Typical really, the day after I do a review of 2017 and mention how AMD announced they would finally release it, but still didn't, they then go and do it today. The 'AMDVLK' is now officially under the MIT license!
The announcement I saw came courtesy of Matthäus G. Chajdas from AMD on Twitter, which was retweeted by another developer I follow to end up in my timeline.
The curious thing now, is what will happen in the open source driver space for AMD GPUs. Since the Mesa RADV driver has come a long way, it will be interesting to see what happens between the two efforts. It is entirely seperate to Mesa, with it being built to work with AMD's Platform Abstraction Library (PAL). That's the awesome thing, their PAL is also now open source under the MIT license.
Since it confused me at first: the xgl repository has the Vulkan implementation that sits on top of the PAL code. The actual AMDVLK repository is information about it all.
AMD have listed support for these GPUs:
- Radeon™ HD 7000 Series
- Radeon™ HD 8000M Series
- Radeon™ R5/R7/R9 200/300 Series
- Radeon™ RX 400/500 Series
- Radeon™ M200/M300/M400 Series
- Radeon™ RX Vega Series
- AMD FirePro™ Workstation Wx000/Wx100/Wx300 Series
- Radeon™ Pro WX x100 Series
- Radeon™ Pro 400/500 Series
AMD are only listing support for Ubuntu 16.04.3 and RedHat 7.4 right now, but hopefully now it's fully open work can be done to enable good support for a wider selection of distributions.
It sure took them a while, but with open source releases they do end up having to go through legal hoops to ensure there's nothing in the code that they're not actually allowed to open source. There's probably plenty of other checks it has to go through too, but it is fantastic that AMD have done this.
You can see the full repository on GitHub here. For some extra background info, see this article on the gpuopen website.
Seeing the PAL code though.
Last edited by STiAT on 22 December 2017 at 12:54 pm UTC
We open sourced our @VulkanAPI driver for lots of Radeon GPUs! That includes PAL, our low-level abstraction that the Vulkan and D3D12 drivers are built on top of!
So it is basically AMD official driver open sourced for both Linux *and* Windows. People making graphics engines for cross platform certainly will like additional ways to debug their problems.
Hmh, I don't see anything released ... am I missing the source code? There is none visible for me in the repo.From the Telegram Group chat, the AMDVLK is just info. The xgl repo is the vulkan api implementation that works with the PAL code.
Seeing the PAL code though.
Nobody should mention this other website now...! ;)
![](https://images.duckduckgo.com/iu/?u=https%3A%2F%2Fmedia.giphy.com%2Fmedia%2FGCjIE9C9Eb9V6%2F200.gif&f=1)
Nobody should mention this other website now...! ;)I'm assuming you mean Phoronix? I feel like I'm missing something?
Hmh, I don't see anything released ... am I missing the source code? There is none visible for me in the repo.From the Telegram Group chat, the AMDVLK is just info. The xgl repo is the vulkan api implementation that works with the PAL code.
Seeing the PAL code though.
Ahh, ye, looks good. I'll start digging :-).
Would be awesome to see AMD becoming a real alternative to NVidia again. I certainly like their way to handle the community better than NVidia's...
Drivers wise it's already a great alternative. Now they need to catch up on hardware. Vega is not very competitive on power consumption.
I'm very interested in going AMD for either/both CPU and/or GPU on my next overhaul.Would be awesome to see AMD becoming a real alternative to NVidia again. I certainly like their way to handle the community better than NVidia's...
Drivers wise it's already a great alternative. Now they need to catch up on hardware. Vega is not very competitive on power consumption.
Would be awesome to see AMD becoming a real alternative to NVidia again. I certainly like their way to handle the community better than NVidia's...
Not until they release a 1080ti competitor under 300tdp. I mention the TDP because its quite important for ITX and temperature control, preferably with 7nm we should get significantly lower power demand and a 1080ti competitor.
A 1080TI has become the baseline for VR gaming AND 4k gaming, I'm in allot of pain with my FuryX but just don't have 1k laying around to spend on a Ti atm :(
preferably with 7nm we should get significantly lower power demand and a 1080ti competitor.
7 nm Vega refresh is supposed to happen in H1 2018, so I'm waiting for Sapphire to make something sensible with it, and not that 3 8-pin power connectors monster.
preferably with 7nm we should get significantly lower power demand and a 1080ti competitor.
7 nm Vega refresh is supposed to happen in H1 2018, so I'm waiting for Sapphire to make something sensible with it, and not that 3 8-pin power connectors monster.
Ha, I wonder if we'll ever hit that nice plateau of power, so we can finally start working toward power consumption. Personally I think the companies would serve customers better by doing a raw power version upon release of a new chip/family, then for their refresh, release a card as powerful, but that runs quieter/less power hungry. These should yield higher overclocks for those that tend to go that route anyhow.
Last edited by Shmerl on 28 December 2017 at 9:58 pm UTC
See more from me