Compare Your Linux System's OpenGL Performance Side-By-Side With The Radeon RX 480

[phoronix]

Phoronix: Compare Your Linux System's OpenGL Performance Side-By-Side With The Radeon RX 480

Here are some 1080p OpenGL results (as opposed to our plethora of 1440p and 4K data today) for the brand new Radeon RX 480 and available via OpenBenchmarking.org so you can easily compare your own Linux system(s) performance against these reference numbers using the open-source driver stack...

http://www.phoronix.com/scan.php?page=news_item&px=RX-480-Side-By-Side-GL

[chuckula]

I have GPU Test installed for some quick benchmarking on my GTX-1080.
All tests are run at 2560x1440 fullscreen:
1. Piano test: 1732 points
2. Furmark: 6149 points
3. Triangle: 8626 points
4. Plot3D: 8626 points
5. Volplosion: 5408 points
6. Tessmark (at fulll 64 samples): 8615 points
7. Gimark (giant donut benchmark): 8495 points


Incidentally, G-sync is turned on here. I noticed that the triangle and plot3d results showed up as 143 FPS with an identical score, and I'm using a 144 FPS monitor. I think those scores may be artificially held back because I'm being frame rate limited, FYI. Tessmark may have been similar since it was also at 143 fps and almost at 8626 points.

[chuckula]

Re-ran a few tests with both G-sync and V-sync turned off.
A few of the previously rate-capped tests liked it.

1. Triangle: 225681 (at 3760 FPS!)
2. Plot3d: 132981 (at 2218 FPS)
3. Tessmark: 23603 points (at 393 FPS)
4. GiMark: 8878 points at 147 FPS, so that one is a close call.

I think it just goes to show the GPU settings you use for benchmarks are not necessarily the ones you want to use in real life.

[Kano]

Dota 2 looks so slow, it barely beats a GTX 650 Ti 1 GB... Not sure if VRAM or system RAM affects the Vulkan speed, i guess the Vulkan codepath allocated too much memory. With a 4 GB Haswell box i even need swap if I try Vulkan, OpenGL is fine. Haswell still has Vulkan rendering issues. Maybe I run some more tests... Tomb Raider and DiRT Showdown would be interesting for me as I played those games lately.

https://openbenchmarking.org/result/1606297-KANO-GTX650T91

[bridgman]

Yeah, DOTA2 is pretty much CPU limited at 1080. Testing at 1440 or 4K probably makes more sense.

[dungeon]

Yeah, DOTA2 is pretty much CPU limited at 1080. Testing at 1440 or 4K probably makes more sense.

So how you suggests to Kano test at those resolutions, when his screen is 1920x1200? Using virtual super resolution?

8K shows best GPUs of course but i guess Michael intentionally choose FullHD resolution, as that is what most people supposed to have - as whole purpose of this is that most people can compare

[Kano]

Maybe Dota 2 Vulkan would be interesting too, with Nvidia it seems to be faster with a newer card. But it is really funny that you can keep a really old and just midrange card if you mainly play this game... Intel seems to suffer extremely in that benchmark, maybe a different preset would give playable results - fps below 20 are not enough for this game type. If you play a single player game like Tomb Raider 30+ fps are playable for me, that's what I try to achieve with the GFX settings. Competitive shooters need always over 60 fps.

[dungeon]

Ignorance is futile Current steam survey shows that 25.72% plays at 1366 x 768 so laptops and that most people 36.81% have 1920x1080... only 1.53% 2560 x 1440 and 4K does not even exist ... it goes to Other column i guess which is in whole 1.96% together with others whatever that can also be

Likely low adoption curve of people who have 4K monitor and/or VR equipment, answering why most people use integrated graphics or some low to mid range dedicated chip/cards as with low res one don't really need more

[bug77]

Ignorance is futile Current steam survey shows that 25.72% plays at 1366 x 768 so laptops and that most people 36.81% have 1920x1080... only 1.53% 2560 x 1440 and 4K does not even exist ... it goes to Other column i guess which is in whole 1.96% together with others whatever that can also be

Likely low adoption curve of people who have 4K monitor and/or VR equipment, answering why most people use integrated graphics or some low to mid range dedicated chip/cards as with low res one don't really need more

Well, since you can't get proper 4k performance without spending well over $500, I can see how even people that do have a 4k monitor use it for anything but gaming. You can do photo/video editing, programming, DTP and what not on a 4k monitor, but gaming is still way too expensive.

[puleglot]

... amd does not care about wine-staging csmt ...

Can you please explain this statement in more detail?