CPU test

Kleaf

Good day! I have i7 6700HQ processor, 16gb DDR4 2133. my cpu test is 300+. at the same time, I do not lag in games, and the ZLO test for replays shows the results as ~cpu 200.
I even found a possible reason. This processor has 4 cores, but due to Intel Hyper-Threading Technology, these 4 cores are divided into 8 logical cores. I tried to disable 7 cores for the game and ran the test, the result did not change. so the test uses only half of one core.
I wanted to disable this technology through the bios to check and improve the cpu test, but unfortunately the advanced bios settings are locked and I can not access them.
I really ask you to fix the cpu test, because I get kicked from most lobbies.
sorry for using google translate

Jip

We're working on it but it is taking quite a bit of time. For your idea: instead of using an artificial benchmark a small scenario will be run instead (outside of the lobby) that looks similar to the test of ZLO. The time it takes for your computer to finish the scenario is your score. I'm not sure when this will be finished.

Kleaf

thanks, looking forward to it

ZLO

he means that he ran a replay test from that cpu-benchmark topic https://forum.faforever.com/topic/328/cpu-performance-tests

and got 21:05 time.
idk his CPU clocks and memory timings tho, and latency...
so he has i7 6700HQ
and i have Xeon E5 1650 (aka Intel Core i7-3930K it is sandy bridge-second gen 32nm)
i checked my CPU score again, it showed 187, ran test again and it now shows 194-183.

i would like to know in what way CPU test was changed, but i am to lazy to read code or github, if something was hardcoded to make old CPUs have less score, well then that is bad... if not, and test is just a load task then i don't know... and not sure why my CPU keeps good score while all other sandy bridge seem to be loosing score... maybe something to do with my quad channel memory... but i don't think quad channel memory is beneficial for ingame performance, even if it is ddr3 quad channel, i have heared latency is important and amount of channels does not improve latency...

Edit: also my cpu is at 3.5 ghz and 1328 ddr3 with 9-9-9-24 timings (huh i thought it was 1600, not some weird 1328)

TLDR: similar performance - different CPU score

Jip

Table management was included, in particular the use of large tables.

Kleaf

did you change something in the test? now the results are 350+ :(((( repeated the ZLO test - 21:25

Jip

Nothing changed the last few months.

Kleaf

This post is deleted!