03-03-2023, 08:39 PM -
A friend of mine considers trying to get a 7800X3D on launch day but seemingly bad performance in RPCS3 in a preview is putting this into question and we need some expertise or maybe even a 7950X3D owner to resolve this.
In their review of the 7590X3D TechPowerUp simulated the upcoming 7800X3D by disabling the 2nd CCD and among many other benchmarks they stated the average FPS in Red Dead Redemption in RPCS3.
While the fully enabled 7590X3D beat the 13900k (49.4 FPS vs 48.2 FPS) the simulated 7800XD fared much worse with only 30.6 FPS.
The fact that the fully enabled CPU supposedly performs over 60% better cannot be explained by around 15% higher boost clocks on the other CCD.
Especially since the best results were archieved with a scheduler favoring cache (only the first CCD with lower frequency has the 3D cache) (49.4 FPS) and the scheduler favoring frequency only averaged 43.4 FPS.
A few other numbers from that bench for orientation:
13900k: 48.2 FPS
12900k: 42.8 FPS
7900X: 32.5 FPS
7700X: 31.4 FPS
7950X: 30.3 FPS (it seems like this comes from an older benchmark done without AVX 512 and the correct number is 39.1 FPS)https://www.reddit.com/r/rpcs3/comments/...are_button
It's unclear whether or not AVX 512 was used for the other 7000 CPUs and which version of RPCS3 was used.
The only explanation why the fully enabled 7950X3D performs so much better seems to be core count but from all I can see on YT RPCS3 doesn't really use the 2nd CCD (at least on normal 7950X).
Can somebody help us figure that one out?
(the best thing would be a 7950X3D owner comparing performance with CCD1 disabled (or whatever the one without extra cache is called) but any expertise on how much a 2nd 8 core-CCD could realistically boost performance would be extremely helpful)
https://www.techpowerup.com/review/ryzen...ew/13.html
In their review of the 7590X3D TechPowerUp simulated the upcoming 7800X3D by disabling the 2nd CCD and among many other benchmarks they stated the average FPS in Red Dead Redemption in RPCS3.
While the fully enabled 7590X3D beat the 13900k (49.4 FPS vs 48.2 FPS) the simulated 7800XD fared much worse with only 30.6 FPS.
The fact that the fully enabled CPU supposedly performs over 60% better cannot be explained by around 15% higher boost clocks on the other CCD.
Especially since the best results were archieved with a scheduler favoring cache (only the first CCD with lower frequency has the 3D cache) (49.4 FPS) and the scheduler favoring frequency only averaged 43.4 FPS.
A few other numbers from that bench for orientation:
13900k: 48.2 FPS
12900k: 42.8 FPS
7900X: 32.5 FPS
7700X: 31.4 FPS
7950X: 30.3 FPS (it seems like this comes from an older benchmark done without AVX 512 and the correct number is 39.1 FPS)https://www.reddit.com/r/rpcs3/comments/...are_button
It's unclear whether or not AVX 512 was used for the other 7000 CPUs and which version of RPCS3 was used.
The only explanation why the fully enabled 7950X3D performs so much better seems to be core count but from all I can see on YT RPCS3 doesn't really use the 2nd CCD (at least on normal 7950X).
Can somebody help us figure that one out?
(the best thing would be a 7950X3D owner comparing performance with CCD1 disabled (or whatever the one without extra cache is called) but any expertise on how much a 2nd 8 core-CCD could realistically boost performance would be extremely helpful)
https://www.techpowerup.com/review/ryzen...ew/13.html