home.social

#amdgpu — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #amdgpu, aggregated by home.social.

  1. @benoit Make sure you run a very recent kernel with Ryzen AI because the #amdgpu driver had rare but nasty crashes with compute workloads.

    Also, the Mediatek WiFi and Bluetooth is crap. I had packaet drops with WiFi at 2.4GHz (my only option for the hotspot) and Bluetooth stopped working in #Linux 7.1-rc3. Yes, it's an RC kernel, but they asked to backport the bogus patch to stable kernels too. Not sure if it made it there already.

    Anyway, sloppy maintainers are sloppy.

  2. @benoit Make sure you run a very recent kernel with Ryzen AI because the #amdgpu driver had rare but nasty crashes with compute workloads.

    Also, the Mediatek WiFi and Bluetooth is crap. I had packaet drops with WiFi at 2.4GHz (my only option for the hotspot) and Bluetooth stopped working in #Linux 7.1-rc3. Yes, it's an RC kernel, but they asked to backport the bogus patch to stable kernels too. Not sure if it made it there already.

    Anyway, sloppy maintainers are sloppy.

  3. @benoit Make sure you run a very recent kernel with Ryzen AI because the #amdgpu driver had rare but nasty crashes with compute workloads.

    Also, the Mediatek WiFi and Bluetooth is crap. I had packaet drops with WiFi at 2.4GHz (my only option for the hotspot) and Bluetooth stopped working in #Linux 7.1-rc3. Yes, it's an RC kernel, but they asked to backport the bogus patch to stable kernels too. Not sure if it made it there already.

    Anyway, sloppy maintainers are sloppy.

  4. @benoit Make sure you run a very recent kernel with Ryzen AI because the #amdgpu driver had rare but nasty crashes with compute workloads.

    Also, the Mediatek WiFi and Bluetooth is crap. I had packaet drops with WiFi at 2.4GHz (my only option for the hotspot) and Bluetooth stopped working in #Linux 7.1-rc3. Yes, it's an RC kernel, but they asked to backport the bogus patch to stable kernels too. Not sure if it made it there already.

    Anyway, sloppy maintainers are sloppy.

  5. @benoit Make sure you run a very recent kernel with Ryzen AI because the #amdgpu driver had rare but nasty crashes with compute workloads.

    Also, the Mediatek WiFi and Bluetooth is crap. I had packaet drops with WiFi at 2.4GHz (my only option for the hotspot) and Bluetooth stopped working in #Linux 7.1-rc3. Yes, it's an RC kernel, but they asked to backport the bogus patch to stable kernels too. Not sure if it made it there already.

    Anyway, sloppy maintainers are sloppy.

  6. CW: AMDGPU: HDMI 2.1 120 Hz+

    »#Valve could secretly "encourage" that open source developer's work.«

    »Considering that one of the devs, in response to a question in the comments of a Reddit post, stated they "declined to comment" on whether Valve was involved… Yeah I think Valve probably had at least a bit of involvement with the community reverse-engineering effort.«

    phoronix.com/forums/forum/phor

    #AMDGPU #HDMI

  7. CW: AMDGPU: HDMI 2.1 120 Hz+

    »#Valve could secretly "encourage" that open source developer's work.«

    »Considering that one of the devs, in response to a question in the comments of a Reddit post, stated they "declined to comment" on whether Valve was involved… Yeah I think Valve probably had at least a bit of involvement with the community reverse-engineering effort.«

    phoronix.com/forums/forum/phor

    #AMDGPU #HDMI

  8. CW: AMDGPU: HDMI 2.1 120 Hz+

    »#Valve could secretly "encourage" that open source developer's work.«

    »Considering that one of the devs, in response to a question in the comments of a Reddit post, stated they "declined to comment" on whether Valve was involved… Yeah I think Valve probably had at least a bit of involvement with the community reverse-engineering effort.«

    phoronix.com/forums/forum/phor

    #AMDGPU #HDMI

  9. CW: AMDGPU: HDMI 2.1 120 Hz+

    »#Valve could secretly "encourage" that open source developer's work.«

    »Considering that one of the devs, in response to a question in the comments of a Reddit post, stated they "declined to comment" on whether Valve was involved… Yeah I think Valve probably had at least a bit of involvement with the community reverse-engineering effort.«

    phoronix.com/forums/forum/phor

    #AMDGPU #HDMI

  10. CW: AMDGPU: HDMI 2.1 120 Hz+

    »#Valve could secretly "encourage" that open source developer's work.«

    »Considering that one of the devs, in response to a question in the comments of a Reddit post, stated they "declined to comment" on whether Valve was involved… Yeah I think Valve probably had at least a bit of involvement with the community reverse-engineering effort.«

    phoronix.com/forums/forum/phor

    #AMDGPU #HDMI

  11. RE: mastodon.social/@clifmo/116456

    Just wrapped a fully local LLM stack on Fedora + AMD RX 7900 XTX (24GB VRAM).

    🔹 Engine: llama.cpp w/ Vulkan/RADV backend
    🔹 Model: Qwen3.6-35B-A3B (MoE, Q4_K_M, ~21.4GB)
    🔹 Config: -ngl 99 --ctx-size 16384 --batch-size 512
    🔹 Perf: ~137 tok/s gen, ~344 tok/s prompt, rock-solid at 16k context
    🔹 All VRAM. Zero CPU offload.

    Powering AnythingLLM for private RAG pipelines. 🐧⚡

    #LocalLLM #Fedora #AMDGPU #Vulkan #OpenSourceAI #RAG #LLM

  12. RE: mastodon.social/@clifmo/116456

    Just wrapped a fully local LLM stack on Fedora + AMD RX 7900 XTX (24GB VRAM).

    🔹 Engine: llama.cpp w/ Vulkan/RADV backend
    🔹 Model: Qwen3.6-35B-A3B (MoE, Q4_K_M, ~21.4GB)
    🔹 Config: -ngl 99 --ctx-size 16384 --batch-size 512
    🔹 Perf: ~137 tok/s gen, ~344 tok/s prompt, rock-solid at 16k context
    🔹 All VRAM. Zero CPU offload.

    Powering AnythingLLM for private RAG pipelines. 🐧⚡

    #LocalLLM #Fedora #AMDGPU #Vulkan #OpenSourceAI #RAG #LLM

  13. RE: mastodon.social/@clifmo/116456

    Just wrapped a fully local LLM stack on Fedora + AMD RX 7900 XTX (24GB VRAM).

    🔹 Engine: llama.cpp w/ Vulkan/RADV backend
    🔹 Model: Qwen3.6-35B-A3B (MoE, Q4_K_M, ~21.4GB)
    🔹 Config: -ngl 99 --ctx-size 16384 --batch-size 512
    🔹 Perf: ~137 tok/s gen, ~344 tok/s prompt, rock-solid at 16k context
    🔹 All VRAM. Zero CPU offload.

    Powering AnythingLLM for private RAG pipelines. 🐧⚡

    #LocalLLM #Fedora #AMDGPU #Vulkan #OpenSourceAI #RAG #LLM