I just built a new rig with ASUS P8Z77-V LE PLUS and an Intel Core i7 3770k 3.5GHz cooled by a refurbished Corsair H100. (I was eager to build it this weekend, and this is what Fry's had in stock ).
After upgrading bios version 0404 to the latest version 0606, I went into Advanced Bios settings and set up a 4.1GHz overclock. When rebooting, the bios post screen still indicates that CPU speed is 3.5GHz, as does Windows 7 System Properties. However, CPUID CPU-Z on the CPU tab displays "Clocks (Core #0): Core Speed 4100.00 MHz.
Unfortunately, my video card won't arrive until next week, and I'm unable to do reliable testing with the Intel onboard video. I ran the Crysis Demo benchmark before and after OC and got a 0.5 to 1.0 FPS improvement @ 4.1GHz. This doesn't tell me much since the (lack of) GPU clearly is the bottleneck: it basically was a 10-16 FPS slideshow.
How can I be sure that OC really is happening? Also, why don't the bios post screen and Windows System properties display the actual overclocked speed that CPU-Z was able to detect? With every other motherboard I've used, including from ASUS, the overclocked speed is pulled by the system and correctly displayed in both locations. This particular motherboard has more Ai Tuner controls than I'm used to from previous ASUS mobos, so perhaps it's a case of RTFM.
Thanks for the reply. I decided to make this more objective by installing povwin-3.7.RC6 and running the 'Render -> Benchmark (All CPUs)' option. POV-Ray is almost completely CPU bound, so this makes for a good torture test. The [comments] are my notes, everything else is pasted directly from POV-Ray. I ran CPU-Z to verify CPU and memory clock speed prior to each test.
[Factory Optimized Defaults via BIOS: 3.9GHz + 1333GHz DDR3]
CPU time used: kernel 0.20 seconds, user 1514.50 seconds, total 1514.71 seconds.
Elapsed time 195.86 seconds, CPU vs elapsed time ratio 7.73.
Render averaged 1338.44 PPS (173.07 PPS CPU time) over 262144 pixels.
[Manual OC set via BIOS: 4.2GHz + 1866GHz DDR3]
CPU time used: kernel 0.08 seconds, user 1324.73 seconds, total 1324.81 seconds.
Elapsed time 170.68 seconds, CPU vs elapsed time ratio 7.76.
Render averaged 1535.88 PPS (197.87 PPS CPU time) over 262144 pixels.
[Manual OC set via BIOS: 4.5GHz + 1866GHz DDR3]
CPU time used: kernel 0.05 seconds, user 1246.04 seconds, total 1246.09 seconds.
Elapsed time 161.44 seconds, CPU vs elapsed time ratio 7.72.
Render averaged 1623.75 PPS (210.37 PPS CPU time) over 262144 pixels.
[Results from 3.9 to 4.2]
07.7% OC 3.9 => 4.2
14.3% Improvement on PPS CPU time
12.5% Improvement on elapsed time
[Results from 3.9 to 4.5]
15.4% OC 3.9 => 4.5
17.6% Improvement on PPS CPU time
17.7% Improvement on elapsed time
Bus Speed @ 100MHz. (Possible Bus Speed settings 80MHz - 300MHz, default 100MHz.)
It looks to me as if OC is making a real difference. Also, the fans on my H100 liquid cooler never went turbo, so perhaps I have room for a higher OC.
Thanks for that observation. I took your advice and bumped up OC to 4.5GHz. I've edited the earlier post with updated POV-Ray results.
This raises an interesting question. Am I cheating myself out of performance by leaving the Bus Speed at the default of 100MHz? Also, should I be disabling all Intel Power Management and Turbo options?