Rewritten OpenGL Drivers Make AMD's GPUs 'Up To 72%' Faster in Some Pro Apps (arstechnica.com) 23
Most development effort in graphics drivers these days, whether you're talking about Nvidia, Intel, or AMD, is focused on new APIs like DirectX 12 or Vulkan, increasingly advanced upscaling technologies, and specific improvements for new game releases. But this year, AMD has also been focusing on an old problem area for its graphics drivers: OpenGL performance. From a report: Over the summer, AMD released a rewritten OpenGL driver that it said would boost the performance of Minecraft by up to 79 percent (independent testing also found gains in other OpenGL games and benchmarks, though not always to the same degree). Now those same optimizations are coming to AMD's officially validated GPU drivers for its Radeon Pro-series workstation cards, providing big boosts to professional apps like Solidworks and Autodesk Maya. "The AMD Software: PRO Edition 22.Q3 driver has been tested and approved by Dell, HP, and Lenovo for stability and is available through their driver downloads," the company wrote in its blog post. "AMD continues to work with software developers to certify the latest drivers." Using a Radeon Pro W6800 workstation GPU, AMD says that its new drivers can improve Solidworks rendering speeds by up to 52 or 28 percent at 4K and 1080p resolutions, respectively. Autodesk Maya performance goes up by 34 percent at 4K or 72 percent at the default resolution. The size of the improvements varies based on the app and the GPU, but AMD's testing shows significant, consistent improvements across the board on the Radeon Pro W6800, W6600, and W6400 GPUs, improvements that AMD says will help those GPUs outpace analogous Nvidia workstation GPUs like the RTX A5000 and A2000 and the Nvidia T600.
Re: (Score:2)
Re: Imagine what else is broken (Score:2)
This is what I harp on all the time about AMD.
And it's exactly why I love AMD: their shit only ripens and gets more powerful over time with driver improvements.
Pretty much the opposite as Intel.
I Find Radeon Drivers Well Written (Score:2)
Getting 72% improvement in specific situations isn't necessarily an indicate of something written poorly, it can sometimes just be not getting the time to make some insane optimization work right. For instance, I was just working on this piece of code recently and got it working 100% correctly. It was clean, easy to understand and bug free. I was told to make it faster, s
Re: (Score:2)
Speaking as a game developer that uses OpenGL on PC, I've never had an issue with Nvidia. Their stuff has always worked as expected for me.
AMD occasionally gets some weird quirks, but not too bad.
Intel's the worst though. For years I could not get a reliable vsync on Intel. I could write code that worked fine on Nvidia/AMD, but Intel wouldn't follow the spec, so I'd have to make special builds for Intel GPUs. Intel also had a habit of adopting new OpenGL versions years after everyone else did.
Re: (Score:2)
Your opinion differs from a lot of peoples opinions then.
Nvidia is overall considered to be the most stable and accesible platform, and has a history of the standard API's actually working(unlike ATI/AMD) as per documentation.
In fact, that very thing you lament, people using nVidia extensions, grew out of nVidia being far more stable and API compliant, and those extensions were allowed for in the OpenGL API.
AMD's history of boneheaded decisions and implementations on the driver/software side has left them q
Windows vs Linux (Score:2)
Their drivers and software sucks ass.
...on Windows.
The Linux experience is quite decent, actually.
Re: (Score:2)
After a number of bad experiences with nVidia I make sure the GPU is AMD when I buy/build a new PC. The people who wrote and maintain the nouveau driver have always been handicapped by nVidia's code of Omerta, AMD has traditionally been open in that respect.
A couple of caveats there: I'm not a gamer so is not as though nVidia are going to be quaking in their boots over my boycott, and I read here a couple of months ago that nVidia have changed that particular policy.
As for Intel, one old laptop is hardly a
nVidia and Mesa (Score:2)
The people who wrote and maintain the nouveau driver have always been handicapped by nVidia's code of Omerta, AMD has traditionally been open in that respect. {...} I read here a couple of months ago that nVidia have changed that particular policy.
So gamers will be finally able to enjoy non-shitty opensource drivers for nVidia... probably by ~2032.
Seriously, look how long it has taken for AMD to get to their current state (specially, see the very long delay in opensourcing their amdvlk, to the point that radv became a thing, and same long story with regards to opencl).
And that's with opensource devs litteraly on their pay roll, and the amdgpu series of drivers and mesa backends being litterally planned to share code with Windows.
On nVidia's side, the
Windows only (Score:1)
Read the article and clicked through to the AMD press releases: it applies to the Windows drivers only, apparently.
Re: (Score:2)
Re: Windows only (Score:2)
Re:Windows only (Score:5, Interesting)
Linux AMD OpenGL performance always been quite excelent due mesa.
On my machine, i get literally 3x the frame rate on minecraft with mods if compared to windows (20 fps vs 60 fps)
I bet they only got both on par.
Re: (Score:3)
Only if you factor in the last 10 years or so(optimistically) as "always".
AMD, and before them ATI, have a history of poor OpenGL performance or even compliance/compatibility, which has historically been mitigated by Mesa etc.
Indeed, there was a time when ATI(and still partway into the AMD ownership), where they either actively worked to supplant OpenGL with DirectX, or just implemented bare minimums. On the OpenGL ARB, they were obstructive.
It was so bad that the Radeon 9700 couldn't run the SpecViewPerf t
Re: (Score:2)
Well, to AMD/ATI's credit, their obstruction to OpenGL did eventually lead to Vulkan, which these days is a far better graphics API to use than OpenGL. Especially if you want to write performant apps - OpenGL can do things in the background at the worst possible time leading to all sorts of
Re: (Score:2)
The problem with that view is that even after Vulkan was first released, AMD still pushed Mantle, especially in the HPC field.
Also, there are use cases where OpenGL is still a good option
Re: (Score:2)
i have a vega56 and i still have significant (but not game-breaking) violations of the openGL spec.
isn't there some kind of conformance test that a vendor has to run successfully before claiming "opengl 4.5 supported" or something?
can I sell a box of rocks with a sticker "opengl" on it?
Re: (Score:2)
Nobody does monoliths these days. It will likely materialize on Linux as well, just a bit later.
Re: Windows only (Score:2)
The Linux opengl drivers were already performing at this level though. AMD just had real shitty windows opengl performance.
Re: (Score:2)
Ah, so they probably ported the core of the driver to Windows. Makes sense.
Tn: (Score:2)
we fixed the performance in the worlds most commercially succesful game
it only took us 10 years
amd's opengl support is roughly on par with intel's. They trade blows in how shitty they are. They straight up violate the spec. Directly.
source: been coding opengl and it so happens i have a vega56 and an intel iris something-or-other
100% CPU usage (Score:2)
Are they EVER going to fix the bug that makes CPU usage spike over 100% every time the mouse is moved? I am told there is a fix for it, but every distro fights like hell to keep you from installing the proprietary drivers with the end result that AMG GPU is a complete failure in Linux.