

I've had at least 5 920's that wouldn't do 4Ghz on water, no matter what. Because some CPUs don't like the 21x/20 multipliers, and they don't run high BCLK either

starting at 4.6 there seems to be some weird barrier
that many chips cant surpass stably even with ln2 and huge vcore, must be some transistor limitation or a fundamental design limitation...


worse or the same is more likely, intel has been binning press samples very very well for the past years

pcrinimal, yes, there is something very weird about cpu multipliers, especially with intel cpus... just ask KP and hipro5 about their 780ES chips... the diferences between the multipliers were insane, some clocked ridiculously low compared to others. im pretty sure it has something to do with internal timings and interface buffers that dont work well with certain multipliers. in the end, multipliers are pretty similar to memory dividers, and we all know what vast diferences memory dividers on chipsets had. multipliers seem to be a lot less weird in that regard but it still happens... i dont know if intel can adjust some timings or settings to improve this, but i think they can... but they obviously spend most if not all their time tweaking everything for the stock multipliers, and not higher multipliers...

I agree on that, my 975 doesn't seem to like more than ~1,53V Vcore for example
But it does run higher than 4,6

[saaya]

I've had at least 5 920's that wouldn't do 4Ghz on water, no matter what. Because some CPUs don't like the 21x/20 multipliers, and they don't run high BCLK either

I agree on that, my 975 doesn't seem to like more than ~1,53V Vcore for example
But it does run higher than 4,6

cpus not "liking" 20/21 multis is a board problem
its when the TDP TDC exceeds the programmed value, then the cpu will become unstable and crash, thats a board problem. on BloodRage you can completely cut the tdp and tdc reading for the cpu, dont know about other boards but id be surprised if they dont do it... i think they adjusted this in the more recent bios files though and at least manipulated the tdp tdc reading of the cpu so you can clock higher.

im sure if youd try those cpus again you could hit 4G easily

whats the max 3d stable bclock pcriminal?

[jcool]

cpus not "liking" 20/21 multis is a board problem
its when the TDP TDC exceeds the programmed value, then the cpu will become unstable and crash, thats a board problem. on BloodRage you can completely cut the tdp and tdc reading for the cpu, dont know about other boards but id be surprised if they dont do it... i think they adjusted this in the more recent bios files though and at least manipulated the tdp tdc reading of the cpu so you can clock higher.

im sure if youd try those cpus again you could hit 4G easily

Lol. Sorry, that's all I can say to that. I run a DFI UT on my testbed with "set current limit max" option enabled and TDP/TDC limits disabled.
Some 920 CPUs just don't run well using the 20/21x mults, nothing to do with the mobo. If you don't disable TDP and TDC limits, the CPU will come down from 21x to 20x, but that's it - doesn't influence the way a CPU handles the multipliers

[PcCI2iminal]

whats the max 3d stable bclock pcriminal?

Ive tested a bit and 228MHz is easy for boot and 3d benching

Welcome to the 5Ghz air club, very nice chip.

Now lap the chip and enjoy the extra 100Mhz or so

hehehe i cant cz its not mine ,i need to send it back soon

relase is about 25.5. i think or so...

31.5 i think

[carpo93]

31.5 i think

max bclk stable in 3d?