cubes158@alien.topBtoHardware•PC Gamer: If RDNA 5 can't turn things around I have a tough time believing AMD will stick around in PC graphics for much longerEnglish
1·
1 year agoAMD is going to be driven by financials.
In simple terms, if Profit > Costs then AMD will stick around regardless of market share or competitors performance.
These guys will even run at a loss if there is a light at the end of the tunnel.
What are the financial numbers? AMDs participation in the market is not driven by market share or hardware surveys. It’s driven by how much value the division returns to the company.
As far as I can tell, It might be working but I don’t think it’s working fully or as intended.
Here is testing I just did on a 13900k;
When I run CS2, the process is detected and there is a nice graphic in Core Director that shows core utilization, and it has cores marked which are e-cores.
I just did a test with CS2 launch commands "-threads #’ where # is the number of threads you want to cap CS2 at.
When I use ‘-threads 8’ I can see that CS2 utilizes 8 p-cores with a p-core usage at 50% and e-core usage 0 or negligible. Seems to be working great - but I’m not sure that CoreDirector is doing this scheduling.
When I use ‘-threads 16’ I can see that CS2 utilizes many e-cores and p-core utilization drops to 40-45%
When I remove the command entirely, I can see that CS2 activates all of the e-cores.
tl;dr - As I increase CS2 thread count, the process clearly increases e-core usage. It seems CoreDirector is failing to keep something in the process off of the e-cores.