IDF 08 SF: Intel Reveals More About Nehalem

[Rob Williams]

Intel opened up about their Nehalem architecture at this summer's IDF, and while we are unable to talk about performance data, what we can talk about is of great interest. Read on as we take a brief look at Turbo Mode, HyperThreading, CPU and Memory overclocking, the triple-channel memory controller and much more.

You can read the full look here and discuss it here!

[Kougar]

That wasn't your hand in the photos was it?

Highly overclocked current-gen processors with the fastest memory on the market will not even reach what Nehalem is capable of... even with DDR3-1066 memory. So while the overall memory frequency is low (to what we are now used to), the intense performance gains negate any downside.

Hm this seems directed at someone... Partly what I was getting at is if we can find a way of mitigating this huge performance hit due to CAS 7 or CAS 8 1066MHz memory, then the performance should jump noticeably, even significantly in some scenarios.

It may just be a pet peeve of mine, but I can't stand not fully optimizing a system regardless of how fast it is, if I know a little tweaking here or changes there can make it faster still for almost nothing. I'll be happy once I can see some 1333Mhz CAS 5 or CAS 6 results compared to 1066MHz CAS 7, which is already "tight" since most kits are CAS 8 and CAS 9 at these speeds.

Depending on certain factors (none of which we can state until we actually know Nehalem's limitations), all current DDR3 memory should work fine, as long as the rest of the system allows it, such as the QPI, or whatever limitation we might run into. DDR3-2000 has been found functional on a Nehalem rig, but it leads to the new quandary of whether or not it's even needed.

Very good to hear! And that is extremely impressive that despite being integrated into the CPU AND that there are now three of them, that Intel's memory controllers had no issues handling 2000Mhz DDR3 kit. Or at least I will assume the no issues part, I think 2200MHz is the current maximum for DDR3.

Today's applications can't even seem to make use of available bandwidth seen on the high-end memory kits available today, so will Nehalem essentially kill what is known as high-end RAM?

Despite DDR2 memory clocking anywhere from 1066Mhz up to 1200MHz showing no tangible performance gains outside of synthetic memory bandwidth tests (compared to DDR2-800Mhz CAS 4), almost every typical enthusiast jumped on 1066MHz PC2-8500 kit or better. I quite agree with you that such high-end DDR3 will be equally unnecessary, but in my view I don't think that is going to stop any of the same enthusiasts from buying it regardless whether to brag, have a "good looking" spec sheet, to play with, or whatever their reasons are.

Last I heard DDR3 hit something around 2200MHz stable, I bet that a good many extreme clockers are going to be dieing to break 2400MHz on a Nehalem rig, and that it will probably happen before the end of this year if Nehalem's IMCs truly do not have any limitation with memory frequencies, or at least are par with their chipset IMCs.

I'll be happy with some ultra-low latency DDR3-1333 or 1600 kit though... amazing but I don't think I've seen any yet.