June 7, 20205 yr I'm trying to learn memory OC, and I'm using maxxmem to test the effect of different timings, but I'd like to corroborate my results with a "Real world" benchmark, and 32m takes forever to run. What are some other 2d benches that are highly sensitive to memory? Preferably with consistent results so I don't have to run it 20 times to get through the variance. I've heard geekbench is good?
June 8, 20205 yr Author 1 hour ago, TerraRaptor said: HyperPi, shorter variants like 8m/16m are rather good if you have anough threads I'll give it a shot thanks
June 8, 20205 yr Crew Geekbenches are a good fast test too, hence why memory sold here usually has a geekbench screenshot with the sale post
June 9, 20205 yr Author 23 hours ago, Leeghoofd said: Geekbenches are a good fast test too, hence why memory sold here usually has a geekbench screenshot with the sale post Do you need win7 for geekbench? I tried HyperPi, thought it would be great to be able to test 32m 4 times as fast But it seems unstable. Out of round at stock clocks for CPU and memory. And I had done 32m. Worth mentioning the CPU is a 2500k that I just got and haven't tested at all so there's an outside chance it might have problems. tried 1.2 vcore and up to 1.9 vdimm, 1.15 vccio. Not much else to adjust. The board is an EVGA z77 stinger that I bought locally as an impulse purchase because it was cute, but the bios is atrocious, doesn't have XMP or PLL overvoltage, and not all the memory timings are there, and some of them have too high minimum values. All that said, it's probably user error, I'll try more tomorrow.
June 9, 20205 yr 11 minutes ago, nnimrod said: But it seems unstable. Out of round at stock clocks for CPU and memory. It won't be 4 times faster, it will launch 4 copies of 32m in parallel, 1 for each cpu core. Make sure you have enough disk space in hyperpi folder location - i would estimate it as 300-400 mb for each thread, so free space should be 1200-1600 mb.
June 9, 20205 yr Author 1 hour ago, TerraRaptor said: It won't be 4 times faster, it will launch 4 copies of 32m in parallel, 1 for each cpu core. Make sure you have enough disk space in hyperpi folder location - i would estimate it as 300-400 mb for each thread, so free space should be 1200-1600 mb. Yes, by 4x faster I meant I could essentially get through a statistically meaningful number of 32m runs 4x as fast, or that's the hope anyway. Of course the times won't be good but they should be useful for comparing the effect of timing changes 1 hour ago, Leeghoofd said: What mem speeds are you testing? 666 and 800 straps, Making some graphs to see how performance varies with tRAS. I'll go higher just started at the bottom.
June 9, 20205 yr Author 4 hours ago, Leeghoofd said: life starts at 1866Mhz and 2133 if the IMC can handle it ? Haha yes, but I'm somewhat unusual and I just want to map out how all the timings affect performance. I like to do things methodically and exhaustively. Any suggestions on a better p67/z68 board? One with a full featured bios like the bios on my giga z97 soc-f or asrock z97 oc-f. And is z77 ok for efficiency with sandy? Thanks for your help
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.