>The OpenGL hardware renderer is slow on AMD GPUs due to

File: 1488664232472.png (84KB, 653x726px) Image search: [Google]

84KB, 653x726px

>>60234057
>We do not recommend running outside the processor specifications, such as by exceeding processor frequency or voltage specifications, or removing of the integrated heat spreader (sometimes called 'de-lidding'). These actions will void the processor warranty."

>Check out the link below for further information.

http://www.pcgamer.com/intels-tells-core-i7-7700k-owners-to-stop-overclocking-to-avoid-high-temps/

>>60234057
DELET

>>60234057

Whgat you fail to realise is because opengl is a broken mess AMD's fairly by the book implementation of it is also a broken mess. Nvidia has so many hacks in their driver for opengl at this point to get good performance from opengl you HAVE to play to Nvidia's tune.

It is a bad API that is broken on everything.

>it's a /v/tard finds out the reason that AMD started developing Mantle five years ago and posts it as if it's breaking new episode

File: Vulkan_2.png (310KB, 3000x1688px) Image search: [Google]

310KB, 3000x1688px

Opengl is a fucking joke on anything. We've had Vulkan for a while now.

>>60234431
>Whgat you fail to realise is because opengl is a broken mess AMD's fairly by the book implementation of it is also a broken mess.

This. It was 2-5 years behind in the 00s and they piled hacks on top of hacks to add in the missing features.

And besides why the fuck would you want to use OpenGL on PCSX2 which is one of the few emulators that can take advantage of extreme core counts to accelerate pixel accurate software emulation.... on top of being one of the first to support DX10/11 in hardware accelerated mode, which allowed it to fix a shitloads of emulation bugs that resulted from the limiting nature of DX9.

Using OpenGL in PCSX2 is reinventing the wheel.

>>60234793
b-but OpenGL is more accurate!

>>60234829
no, software mode is more accurate.

Mesa RadeonSI is the best OpenGL driver for AMD hardware, and has been for a while.
On Linux they're even telling users that their official driver is now really only intended for a few pro apps.

>>60234793

The opengl renderer also is more feature complete than the DX renderer as that is where Gregory puts in his effort. For something like PCSX2 vulkan's main benefit is not being spaghetti city but vulkan's documentation is still (iirc) sparse on the ground compared to opengl and DX (including DX12).

This does overlook still AMD hasn't exactly supported opengl as well as they could've which is only compunded on linux where opengl is kind of important.

>>60235101

AMD's turn around for their linux support has been impressive - but the cost for that is basically declaring anything older than hawaii (and sometimes even hawaii) dead and writing it off. Right now unlike windows the only good AMD gpus on linux are the polaris based ones.

>>60235346
Not entirely true, RadeonSI as the name implies supports all GCN hardware back to SI. The AMDGPU kernel module now has support for all GCN hardware, though in some cases its still marked as experimental because the old Radeon module still supports more features for them.

There are still so few Vulkan games, and RadV is still so unoptimized that its still not that big of a deal to use the older kernel module without Vulkan support.
As soon as the HDMI/Display Port audio issues are sorted out they'll probably retire Radeon for GCN hardware and everything will be on a single code base.

You basically want to have a 7950/7970/280/280X, a 470/480/570/580, or a Nano/Fury/Fury X for the open drivers. Everything else is slow or problematic and Vega seems like its going to be under the 'problematic' category until the display code gets merged, or somebody else steps in and writes their own module to go upstream because AMD's in house Windows-centric devs are being their usual incompetent selves.