Quote from: nizzuu on July 04, 2017, 07:46:38 AM

Quote from: dbc23 on July 03, 2017, 08:49:50 PM

I'm seeing about 600h/sec on an i7-7700

Quote from: Elder III on July 04, 2017, 12:21:21 AM

I get 580-630 on an i7 6700 at stock clocks (without locked pages enabled

That's pretty slow. On i5-4570 (all stock, 16Gb ram @ 2 channels, win10, locked pages enabled) it's ~745h/s (without any other load), or ~748h/s after reboot and miner's clean start. Drops to ~550 with regualar web browsing though.

http://www.pixic.ru/i/A00104s111l34914_preview.png

How to?

1) Use aeon-stack or xmrig
2) Enable lock pages (if not, may give up to ~30% hasharte drop depending on system)
3) Threads/cores/cache count:

2Mb L3, 2 cores (celerons etc.) = 2 single threads
3Mb L3, 2 cores, 2 threads (pentiums) = 2 threads, one double, one single
3Mb L3, 2 cores, 4 threads (modern pentiums) = 3 single threads, affinity 7 (--cpu-affinity 7 using miner key, or just using cmd: START /affinity 7 cmd /C "xmr-stak-cpu-notls.exe"). Or, one double + one single - about 3-5% slower, but less drop under system load.
4Mb L3, 2 cores, 2 threads (i3s) = 2 double threads
4Mb L3, 2 cores, 4 threads (i3s) = 4 single threads (or two double)
6Mb L3, 4 cores, 4 threads (i5s) = 3 double threads, affinity 7
8Mb L3, 4 cores, 4 or 8 threads (i7s) = 4 double threads

If you miner cannot use both single and double threads at the same time, launch two individual instances.

Lock pages in linux is necessary too?