Hello, I just finished building a new pc, it posts and can handle some games well however on other games it performs very poorly. For instance, when I run red dead redemption 2 on ultra settings 1080p my CPU usage is at 30% and I get ~100 fps (RTX 3080 TI GPU), but when I run cyberpunk on medium settings at 1080p my CPU usage is at 95% and I get 30 fps.
I then tried running 3Dmark (TimeSpy) I got a good score for the graphics (18431) but I got a very low CPU score: 2390. I also tried running Cinebench and got a multi core score of 986 (I should be getting 18k score). I also had Ryzen master open when running Cinebench and found that my CPU power was at 50W while running the benchmark.
I have tried enabling XMP, updating my chipset, flashing my motherboard BIOS and updating CPU drivers, I have installed latest CPU and GPU graphics drivers but still get these results.
Below are my specs:
MSI Pro B650M-A Wifi Motherboard
T-FORCE Vulcan DDR5 6000MHz CL30 (2*16GB) Ram
Ryzen 7 7800x3D CPU
EVGA XC3 RTX 3080 Ti
Cooler Master Hyper 212 Black Edition CPU cooler
Corsair RM850x PSU (Refurbished)
Below are screenshots of benchmark results and Ryzen master overview:
TimeSpy 3Dmark:
https://imgur.com/a/S5TnuSN
CineBench multicore:
https://imgur.com/a/gKhuWPn
Ryzen Master output during Cinebench:
https://imgur.com/a/Udc0Yo1
So, I wouldn’t just jump to CB r24 just yet, as the vast majority of comparison data will be with r23 still.
But, it’s also the only benchmark in your case that “matters” with regards to CPU. And I didn’t really see anything on your r24 score that made me go, “whoa, that’s wrong”. Seeing that some with the same GPU are getting over 1000, sure, you might be able to say you’re “low”, but perhaps well within “range” of typical (?)
So it looks like you’re right, I was comparing my r24 score with ones on r23. I just ran a CB r23 multi core benchmark and got 16107 which is much better than I thought but the normal score for my CPU should be ~18300.
Any idea why this would be happening?
Not sure. The bad part about benchmarks is that you can be competing with “the impossible”, so, sometimes helps if you can compare with someone you can actually trust, instead of the guy who got one run out of 1,000,000 stable using LN2.