LICENSE
README.md
THIRDPARTYNOTICES
pyproject.toml
python/tokenspeed_kernel_amd/__init__.py
python/tokenspeed_kernel_amd/_scheduling.py
python/tokenspeed_kernel_amd/_triton.py
python/tokenspeed_kernel_amd/sched_barrier.ll
python/tokenspeed_kernel_amd.egg-info/PKG-INFO
python/tokenspeed_kernel_amd.egg-info/SOURCES.txt
python/tokenspeed_kernel_amd.egg-info/dependency_links.txt
python/tokenspeed_kernel_amd.egg-info/requires.txt
python/tokenspeed_kernel_amd.egg-info/top_level.txt
python/tokenspeed_kernel_amd/ops/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/_common.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/dsa/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/dsa/attention.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/dsa/indexing.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/dsa/sparse_mla.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/dsa/standard_cache_logits.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/dsv4/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/dsv4/decode.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/dsv4/indexer.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/dsv4/prefill.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/dsv41/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/dsv41/indexer.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/dsv41/selected.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/kda/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/kda/attn_res.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/kda/decode.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/kda/prefill.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/mha/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/mha/decode.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/mha/prefill.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/mla/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/mla/_common.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/mla/decode.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/mla/extend.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/mla/kv_pack.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/mla/normalize_project_query.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/mla/prefill.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/mla/project_value.py
python/tokenspeed_kernel_amd/ops/gfx1250/attention/mla/reduce_project_value.py
python/tokenspeed_kernel_amd/ops/gfx1250/gemm/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/gemm/fp16/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/gemm/fp16/linear_attnres_partials_gfx1250.py
python/tokenspeed_kernel_amd/ops/gfx1250/gemm/fp16/mm.py
python/tokenspeed_kernel_amd/ops/gfx1250/gemm/mxfp8/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/gemm/mxfp8/decode_mm.py
python/tokenspeed_kernel_amd/ops/gfx1250/moe/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/moe/_common.py
python/tokenspeed_kernel_amd/ops/gfx1250/moe/fp16/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/moe/fp16/_common.py
python/tokenspeed_kernel_amd/ops/gfx1250/moe/fp16/latent_input_decode.py
python/tokenspeed_kernel_amd/ops/gfx1250/moe/fp16/latent_input_largem.py
python/tokenspeed_kernel_amd/ops/gfx1250/moe/mxfp4/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/moe/mxfp4/_common.py
python/tokenspeed_kernel_amd/ops/gfx1250/moe/mxfp4/decode.py
python/tokenspeed_kernel_amd/ops/gfx1250/moe/mxfp4/fused.py
python/tokenspeed_kernel_amd/ops/gfx1250/moe/mxfp4/routing.py
python/tokenspeed_kernel_amd/ops/gfx1250/moe/mxfp4/weight_preprocess.py
python/tokenspeed_kernel_amd/ops/gfx1250/sampling/__init__.py
python/tokenspeed_kernel_amd/ops/gfx1250/sampling/argmax.py
python/tokenspeed_kernel_amd/ops/gfx950/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/mhc.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/_common.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsa/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsa/attention.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsa/indexing.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsa/sparse_mla.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsa/standard_cache_logits.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsv4/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsv4/decode.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsv4/indexer.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsv4/prefill.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsv4/sparse_prefill.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsv41/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsv41/indexer.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/dsv41/selected.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/kda/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/kda/attn_res.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/kda/decode.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/kda/prefill.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mha/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mha/decode.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mha/extend.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mha/prefill.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mla/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mla/decode.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mla/extend.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mla/kv_pack.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mla/normalize_project_query.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mla/prefill.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mla/prefill_8wave.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mla/project_value.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/mla/reduce_project_value.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/rmha/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/rmha/decode.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/rmha/extend.py
python/tokenspeed_kernel_amd/ops/gfx950/attention/rmha/prefill.py
python/tokenspeed_kernel_amd/ops/gfx950/gemm/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/gemm/fp16/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/gemm/fp16/largem.py
python/tokenspeed_kernel_amd/ops/gfx950/gemm/fp16/linear_attnres_partials_gfx950.py
python/tokenspeed_kernel_amd/ops/gfx950/gemm/fp16/mm.py
python/tokenspeed_kernel_amd/ops/gfx950/gemm/fp16/rmsnorm_linear_add.py
python/tokenspeed_kernel_amd/ops/gfx950/gemm/mxfp8/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/gemm/mxfp8/mm.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/_common.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/sqrt_softplus_topk.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/_grid.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/latent_input_decode.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/latent_input_largem.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/latent_input_mediumm.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/latent_input_small_batch.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/moe.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/moe_align_device.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/moe_align_fused.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/stage1_kernel.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/stage1_splitk_kernel.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/stage2_decode_kernel.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/stage2_kernel.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp16/warp_decode_gluon_kernel.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp8/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp8/exact_mfma.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/fp8/warp_decode.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/decode_common.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/decode_stage1.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/decode_stage2.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/expert_mesh.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/latent_shared_decode.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/moe_sorting.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/mxfp8_gemm.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/mxfp8_quantize.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/n16_weights.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/prefill_mxfp8.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/prefill_stage1.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/prefill_stage2.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/preprocess.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/quantize_gluon.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/routing.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/scale.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/scale_layout.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/situ_decode.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/situ_grouped.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/weight_preprocess.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/_common.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/_layouts.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/gemm_api.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/launch.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/medium_decode.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/moe.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/pipelined_kernel.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/pipelined_program.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/quantize.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/routing.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/tuning.py
python/tokenspeed_kernel_amd/ops/gfx950/moe/mxfp4/fused/warp_decode.py
python/tokenspeed_kernel_amd/ops/gfx950/sampling/__init__.py
python/tokenspeed_kernel_amd/ops/gfx950/sampling/argmax.py