RadeonSI Gallium3D Might Run As Well As Catalyst For CS:GO Linux

[AnAkIn]

As oposed to that of CS:GO, in the TF2 benchmark I can see the game actual rendering, not the game menu. However, considering that the fps are very similar in both benchmarks, I presume the CS:GO benchmark is correct.

I get 76.17 fps in TF 2 at 1920x1080 (http://openbenchmarking.org/result/1...PL-TEAMFORTR49).

The benchmark should use timedemo. You should see a demo of gameplay footage, otherwise there's a problem, and I doubt the result is correct.

[jsa1983]

As oposed to that of CS:GO, in the TF2 benchmark I can see the game actual rendering, not the game menu. However, considering that the fps are very similar in both benchmarks, I presume the CS:GO benchmark is correct.

I get 76.17 fps in TF 2 at 1920x1080 (http://openbenchmarking.org/result/1...PL-TEAMFORTR49).

[jsa1983]

The fps are 70.88. Uploaded to http://openbenchmarking.org/result/1...PL-CSGO1773885
My system main specs are AMD FX 8150 @ 3.7GHz (i.e. slightly overclocked), 16GB RAM @ 1600 MHz with a Radeon HD 7870 2 GB with mesa git (from September 20) from oibaf, but recompiled with -march=native and llvm-3.6, linux 3.17 rc5 and KDE.

By the way, just to confirm: when the benchmark is starting I get the usual loading screen but then, when all is loaded, the main menu apears and warns about using insecure or having launched the game from outside steam. No real game images are shown (just the menu with it's blueish background) but the fps are shown on the left upper corner.
I guess that is all correct and the benchmark is still valid, right? (I presume so because the fps figure makes sense judging from my game experience).

[smitty3268]

I always see readonsi news. Is r600 still being improved or does it benefit from radeonsi improvements, too?

http://cgit.freedesktop.org/mesa/mesa/log/?qt=grep&q=r600g

[Michael]

Thank you Michael. For that to work I need steam already up and running, right?

You need steam installed, yes. PTS will attempt to install CSGO (via Steam) if you don't have it installed.

[jsa1983]

phoronix-test-suite benchmark csgo

Thank you Michael. For that to work I need steam already up and running, right?

[Michael]

Thanks for the tip! I already knew that one. It's just that I didn't think about benchmarking the fps when I played the game and now when I read the article I didn't have any figures to share, just the great overall impression of the performance.
I'll measure the fps later tonight and share them here in the forum.

phoronix-test-suite benchmark csgo

[jsa1983]

Start steam with:

GALLIUM_HUD=fps steam

and you will know

Thanks for the tip! I already knew that one. It's just that I didn't think about benchmarking the fps when I played the game and now when I read the article I didn't have any figures to share, just the great overall impression of the performance.
I'll measure the fps later tonight and share them here in the forum.

[darkbasic]

I don't know the fps, but probably over 60.

Start steam with:

GALLIUM_HUD=fps steam

and you will know

[gutigen]

I understand the frustration, but unless you've got a 120 Hz monitor, anything over 60fps is redundant anyway.

It is not, higher framerate is actually measured from frame latency which at 60fps is 16.7ms, while for 120 fps it's only 8.3ms which directly translates to input latency which for fast paced games is crucial. Also when machines achieves higher framerate than 60 at most times it's most likely to not drop below 60 when lots of shit happens on the screen (unless game is badly optimized), which results in not dropping frame latency which results in keeping your input latency somehow consitent and responsive which results in me shooting your head faster than you with 60fps or below.

This is not my area of expertise, but ask anyone who has something to do with game or graphics stack development and he will confirm.