#zluda — Public Fediverse posts

#ZLUDA Boasts Full #Llamacpp Support, Better Windows Handling For #CUDA On Non-NVIDIA GPU
ZLUDA developers had hoped to provide robust PyTorch support by end of 2025 but that didn't quite turn out. Llama.cpp support using NVIDIA CUDA back-end is now considered complete. Llama.cpp has #AMD ROCm but having ZLUDA support for Llama.cpp allows for performance comparisons. ZLUDA devs report that their route achieves "nearly identical" performance to the native ROCm back-end.
https://www.phoronix.com/news/ZLUDA-Q4-2025-Report

#ZLUDA Adds #CUDA 13.1 Compatibility For Running CUDA Apps On Non-NVIDIA Hardware
The current form of ZLUDA is on delivering CUDA for non-NVIDIA GPUs but not explicitly being tied to one #GPU target/vendor like in the past. This multi-GPU CUDA implementation though has been taking a special focus on AI workloads but ultimately this implementation will hopefully allow most unmodified CUDA software to run gracefully on other GPU hardware.
https://www.phoronix.com/news/ZLUDA-CUDA-13.1-Compatibility

#ZLUDA For #CUDA On Non-NVIDIA GPUs Enables #AMD #ROCm 7 Support
ZLUDA is working on bringing CUDA to non-NVIDIA GPUs. While there were prior versions focused originally on Intel GPUs and then for a while AMD-financed work on Radeon/ROCm support, the current take is on being a multi-vendor CUDA implementation and with a special focus on getting CUDA #AI workloads up and running.
https://www.phoronix.com/news/ZLUDA-ROCm-7

ZLUDA: CUDA ohne NVIDIA-GPU unterstützt ROCm7
https://linuxnews.de/zluda-cuda-ohne-nvidia-gpu-unterstuetzt-rocm7/ #zluda #cuda #nvidia #amd #rocm #linux #linuxnews

#ZLUDA 5 Released With An Offline #Compiler For #CUDA On Non-NVIDIA #GPUs
#ZLUDA5 introduces "zoc" as an offline compiler for ZLUDA. #Zoc will take a NVIDIA PTX file as input and output the #LLVM intermediate representation (IR) and the #RDNA Assembly for the GPU in use generated in turn by the #ROCm compiler.
https://www.phoronix.com/news/ZLUDA-5-Released

#ZLUDA 5 Released With An Offline #Compiler For #CUDA On Non-NVIDIA #GPUs
#ZLUDA5 introduces "zoc" as an offline compiler for ZLUDA. #Zoc will take a NVIDIA PTX file as input and output the #LLVM intermediate representation (IR) and the #RDNA Assembly for the GPU in use generated in turn by the #ROCm compiler.
https://www.phoronix.com/news/ZLUDA-5-Released

#ZLUDA 5 Released With An Offline #Compiler For #CUDA On Non-NVIDIA #GPUs
#ZLUDA5 introduces "zoc" as an offline compiler for ZLUDA. #Zoc will take a NVIDIA PTX file as input and output the #LLVM intermediate representation (IR) and the #RDNA Assembly for the GPU in use generated in turn by the #ROCm compiler.
https://www.phoronix.com/news/ZLUDA-5-Released

#ZLUDA 5 Released With An Offline #Compiler For #CUDA On Non-NVIDIA #GPUs
#ZLUDA5 introduces "zoc" as an offline compiler for ZLUDA. #Zoc will take a NVIDIA PTX file as input and output the #LLVM intermediate representation (IR) and the #RDNA Assembly for the GPU in use generated in turn by the #ROCm compiler.
https://www.phoronix.com/news/ZLUDA-5-Released

#ZLUDA 5 Released With An Offline #Compiler For #CUDA On Non-NVIDIA #GPUs
#ZLUDA5 introduces "zoc" as an offline compiler for ZLUDA. #Zoc will take a NVIDIA PTX file as input and output the #LLVM intermediate representation (IR) and the #RDNA Assembly for the GPU in use generated in turn by the #ROCm compiler.
https://www.phoronix.com/news/ZLUDA-5-Released

Holy shit it lives: #ZLUDA ( #CUDA on non-NVIDIA GPUs ) got a update

https://github.com/vosen/ZLUDA/releases/tag/v5

ZLUDA Implements Kernel Cache Support To Help With Performance
#ZLUDA as the #opensource solution bringing #CUDA to non-NVIDIA hardware has been seeing a nice uptick in activity the past several months for its latest take on life. The latest feature merged to ZLUDA is the all-important kernel cache to help with performance.
https://www.phoronix.com/news/ZLUDA-Kernel-Cache

New #ZLUDA 5 Preview Released For #CUDA On Non-NVIDIA #GPU
For now this ability to run unmodified CUDA apps on non-#NVIDIA GPUs is focused on #AMD GPUs of the #Radeon RX 5000 series and newer, which is AMD Radeon GPUs with #ROCm. Besides CUDA code samples, GeekBench has been one of the early targets for testing.
https://www.phoronix.com/news/ZLUDA-5-preview.43

Uninstalling the #MicrosoftStore version of #Python and disabling #WindowsDefender so the #torch dlls properly install so I can run comfyui-zluda, a fork of #comfyui which uses #zluda which is a shim for #CUDA applications to use #AMD #HIP SDK so I can download #stableDiffusion and run it on my #Radeon #GPU because #Amuse won't let me generate porn.

#ai #llm #generativeAI

Uninstalling the #MicrosoftStore version of #Python and disabling #WindowsDefender so the #torch dlls properly install so I can run comfyui-zluda, a fork of #comfyui which uses #zluda which is a shim for #CUDA applications to use #AMD #HIP SDK so I can download #stableDiffusion and run it on my #Radeon #GPU because #Amuse won't let me generate porn.

#ai #llm #generativeAI

Uninstalling the #MicrosoftStore version of #Python and disabling #WindowsDefender so the #torch dlls properly install so I can run comfyui-zluda, a fork of #comfyui which uses #zluda which is a shim for #CUDA applications to use #AMD #HIP SDK so I can download #stableDiffusion and run it on my #Radeon #GPU because #Amuse won't let me generate porn.

#ai #llm #generativeAI

Uninstalling the #MicrosoftStore version of #Python and disabling #WindowsDefender so the #torch dlls properly install so I can run comfyui-zluda, a fork of #comfyui which uses #zluda which is a shim for #CUDA applications to use #AMD #HIP SDK so I can download #stableDiffusion and run it on my #Radeon #GPU because #Amuse won't let me generate porn.

#ai #llm #generativeAI

Uninstalling the #MicrosoftStore version of #Python and disabling #WindowsDefender so the #torch dlls properly install so I can run comfyui-zluda, a fork of #comfyui which uses #zluda which is a shim for #CUDA applications to use #AMD #HIP SDK so I can download #stableDiffusion and run it on my #Radeon #GPU because #Amuse won't let me generate porn.

#ai #llm #generativeAI

#ZLUDA Making Progress In 2025 On Bringing #CUDA To Non-NVIDIA #GPU
ZLUDA #opensource effort that started half-decade ago as drop-in CUDA implementation for #Intel GPUs and then for several years was funded by ##AMD as a CUDA implementation for #Radeon GPUs atop #ROCm and then open-sourced but then reverted has been continuing to push along a new path since last year. Current take on ZLUDA is a multi-vendor CUDA implementation for non-NVIDIA GPUs for #AI workloads & more.
https://www.phoronix.com/news/ZLUDA-Q2-2025-Update

#ZLUDA v4 Released For Initial #CUDA Support On Non-NVIDIA #GPU
Back in October ZLUDA decided to pursue a new life as an #opensource multi-GPU CUDA implementation with an emphasis on AI workloads. ZLUDA v4 was released as the first step to that new codebase. But before getting too excited, ZLUDA v4 support is quite limited and the only known intended program to be currently working is Geekbench with its CUDA benchmarks.
https://www.phoronix.com/news/ZLUDA-v4-Released

Another great news for this 2025: #ZLUDA lives

I'm eager for this library to take off. While NVIDIA dominates the market, there are decent iGPU like the Radeon 680M/780M/870M and Intel Arc on Ultra 200 Series.

https://www.phoronix.com/news/ZLUDA-v4-Released

#CUDA #NVIDIA #RTX #AI #ML #LM #LLM #LargeLanguageModels #AMD #Radeon #Intel #IntelArc #Battlemage #B570 #PC #PCHardware #Hardware #GPU #iGPU #dGPU

Another great news for this 2025: #ZLUDA lives

I'm eager for this library to take off. While NVIDIA dominates the market, there are decent iGPU like the Radeon 680M/780M/870M and Intel Arc on Ultra 200 Series.

https://www.phoronix.com/news/ZLUDA-v4-Released

#CUDA #NVIDIA #RTX #AI #ML #LM #LLM #LargeLanguageModels #AMD #Radeon #Intel #IntelArc #Battlemage #B570 #PC #PCHardware #Hardware #GPU #iGPU #dGPU

Another great news for this 2025: #ZLUDA lives

I'm eager for this library to take off. While NVIDIA dominates the market, there are decent iGPU like the Radeon 680M/780M/870M and Intel Arc on Ultra 200 Series.

https://www.phoronix.com/news/ZLUDA-v4-Released

#CUDA #NVIDIA #RTX #AI #ML #LM #LLM #LargeLanguageModels #AMD #Radeon #Intel #IntelArc #Battlemage #B570 #PC #PCHardware #Hardware #GPU #iGPU #dGPU

Another great news for this 2025: #ZLUDA lives

I'm eager for this library to take off. While NVIDIA dominates the market, there are decent iGPU like the Radeon 680M/780M/870M and Intel Arc on Ultra 200 Series.

https://www.phoronix.com/news/ZLUDA-v4-Released

#CUDA #NVIDIA #RTX #AI #ML #LM #LLM #LargeLanguageModels #AMD #Radeon #Intel #IntelArc #Battlemage #B570 #PC #PCHardware #Hardware #GPU #iGPU #dGPU

Another great news for this 2025: #ZLUDA lives

I'm eager for this library to take off. While NVIDIA dominates the market, there are decent iGPU like the Radeon 680M/780M/870M and Intel Arc on Ultra 200 Series.

https://www.phoronix.com/news/ZLUDA-v4-Released

#CUDA #NVIDIA #RTX #AI #ML #LM #LLM #LargeLanguageModels #AMD #Radeon #Intel #IntelArc #Battlemage #B570 #PC #PCHardware #Hardware #GPU #iGPU #dGPU

I just had a thought. if #ZLUDA gets good enough, would it be possible to inject it into a game and get #DLSS instead of #FSR on #AMD?

@dexter Here's the AMD w/ CUDA support/library/api/shim/translation/thing that came up the other day:

> ZLUDA is a drop-in replacement for CUDA on non-NVIDIA GPU. ZLUDA allows to run unmodified CUDA applications using non-NVIDIA GPUs with near-native performance.

- https://vosen.github.io/ZLUDA/blog/zludas-third-life/
- https://github.com/vosen/ZLUDA

#forDiscussion #gpu #ai #cuda #amd #nvidia #zluda #opencl #machineLearning

#ZLUDA Takes On Third Life: #OpenSource Multi-GPU #CUDA Implementation Focused On AI
ZLUDA is being rebuilt to focus on multi-GPU vendor support and will take a particular emphasis on #machinelearning / #AI type workloads. Previously ZLUDA was more focused on enabling professional creator workloads while now it will be more focused on CUDA-based AI/#ML software. The new ZLUDA code will be focused on #RDNA1 and newer support along with #ROCm 6.1+ compute stack support.
https://www.phoronix.com/news/ZLUDA-Third-Life

#AMD lawyers claw back #CUDA compatibility layer #ZLUDA

https://www.theregister.com/2024/08/09/amd_zluda_take_down/

#AMD asks #developer to take down #ZLUDA, dev vows to rebuild
Earlier this year, AMD quietly stopped funding ZLUDA, #opensource #CUDA #translationlayer project allowed programscompiled for #Nvidia #CUDA #GPU to run on on Radeon GPU with the #ROCm software stack.
"The code that was previously here has been taken down at AMD's request," the developer wrote. "The code was released with AMD's approval through an email. AMD's legal department now says it's not legally binding"
https://www.tomshardware.com/pc-components/gpus/amd-asks-developer-to-take-down-open-source-zluda-dev-vows-to-rebuild-his-project

Open-Source AMD CUDA Implementatio Taken Down

I honestly don’t know what AMD is thinking here, only a few months after this being released and making decent progress. (...)

#amd #cuda #legal #madness #opensource #zluda

https://taoofmac.com/space/links/2024/08/08/0910

iGPU Compute and LLMs on the AceMagic AM18

As part of my forays into LLMs and GPU compute “on the small” I’ve been playing around with the AceMagic AM18 in a few unusual ways. If you missed my initial impressions, you might want to check them out first.(...)

#am18 #amd #amdgpu #gpu #hardware #homelab #llm #ml #ollama #proxmox #review #ryzen #vllm #zluda

https://taoofmac.com/space/blog/2024/04/13/2100

Das Neuste aus der OpenSource-Community könnt ihr direkt bei uns nachhören. U.a. zu #ZLUDA – #AMD hat leise ein #CUDA Translation Layer finanziert

https://theradio.cc/blog/2024/03/11/ll279-quack/

#opensource #nvidia #llm #ai #ki #machinelearning #freesoftware #floss #foss #protocols

Nvidia CUDA 的 EULA 禁止跑在模擬層上面

在 Hacker News 上看到「Nvidia bans using translation layers for CUDA software to run on other chips (tomshardware.com)」這個討論串，原報導「Nvidia bans using translation layers for CUDA software — previously the prohibition

https://blog.gslin.org/archives/2024/03/07/11689/nvidia-cuda-%e7%9a%84-eula-%e7%a6%81%e6%ad%a2%e8%b7%91%e5%9c%a8%e6%a8%a1%e6%93%ac%e5%b1%a4%e4%b8%8a%e9%9d%a2/

#Computer #Murmuring #Software #amd #cuda #eula #intel #nvidia #zluda

Finally i don't have to trouble myself buying nvidia gpu's just because the renderer everyone uses at the office requires CUDA.

And also i had a lot of pain trying to use ROCm on linux (debian).

https://www.phoronix.com/review/radeon-cuda-zluda

#amd #computing #cuda #zluda

讓 <del>Intel</del>AMD GPU 直接跑 CUDA 程式的 ZLUDA

先前提過「在 Intel 內顯上面直接跑 CUDA 程式的 ZLUDA」，結果後來事情大翻轉，AMD 跑去贊助專案，變成支援 AMD GPU 了：「AMD Quietly Funded A Drop-In CUDA Implementation Built On ROCm: It's Now Open-Source」，專案在 GitHub 的 vosen

https://blog.gslin.org/archives/2024/02/13/11655/%e8%ae%93-intelamd-gpu-%e7%9b%b4%e6%8e%a5%e8%b7%91-cuda-%e7%a8%8b%e5%bc%8f%e7%9a%84-zluda/

#Computer #Murmuring #Software #amd #blender #cuda #driver #intel #nvidia #open #opencl #rocm #source #zluda

Wow, this seems quite amazing, especially with performance sometimes being better than pure Rocm:
https://www.phoronix.com/review/radeon-cuda-zluda

#rocm #rust #zluda

AMD Funded A Drop-In CUDA Implementation

The craziest thing for me is that they stopped funding it. You would expect them to want a viable drop-in alternative to their biggest rival’s API to ease migration, but no, they let it go.(...)

#machinelearning #opensource #ai #amd #gpu #zluda #cuda #hpc

https://taoofmac.com/space/links/2024/02/12/1930