mi50
Here are 26 public repositories matching this topic...
Strata 0.2.3 (engine 0.1.48) for AMD gfx906 (MI50/MI60/Radeon VII): rocBLAS solution table, Qwen3.6/Ornith engine restored, DFlash2 build fix, web configurator, portable (no AVX-VNNI needed). Docker: xxdoman/strata-mi50:latest
-
Updated
Nov 1, 2026 - Shell
FlashAttention-style custom attention backend for vLLM on AMD MI50/MI60/Radeon VII (gfx906). Downstream fork of mixa3607/ML-gfx906 with replacement HIP kernels and a vllm.general_plugins entry point.
-
Updated
Apr 22, 2026 - Python
Open-source local AI server configs, GFX906 runtime maintenance, reproducible benchmarks, and QC methods for affordable AI research infrastructure.
-
Updated
Jul 1, 2026 - Python
Closed. AMD Instinct MI50/MI60 (gfx906, ROCm) port of NInfer. Works on one or two cards but runs slower than llama.cpp on this hardware; the README explains why and links better options.
-
Updated
Oct 8, 2026 - C++
Modern AMD/NVIDIA GPU bring-up for PowerPC Mac OS X (Tiger/Leopard) as Darwin KEXTs. Big-endian. MI50 compute accelerator on a Power Mac G5.
-
Updated
Aug 6, 2026 - C
ROCm/Unsloth/bitsandbytes 4-bit lab and VRAM benchmarks for AMD MI50/gfx906 LLM fine-tuning
-
Updated
Jun 3, 2026 - Python
Strata fork for AMD Radeon Instinct MI50 / MI60 (gfx906, wave64): HIP build, exact decode speed-ups, 4-GPU layer split, 12 parallel requests at 256K context
-
Updated
Nov 8, 2026 - C++
Local Paperless-ngx + Ollama integration for AI-based document titles, correspondents, document types, tags, review workflows, and backfill.
-
Updated
Apr 11, 2026 - Python
openPangu Flash92 — C++/CUDA & C++/HIP ROCm Engines. Fully resident, Blackwell tensor cores, no Python. Powered by openPangu.
-
Updated
Sep 13, 2026 - HIP
Run the ComfyUI Aki bundle (秋叶整合包, 绘世 launcher) on an AMD Instinct MI50 / Radeon Pro VII (gfx906) under Windows via ZLUDA — cu118 torch pinning, a drop-in compatibility patch, measured FP16 / attention tuning, and a troubleshooting guide.
-
Updated
Nov 6, 2026 - Python
Add this topic to your repo
To associate your repository with the mi50 topic, visit your repo's landing page and select "manage topics."