How To Make CS:GO Run Much Faster On AMD Catalyst For Linux

[varikonniemi]

AMD should be ashamed.

[dungeon]

AMD should be ashamed.

I guess that is copy-paste technology?

http://www.phoronix.com/forums/forum...393#post809393

[paupav]

Does this work for the Dota?

[xeekei]

AMD is going down fast. The programmers assigned to linux are clearly inferior. Open source will not save them. There are plenty of bugs in linux that have been reported 5 years ago and nobody does anything about them. So stop lying with those 5 min shit.

I'm see this post even though Master5000 is on my ignore list.

Doesn't the ignore list work with the new forum?

[carewolf]

You are saying that vulkan suddenly will make developers less bad at coding?

Well, the vulkan API has fewer ways of doing the same thing, and hopefully less reason to modify shaders using a way that is slower on a particular device.

But yeah. I doubt it will go away completely.

[humbug]

Does this work for the Dota?

Even if it does there is no point because Dota 2 is currently in the process of being moved to a new engine, and then away from openGL to Vulkan.

[varikonniemi]

I guess that is copy-paste technology?

http://www.phoronix.com/forums/forum...393#post809393

So exactly like every catalyst release?

In the future i will also point out how they should be ashamed every time we hear about them employing executable name detection or other lame solutions instead of properly fixing their product to work well with all standards compliant code.

Driver side detection and workaround should only be put in place after the game vendor has publicly declined to fix their broken game. And i doubt source engine is broken.

[Kano]

I think AMD just did not ask for the source code at the correct time, Intel did that, Nvidia most likely as the multi threaded GL hack (since 310 driver) was first used by Source engine games. If the code is "fixed" they can only optimize afterwards which requires application profiles. At least they work on it now, better late than never. But i think Nvidia drivers doesn't use that approach by filename, they had an environment var for Doom 3. But maybe I am wrong, just use a random name for a game with benchmark and try yourself.

[dungeon]

@varikonniemi

Don't know if you are aware that those profiles are there for quite some time... let me choose opscur unpopular old example i can remember that it still works, rename tuxracer binary (or some game based on tuxracer) to 'openarena.x86_64' and see what you get

There is nothing common of those two games and engines, but still performance bump is the same for me around 25-30%

Does openarena profile bump performance for any other ioquake based game? Yes, it works. But also for many other games too

Do I recommend doing this renaming? No, but is yours so fill free to experiment if you like... some games might refuse to start or it can be slower actually, etc.

[cutterjohn]

Good article.

This is extremely disappointing from AMD. A case where they already know how to make things work faster but it's not applied due to negligence.

imnho i beg to differ. i fully believe that they're just incompetent given the 5y that iwas stuck with them with a 4850m notebook.

i USED to beleve that their hw was good too, but beyond the shitastic crapalyst that thing would HARD lock hw in certain specific rendring situations in both linux and windows. heat and memory were eliminated, but at the end of the day i claim hw bug as it was never ever fixed along with a fsckton of other bugs.

add to this the fury is still hot running and power hungry even though they claim power efficiency was a goal? add to this crapalyst is still as shitastic as ever, possibly even worse as my oft mentioned 280 experience where a known good/clean catalyst install package would fail in different ways almost every time run meanwhile the old package installed as expected, w7x64p