MI210 (gfx90a) Qwen3.8-27B W4A16 + DFlash2 inference acceleration: int8 MFMA Marlin kernels, SGLang integration, benchmarks and research notes
-
Updated
Sep 15, 2026 - Python
MI210 (gfx90a) Qwen3.8-27B W4A16 + DFlash2 inference acceleration: int8 MFMA Marlin kernels, SGLang integration, benchmarks and research notes
To associate your repository with the amd-mi210 topic, visit your repo's landing page and select "manage topics."