Interesting findings, Core usage & different affinities with different Mouse Polling rates (125,500,1000hz)

Otto_iii · October 20, 2019

Recently i built a R5 3600 system, first build i've done by myself, very happy with it. One thing i had to deal with coming from my old Intel i5-3570 was significant CPU bottlenecking, even with a 1050ti in most new games, so to avoid hitching during rapid gameplay with me throwing the mouse around id often turn down my polling rate of my Zowie mouse from 1000 to 500hz as i noticed 500hz was about a 5-6% peak usage increase, and 1000hz was about 9-11%, at the time i needed all CPU power i could get but still wanted the latency reduction.

So recently i revisited this with the 3600, not because i presumed it really hammered the cores, obviously it doesn't as i could bearly see a increase in Task Manager (125hz and 500hz report 1% usage increase, 1000hz is 2% reported usage, nothing note worthy), but felt like it would be nice to see exactly what is going on.

Here is where things got odd, using Ryzen master i found that with Windows 10 open all cores remain active running at 165-225 or so, atleast for me this is normal.

Spoiler

Pretty normal, with 125hz, the 2nd Core goes to sleep, basically looks similar to idle, very mild increase core 3 and 4

Spoiler

With 1000hz 3 Cores go to sleep and on my PC Core 5(ccx1) and 3(ccx0) ramp up to around 500

Spoiler

Now here is the part that had me scratching my head, with 500hz and 500hz only, literally the moment i start making aggressive mouse movements at 500hz all the other cores beside core 3 go to sleep, Core 3 getting used moderately. For whatever reason it seems to be the only polling rate that is bound to a single core, mind you the 2nd best core on the lesser of the two ccx in my case, but whatever, its consistent.

Spoiler

I bring this up in part because there has been discussion for a long time that 500hz is arguably the sweetspot, with people claiming to still prefer it even if 1000hz should be superior. They are close enough that serious mouse reviewers like RocketJumpNinja are still claiming 500hz limited mice like the FinalMouse series are or can be the best they've ever used (i don't say this as a finalmouse fan, i dislike the company.) So assuming that this behavior is similar on most PCs, with input from a mouse polling at 500hz locking itself to a single core, i do wonder if that could effect some level of consistency during gameplay that 1000hz doesn't offer, or atleast could of in the past when certain CPUs were more bottlenecked by games.

I have no clue whats actually really going on here, just thought it was interesting so if anybody understands the computer science better maybe they could comment on it.

Tan3l6 · October 20, 2019

I have an interesting find/bug too - when polling rate is 1000Hz and I try to move GPU-Z window it moves like on rubber band - moves slower than mouse cursor and takes some time to reach the place mouse actually moved.

500Hz - no rubber banding.

Also GPU-Z window is the only program window that rubberbands.

Might not really be related to OP story/problem, apologies.

Otto_iii · October 20, 2019

4 hours ago, Tan3l6 said:

I have an interesting find/bug too - when polling rate is 1000Hz and I try to move GPU-Z window it moves like on rubber band - moves slower than mouse cursor and takes some time to reach the place mouse actually moved.

500Hz - no rubber banding.

Also GPU-Z window is the only progrm window that rubberbands.

Might not really be related to OP story/problem, apologies.

Definitely different then mine, but its interesting none the less.