• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Is a 5900X/5950X still a bottleneck for a 3090?

Soldato
Joined
26 Sep 2013
Posts
10,755
Location
West End, Southampton
Hi guys.

Reason I ask is, In so many games I'm seeing my 3090 not breaking a sweat but I've still got lots of headroom before hitting 120fps/144fps cap I've set. I very rarely see GPU usage go above 70% in the games I play yet looking at my fps overlay I'm never locked at 120 or 144 whatever I chose to set (when I stream and record I lock at 120 as it makes it smoother than 144. The thing I don't understand is my CPU usage, average across all cores will be like <20%

I've two rigs one with a 5900x and the other with a 5950x but moving the GPU to either system to test is the same. I game at 1440p, both CPUs hitting 4950mhz with the 5900x hitting a bit higher. Can you make me understand why this happens?
 
But how does that happen if my individual cores aren't being pegged at 100%, actually nowhere near that. I'm just trying to understand it.
There must be some other bottleneck in the CPU's architecture that is preventing the cores being used 100% (just my speculation).
 
Driving a 3090 to 100% usage at 1440p needs - besides a very fast CPU (which you kinda have), a high end / tuned ddr kit. What's your ram, and have you done any tuning with it?
 
But how does that happen if my individual cores aren't being pegged at 100%, actually nowhere near that. I'm just trying to understand it.
A common 'issue' with CPUs with many threads is that the default timing resolution on the monitoring software is too coarse. Try changing it to refresh much more frequently, e.g. if it's only updating cpu usage every 1s try changing it to 300ms. What you may then find is the peak CPU usage per core is a lot higher. The problem is that normally he cpu is constantly moving work between cores and hence the average load over a 1000ms period is significantly below 100% but over 300ms it might be a lot closer.

I had similar observations when I had a 3090 and it was only making afterburner refresh the utilisation stats much more rapidly that I realised single cores were in fact getting maxed out. By default I'd see only like 75% utilisation on a core because it would actually be (made up example) running at 100% for half a second and then 50% for another half a second. So it was in fact a bottleneck.
 
Interesting @everyone.

So a better upgrade would be to upgrade to a next gen CPU and platform rather than a 40 series GPU which will obviously exaggerate this CPU bottleneck even more. At times it feels like my 3090 is only being used at 50% of what it's capable of.
 
Driving a 3090 to 100% usage at 1440p needs - besides a very fast CPU (which you kinda have), a high end / tuned ddr kit. What's your ram, and have you done any tuning with it?
Sorry for the late reply, long working hours and a 5 month old baby lol.

I'm using Corsair Vengeance 3600mhz DDR4 32gb (2x16) overclocked to 3800mhz using just default timings.

@stooeh i play a lot of racing sims, Iracing on triple 1440p setup (7680x1440p), Assetto Corsa, Assetto Corsa Competizione. Then just pretty much anything and everything really, bit of Borderlands 3, Rust, 7days 2 Die, Vampire Rising, Valheim.
 
Last edited:
Interesting @everyone.

So a better upgrade would be to upgrade to a next gen CPU and platform rather than a 40 series GPU which will obviously exaggerate this CPU bottleneck even more. At times it feels like my 3090 is only being used at 50% of what it's capable of.
3090 is a good card so I think you would struggle to get value from a 40 series upgrade.
Best upgrade would be a 4k screen which will hit the GPU harder.
 
Back
Top Bottom