For many months now there's been work on an AMD HSA IL front-end for GCC with supporting the BRIG binary form of the Heterogeneous System Architecture Intermediate Language (HSA IL). It's getting late into GCC 7 development and onwards to its final development stage while this new front-end has yet to be merged.
It doesn't look like the Talos Secure Workstation will see the light of day with it's crowdfunding campaign ending this week and it's coming up more than three million dollars short of its financing goal. Now there's another effort to offer a libre system but using off-the-shelf x86 hardware.
AMD isn't using CES 2017 to launch their Ryzen (Zen) processors or Vega graphics cards, but at least they have opened up more Vega architecture details for this busy week in Las Vegas.
AMD's upcoming Ryzen (Zen) processors appear to be in good enough shape that they are working on the current mainline kernel as far as I can tell based upon limited information available prior to getting my hands on the CPUs or getting any official announcement from AMD, but some Linux kernel patches have yet to be mainlined. The yet-to-be-merged work appears to be more for non-core features and Zen server functionality with those CPUs shipping later than the upcoming Ryzen desktop CPUs.
Patches were published today for supporting "Polaris 12" graphics cards within the AMDGPU DRM kernel driver on Linux.
The Radeon Open Compute platform has been updated and quietly released prior to the weekend. The ROCm 1.4 release comes with preliminary OpenCL support.
In recent days there have been a few Phoronix readers inquiring why I am not testing with my Radeon R9 290 graphics card in all our frequent comparisons and driver benchmarks. The short story is that the regression since Linux 4.7 remains and for my Radeon R9 290 and others with select Hawaii graphics cards, there still is a performance regression. Though over Christmas I hope to finally find the time to bisect it.
Well this weekend is exciting for AMDGPU users and open-source AMD fans. Yesterday was the news we published about Valve looking to improve AMDGPU/RADV for their Vulkan-based VR experience while the latest is work from AMD that implements GPU virtualization support within the AMDGPU driver.
AMD's Zen New Horizon event is going on right now. For those missing out on the livestream, here are my live details so far on Zen, or now officially known as Ryzen.
For those interested in the upcoming Zen processors, a quick reminder that later today is AMD's livestream event where they will be giving a "sneak preview" of the upcoming Zen CPU.
With Linux 4.9 expected for release this weekend and the 4.10 merge window to then immediately open, Alex Deucher of AMD sent in an early batch of fixes atop the earlier feature material of AMDGPU/Radeon DRM changes for Linux 4.10.
This week's news of Qualcomm sampling a 10nm 48-core ARMv8 SoC for servers made me wonder where are AMD's ARM SoCs -- and the long-awaited development boards -- and thus been following up with a few sources this week.
AMD ran into a snag getting out the updated proprietary hybrid Linux driver stack this morning, but it's now available for download from AMD.
The latest patches for the AMDGPU DRM driver's DC code -- what was previously known as DAL -- have been published and they reduce the size of the code-base some more.
AMD's big display abstraction layer (DAL) code-base that's used by AMDGPU-PRO but not yet mainlined in the Linux kernel for providing HDMI 2.0, future FreeSync/Adaptive-Sync, HDMI/DP audio, and other modern display features is DAL no more.
Earlier this year I exclusively reported on the "Radeon Settings" GUI control panel may be open-sourced for AMD Linux users but since then I hadn't heard anything publicly or privately about getting this graphics driver control panel on Linux for AMDGPU-PRO and the fully-open AMDGPU stack. But it looks like that it's still being worked on internally at AMD.
While many in our forums and other Linux communities want to see "AMDGPU-PRO die" or for AMD to stop supporting the hybrid/proprietary driver given the pace of RadeonSI development for OpenGL and the emerging RADV for (unofficial) Vulkan support, OpenCL remains one of AMDGPU-PRO's strongholds. AMD has been working on opening up their proprietary compute stack, but for now it's there. Here are some fresh AMDGPU-PRO 16.40 benchmarks versus NVIDIA in LuxMark, one of the real-world OpenCL workloads where the AMD blob does very well.
Quietly landing last week into the mainline Linux kernel as part of the AMDGPU fixes is support for tear-free PRIME offloading between Intel and AMDGPU.
AMD used the SC16 super-computing conference today announce version 1.3 of the Radeon Open Compute platform.
The recent slowdowns seen with AMDGPU-PRO 16.40 on my test systems may be attributed to the Linux 4.8 kernel being not properly supported by this hybrid kernel driver.
The AMDGPU backend within LLVM that's used by the open-source Radeon graphics driver stack has landed half-precision / FP16 support.
There is finally a new release of the AMDGPU-PRO hybrid Radeon graphics driver stack for Linux.
With AMD's forthcoming Zen processors is support for some new memory encryption technologies that are of particular benefit for virtualized environments.
Harry Wentland of AMD just presented at the XDC2016 conference about DAL, the big Display Abstraction Layer code-base, which many AMD Linux users have been waiting to see merged in order to have Polaris audio support and this is one of the stepping stones for seeing FreeSync/Adaptive-Sync and other modern display capabilities.
The AMDGPU DRM code for Linux 4.9 is already queued in DRM-Next with virtual display support and other features as well as experimental GCN Southern Islands support while today another feature pull request was submitted to DRM-Next with more Radeon/AMDGPU changes.
One month after the first AMDGPU feature pull of new functionality for DRM-Next to in turn land in Linux 4.9, the second feature pull request has now been sent out and it presents experimental Southern Islands (GCN 1.0) support for AMDGPU.
Last week a Mesa fix landed to target the Radeon R9 290/390 performance regression that's been covered a few time on Phoronix since the issue was originally noticed. While the Mesa fix is working for some users, it didn't fix all problems, including with my Hawaii test card.
The HIP project has made good progress over the summer. HIP from AMD's GPUOpen project is part of the puzzle for converting CUDA to portable C++ code. That source code can then run on AMD GPUs while having little to no performance impact, at least according to AMD.
Things don't appear to be looking up for AMD's ARM efforts. It's looking like we probably won't be seeing AMD ARM development boards publicly available this year, if not the end of 2016, and there won't be many of them going around.
AMD this week open-sourced the Advanced Media Framework (AMF) as their replacement to the earlier AMD Media SDK. But before getting too excited about this latest AMD open-source project, there isn't yet any Linux support.
770 AMD news articles published on Phoronix.