Originally posted by dungeon
View Post
Announcement
Collapse
No announcement yet.
Intel Pentium vs. AMD Ryzen 3 Performance For Linux Gaming
Collapse
X
-
Originally posted by dungeon View Post
Real questions require real answers i guess Openarena have no mesa_glthread profile by default as i see here:
https://cgit.freedesktop.org/mesa/me...src/util/drirc
While Bioshock have it, but that does not matter... as again same thing as you see - nVidia again scale up, but AMD scale down
Something is still missing in amdgpu multithread optimizations it seems or might be just not tested on AMD CPUs or something. While nVidia seems value both singlethread IPC but also more whatever cores, AMD value mostly single IPC but not more cores glthread profile seems helped here on Intel CPU, but on AMD one and with more cores it beceme worse, etc... while opposite happens with nVidia.
Which means or at least that is how these cases looks like to me, is that more of whatever CPU cores you have nVidia will win every timeLast edited by duby229; 24 January 2018, 12:22 AM.
Comment
-
Originally posted by duby229 View Post
I don't think that's quite right, as you can see that Intel CPU has a higher clock speed and slightly better IPC. I think that's the reason you see AMD cards perform slightly better on the Pentium. That 18FPS difference seems about the right margin for the actual performance difference. But I do think you are right about nVidia though.
Problem is amdgpu driver here, as if someone change let say his 2 core CPU he don't know what he will get as some CPU with 4 cores upgrade so to more cores might also be performance downgrade in games
I think common sense is that more cores are better and as we can see nvidia driver take advantage of that, meanwhile with amdgpu driver it could easely be and by common sense unexpectedly worse, just by looking at these resultsLast edited by dungeon; 24 January 2018, 01:07 AM.
Comment
-
Originally posted by dungeon View Post
Both CPUs are just fine, proved by nvidia driver as that scales there on both. 2 core or 4 does not matter of CPU vendor whatever - it just scales up
Problem is amdgpu driver here, as if someone change CPU he don't know what he will get as some CPU upgrade to more cores might also be performance downgrade in games
EDIT: If Michael had more equally matched CPU's to compare I think you'd see. The AMD Ryzen does not have the same number of cores or the same clock or the same IPC. If you doubled the number of cores on that Pentium then it would be faster than that Ryzen even with the AMD cards.Last edited by duby229; 24 January 2018, 01:15 AM.
- Likes 2
Comment
-
Originally posted by duby229 View Post
I don't think you're right, it's clock speed and IPC difference you are seeing on the AMD card results. The Pentium is clocked higher and has a slightly higher IPC. That's what you're seeing. I doubt very strongly it has anything to do with amdgpu at all, it's more likely that mesa doesn't multithread its load as effectively as nvidias driver does.
Either that glthread is not tested on AMD CPUs or some unusual shit happens in amdgpu with 2 to 4 core thread optimizations or... well, could be whatever else, but this test clearly shows this scale up and scale down tendency difference between these driversLast edited by dungeon; 24 January 2018, 01:17 AM.
Comment
-
Originally posted by dungeon View Post
I think i am right nvidia scale up there, while amdgpu scale down in performance by increasing number of cores.
Either that glthread is not tested on AMD CPUs or some unusual shit happens in amdgpu with 2 to 4 core switch or... well, could be whatever else, but this test clearly shows this scale up and scale down tendency difference between these drivers
IF it was exactly the same CPU between those results I would totally agree with you, but they aren't the same and the differences in the benchmarks correlate with the differences in hardware.
EDIT: I don't think it has anything to do with amdgpu, it most likely has to do with mesa not optimizing for multithreading as effectively as nvidia's driver does.Last edited by duby229; 24 January 2018, 01:24 AM.
- Likes 6
Comment
-
Originally posted by duby229 View PostIf Michael had more equally matched CPU's to compare I think you'd see. The AMD Ryzen does not have the same number of cores or the same clock or the same IPC. If you doubled the number of cores on that Pentium then it would be faster than that Ryzen even with the AMD cards.Last edited by dungeon; 24 January 2018, 01:26 AM.
Comment
-
Originally posted by dungeon View Post
Please forget about CPUs, put an ARM CPU there and test cards if you want does not matter and I don't care really which one CPU is better here I only talk about GPU drivers diff here and this test is more than enough to me.
Comment
-
Originally posted by duby229 View Post
It did -NOT- scale down. That Ryzen is slower than that Pentium. That's what you're seeing. That Pentium has a higher clock speed and slightly higher IPC. The differences in FPS correlate with the actual difference in performance roughly.
IF it was exactly the same CPU between those results I would totally agree with you, but they aren't the same and the differences in the benchmarks correlate with the differences in hardware.
EDIT: I don't think it has anything to do with amdgpu, it most likely has to do with mesa not optimizing for multithreading as effectively as nvidia's driver does.Last edited by dungeon; 24 January 2018, 01:35 AM.
Comment
-
Originally posted by dungeon View Post
I agree, you don't think
edit: It did -not- scale down, that Pentium is faster than that Ryzen. And the difference in performance correlates with the difference in hardware.Last edited by duby229; 24 January 2018, 01:36 AM.
- Likes 1
Comment
Comment