Company of Heroes 2 Is The Latest Linux Game Showcasing AMD's Performance Wreck With Catalyst

[Martin76]

That is actually better than i was expecting !
Seems the game needs lots of VRAM....tell me, is the iGPU at stock speed or did you OC it to +/- 950MHz ? If still stock, OC it, you can expect +10% minimum of performance increase.

Can you run game at 1024x768, 1280x720 and 1280x1024 (or 1280x960) to see how it goes ?

TIA,
AJSB

Stock speed, 2x4 GiB 1866 Mhz memory.
I have the stock cooler and i dont know the actual temperature of the cpu, i dont like OC.

Image Quality Minimum:
1920x1080 average 10.5, max 17.9
1280x1024 average 12.5, max 15.5
1280x720 average 14.4, max 23.8

Image Quality Medium:
1080p average 8.8, max 15.9
720p average 11.2, max 19.4

Image Quality Maximum:
1080p average 7.3, max 14.3
720p average 10.4, max 18


1080p with everything low:
Antialiasing off 11.7, 19.2
Antialiasing low (FXAA) 11.7, 19.6
Antialiasing med (SSAA 225%) 8, 14.8
Antialiasing high (SSAA 400%) 5.6, 9.4



Windows 7 (for reference purposes)

1080p Image Quality Minimum 22.2, 37.5
1080p Image Quality Medium 15.2, 26.1
1080p Image Quality Maximum 12.5, 23.2

Ingame tutorial: 24 fps maximum quality, 30 fps medium quality, 45 fps minimum quality

[GreekGeek]

Hi Michael & Yall,

keep up the excellent work Michael.

Objective data on BS ATI/AMD performance is essential, to guide real world buying decisions of peeps on Linux. When it comes to ATI/AMD, we are clearly not in the "....Company of Heroes."

The usual Astroturfers and noise generators, are going to do there stuff. Pity the flat earthers really.... FYI, there is no Santa or Easter Bunny either! LOL.

After ATI/AMD management sort there !hit out, all power to them.


GreekGeek :-)

[vadimg]

Hi Michael & Yall,

keep up the excellent work Michael.

Objective data on BS ATI/AMD performance is essential, to guide real world buying decisions of peeps on Linux. 

Providing a real benchmark data is one thing, making an every day sensationalist articles of it like "AMD is a wreck" is another thing. Just hope you understand the difference. No one asks Michael to lie, just maybe to not make too much useless noise of it.

FWIW, I'm grateful to Michael for all benchmarks. But the point stated above still stands.

[GreekGeek]

Hi Yall & "Vadimg,"

who argues "making an every day sensationalist articles of it like "AMD is a wreck" is another thing. Just hope you understand the difference. No one asks Michael to lie, just maybe to not make too much useless noise of it.," on post #74.

There is nothing whatsoever sensationalist about his title, simply because the data supports it. Anything less would be astroturfing, marketing or a lie...

The obvious reason he published the article is, I quote Michael, "....this game that's now available on Linux as of earlier today is a disaster if trying to use the AMD Catalyst Linux driver." What better reason to publish and unfortunately point out that Catalyst is train wreak city again.

If "vadimg" comes up with something half way reasonable to say, I may well listen. But so far, all I am hearing is yet more white noise.

GreekGeek :-)

[Pontostroy]

hd 7970 high settings - 21 fps (gamingonlinux.com gtx 970 - 41 fps )

http://www.gearsongallium.com/?p=2382

[whitecat]

Probably something goes out of VBO memory, try different lower down settings maybe if possibile.

It crash even with the lowest settings.

[whitecat]

Thanks. Doesn't look like my fault with SB though, so hopefully someone else looks into it. I'm just trying to have some rest currently. But I own that game, so maybe I'll try to look into it.

In fact, the former crash I got is different from the one I filled in bugzilla.
The bug reported on bugzilla happens when zooming on the map and doesn't leads to any dmesg message. And it doesn't looks like do be sensitive to sb.
The former crash occurred once the map loaded (I can't even see anything before it crashed). That said I can't reproduce anymore this crash, I don't now why. Can the game perform a shader cache ? Which, since compiled without sb, are then now reused every time even if sb is re-enabled ? I don't now if an app can do this. And I don't remember Gallium has a shader cache implemented.

[Kano]

Mesa has a shader cache in the home since 10.3.

[log0]

Hi Yall & "Vadimg,"

who argues "making an every day sensationalist articles of it like "AMD is a wreck" is another thing. Just hope you understand the difference. No one asks Michael to lie, just maybe to not make too much useless noise of it.," on post #74.

There is nothing whatsoever sensationalist about his title, simply because the data supports it. Anything less would be astroturfing, marketing or a lie...

The obvious reason he published the article is, I quote Michael, "....this game that's now available on Linux as of earlier today is a disaster if trying to use the AMD Catalyst Linux driver." What better reason to publish and unfortunately point out that Catalyst is train wreak city again.

If "vadimg" comes up with something half way reasonable to say, I may well listen. But so far, all I am hearing is yet more white noise.

GreekGeek :-)

Sure, because you and Michael can show without any uncertainty and doubt that the driver is at fault here and not the game doing something stupid.

[whitecat]

Mesa has a shader cache in the home since 10.3.

Are you sure ?
I remember that an Intel developer tried to implement a shader cache some months ago and I tested its patches for Gallium, but I don't remember the work has been commited.
In any case, where the cache lives more exactly ?