TensorRT-LLM quantization - Hardware Support Matrix (per-GPU-arch Y/. cells)

The per-architecture support matrix, extracted verbatim. Blackwell(sm120): NVFP4 Y, MXFP4 Y, FP8(per tensor) Y, FP8(block scaling) '.', FP8(rowwise) '.', FP8 KV Y, NVFP4 KV '.', all four AWQ/GPTQ cells '.'. Blackwell(sm100/103): NVFP4 Y, MXFP4 Y, FP8(per tensor) Y, FP8(block scaling) Y, FP8(rowwise) '.', FP8 KV Y, NVFP4 KV Y, W4A8 AWQ Y, W4A16 AWQ Y, W4A8 GPTQ Y, W4A16 GPTQ Y. Rubin(sm107): same as sm100/103 except NVFP4 KV '.'. Hopper: NVFP4 '.', MXFP4 '.', FP8(per tensor) Y, FP8(block scaling) Y, FP8(rowwise) Y, FP8 KV Y, NVFP4 KV '.', AWQ and GPTQ both Y. Ada Lovelace: FP8(per tensor) Y, FP8 KV Y, AWQ and GPTQ both Y, NVFP4/MXFP4 '.'. Ampere: FP8(per tensor) '.', FP8 KV Y, W4A16 AWQ Y, W4A16 GPTQ Y, everything else '.'. Key doc footnote: 'FP8 block wise scaling GEMM kernels for sm100/103/107 are using MXFP8 recipe (E4M3 act/weight and UE8M0 act/weight scale), which is slightly different from SM90 FP8 recipe (E4M3 act/weight and FP32 act/weight scale).' This is the direct confirmation that sm120 has no FP8 block-scaled GEMM while sm100/103/107 do. Accessed 2026-10-04.