.gitignore
.pre-commit-config.yaml
AGENTS.md
LICENSE
MANIFEST.in
NOTICE
README.md
pyproject.toml
reference-dependencies.json
licenses/AUTHORS.flash-attention.txt
licenses/LICENSE.cudnn-frontend.txt
licenses/LICENSE.deepep.txt
licenses/LICENSE.deepgemm.txt
licenses/LICENSE.fast-cu.txt
licenses/LICENSE.flash-attention.txt
licenses/LICENSE.flashinfer.txt
licenses/LICENSE.flashmla.txt
licenses/LICENSE.msa.txt
licenses/LICENSE.sglang.txt
scripts/install_reference_dependencies.py
tests/check_distribution.py
tests/test_bench_suite_ab.py
tests/test_bench_suite_remote.py
tests/test_bench_suite_run.py
tests/test_correctness.py
tests/test_descriptor_scalar_access.py
tests/test_low_level_ir.py
tests/test_mla_numerics.py
tests/test_registry.py
tests/test_rubin_fp4_descriptor.py
tests/test_runner_cuda_compile.py
tests/test_runner_interrupts.py
tests/test_tirx_lite_api.py
tests/lint/check_fast_topk_clusters_no_tile.py
tests/lint/check_filtered_topk_no_tile.py
tests/lint/check_flashinfer_rmsnorm_no_tile.py
tests/lint/check_gdn_decode_bf16_ilp4_no_tile.py
tests/lint/check_gdn_decode_bf16_wide_vec_t1_no_tile.py
tests/lint/check_gdn_decode_fp32_mtp_warp_no_tile.py
tests/lint/check_license_headers.py
tests/lint/check_moe_blockscaled_grouped_gemm_dglu_dbias_no_tile.py
tests/lint/check_moe_grouped_gemm_dglu_dbias_no_tile.py
tests/lint/check_radix_topk_multi_cta_no_tile.py
tests/lint/check_radix_topk_single_cta_no_tile.py
tests/lint/check_readme_kernels.py
tests/lint/check_stable_sort_topk_by_value_no_tile.py
tirx_kernels/__init__.py
tirx_kernels/_protocol.py
tirx_kernels/reference_requirements.py
tirx_kernels/registry.py
tirx_kernels/runner.py
tirx_kernels.egg-info/PKG-INFO
tirx_kernels.egg-info/SOURCES.txt
tirx_kernels.egg-info/dependency_links.txt
tirx_kernels.egg-info/requires.txt
tirx_kernels.egg-info/scm_file_list.json
tirx_kernels.egg-info/scm_version.json
tirx_kernels.egg-info/top_level.txt
tirx_kernels/bench/__init__.py
tirx_kernels/bench/__main__.py
tirx_kernels/bench_suite/README.md
tirx_kernels/bench_suite/__init__.py
tirx_kernels/bench_suite/__main__.py
tirx_kernels/bench_suite/_remote_shim.py
tirx_kernels/bench_suite/ab.py
tirx_kernels/bench_suite/baseline.json
tirx_kernels/bench_suite/baseline.md
tirx_kernels/bench_suite/baseline_view.py
tirx_kernels/bench_suite/impls.py
tirx_kernels/bench_suite/promote_baseline.py
tirx_kernels/bench_suite/provenance.py
tirx_kernels/bench_suite/ratio_diff.py
tirx_kernels/bench_suite/remote.py
tirx_kernels/bench_suite/run.py
tirx_kernels/bench_suite/config/gemm/allgather_gemm.yaml
tirx_kernels/bench_suite/config/gemm/fp16_bf16_gemm.yaml
tirx_kernels/bench_suite/config/gemm/gemm_reduce_scatter.yaml
tirx_kernels/bench_suite/config/gemm/nvfp4_gemm.yaml
tirx_kernels/bench_suite/config/kda/kda_backward_packed.yaml
tirx_kernels/bench_suite/config/kda/kda_decode_multishape.yaml
tirx_kernels/bench_suite/config/kda/kda_forward_portfolio_multishape.yaml
tirx_kernels/bench_suite/config/mla/mla_dsv4_multishape.yaml
tirx_kernels/bench_suite/config/moe/alphamoe_fp8_blockscale_qwen3next.yaml
tirx_kernels/bench_suite/config/msa/msa_decode_multishape.yaml
tirx_kernels/bench_suite/config/msa/msa_prefill_multishape.yaml
tirx_kernels/bench_suite/config/norm/rmsnorm.yaml
tirx_kernels/bench_suite/config/ported/cudnn/cudnn_sm100_bsa_backward_blk128.yaml
tirx_kernels/bench_suite/config/ported/cudnn/amax/cudnn_sm100_dense_blockscaled_gemm_persistent_amax.yaml
tirx_kernels/bench_suite/config/ported/cudnn/bsa/cudnn_sm100_bsa_backward_blk64.yaml
tirx_kernels/bench_suite/config/ported/cudnn/bsa/cudnn_sm100_bsa_forward_blk128.yaml
tirx_kernels/bench_suite/config/ported/cudnn/bsa/cudnn_sm100_bsa_forward_blk64.yaml
tirx_kernels/bench_suite/config/ported/cudnn/bsa/cudnn_sm100_bsa_forward_combine_blk64.yaml
tirx_kernels/bench_suite/config/ported/cudnn/csa/cudnn_sm100_csa_compressor_fwd.yaml
tirx_kernels/bench_suite/config/ported/cudnn/dglu/cudnn_sm100_moe_blockscaled_grouped_gemm_dglu_dbias.yaml
tirx_kernels/bench_suite/config/ported/cudnn/dglu/cudnn_sm100_moe_grouped_gemm_dglu_dbias.yaml
tirx_kernels/bench_suite/config/ported/cudnn/dsa/cudnn_sm100_dsa_sparse_attention_backward.yaml
tirx_kernels/bench_suite/config/ported/cudnn/dsrelu/cudnn_sm100_dense_blockscaled_gemm_persistent_dsrelu_quant.yaml
tirx_kernels/bench_suite/config/ported/cudnn/flex_attention/cudnn_sm100_flex_attention_backward.yaml
tirx_kernels/bench_suite/config/ported/cudnn/flex_attention/cudnn_sm100_flex_attention_forward_hd256.yaml
tirx_kernels/bench_suite/config/ported/cudnn/flex_attention/cudnn_sm103_flex_attention_forward.yaml
tirx_kernels/bench_suite/config/ported/cudnn/linear_attention/cudnn_sm100_gdn2_bprop_f16.yaml
tirx_kernels/bench_suite/config/ported/cudnn/linear_attention/cudnn_sm100_gdn2_prefill_f16.yaml
tirx_kernels/bench_suite/config/ported/cudnn/linear_attention/cudnn_sm100_gdn2_recompute_f16.yaml
tirx_kernels/bench_suite/config/ported/cudnn/linear_attention/cudnn_sm100_gdn_bprop_f16.yaml
tirx_kernels/bench_suite/config/ported/cudnn/linear_attention/cudnn_sm100_gdn_prefill_f16.yaml
tirx_kernels/bench_suite/config/ported/cudnn/linear_attention/cudnn_sm100_gdn_recompute_f16.yaml
tirx_kernels/bench_suite/config/ported/cudnn/linear_attention/cudnn_sm100_kda_bprop_f16.yaml
tirx_kernels/bench_suite/config/ported/cudnn/proj_rope_mxfp8/cudnn_sm100_gemm_proj_rope_mxfp8_bf16in.yaml
tirx_kernels/bench_suite/config/ported/cudnn/proj_rope_mxfp8/cudnn_sm100_gemm_proj_rope_mxfp8_mxfp8in.yaml
tirx_kernels/bench_suite/config/ported/cudnn/srelu/cudnn_sm100_dense_blockscaled_gemm_persistent_srelu_quant.yaml
tirx_kernels/bench_suite/config/ported/cudnn/swiglu/cudnn_sm100_dense_blockscaled_gemm_persistent_swiglu_interleaved_quant.yaml
tirx_kernels/bench_suite/config/ported/cudnn/swiglu/cudnn_sm100_dense_gemm_persistent_swiglu.yaml
tirx_kernels/bench_suite/config/ported/deepep/deepep_combine.yaml
tirx_kernels/bench_suite/config/ported/deepep/deepep_dispatch.yaml
tirx_kernels/bench_suite/config/ported/deepgemm/deepgemm_sm100_fp4_mqa_logits.yaml
tirx_kernels/bench_suite/config/ported/deepgemm/deepgemm_sm100_fp4_paged_mqa_logits.yaml
tirx_kernels/bench_suite/config/ported/deepgemm/deepgemm_sm100_fp8_bmm.yaml
tirx_kernels/bench_suite/config/ported/deepgemm/deepgemm_sm100_fp8_gemm_1d1d.yaml
tirx_kernels/bench_suite/config/ported/deepgemm/deepgemm_sm100_fp8_mqa_logits.yaml
tirx_kernels/bench_suite/config/ported/deepgemm/deepgemm_sm100_fp8_paged_mqa_logits.yaml
tirx_kernels/bench_suite/config/ported/deepgemm/deepgemm_sm100_k_grouped_fp8_gemm_contiguous.yaml
tirx_kernels/bench_suite/config/ported/deepgemm/deepgemm_sm100_m_grouped_fp8_gemm_contiguous.yaml
tirx_kernels/bench_suite/config/ported/deepgemm/deepgemm_sm100_m_grouped_fp8_gemm_masked.yaml
tirx_kernels/bench_suite/config/ported/deepgemm/deepgemm_sm100_tf32_hc_prenorm_gemm.yaml
tirx_kernels/bench_suite/config/ported/deepgemm/sm100_fp8_fp4_mega_moe.yaml
tirx_kernels/bench_suite/config/ported/fastcu/fastcu_nvfp4_gemm_gb300.yaml
tirx_kernels/bench_suite/config/ported/flashattention/flash_attention4.yaml
tirx_kernels/bench_suite/config/ported/flashattention/flash_attention4_fp4.yaml
tirx_kernels/bench_suite/config/ported/flashattention/flash_attention_backward_sm100.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/activation/act_and_mul.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/activation/silu_and_mul_nvfp4_experts_quantize.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/cascade/merge_state.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/fused_moe/blockscaled_contiguous_gather_grouped_gemm_swiglu_fusion_rubin.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/gdn_decode/gdn_decode_bf16_ilp4.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/gdn_decode/gdn_decode_bf16_wide_vec_mtp.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/gdn_decode/gdn_decode_bf16_wide_vec_t1.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/gdn_decode/gdn_decode_fp32_mtp_warp.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/gdn_prefill/gdn_cp_prefill_sm100.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/gdn_prefill/gdn_prefill_sm100.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/gemm/bmm_fp8_rubin.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/gemm/dense_blockscaled_gemm_sm107.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/gemm/grouped_gemm_masked_rubin.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/kda/recurrent_kda_decode_grouped.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/kda/recurrent_kda_decode_one_warp.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/mamba/selective_state_update_mtp_horizontal.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/mamba/selective_state_update_mtp_simple.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/mamba/selective_state_update_mtp_vertical.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/mamba/selective_state_update_stp_horizontal.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/mamba/selective_state_update_stp_simple.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/mamba/selective_state_update_stp_vertical.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/norm/flashinfer_add_rmsnorm_fp4quant.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/norm/flashinfer_fused_add_rmsnorm.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/norm/flashinfer_fused_add_rmsnorm_quant.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/norm/flashinfer_fused_dit_layernorm.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/norm/flashinfer_layernorm.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/norm/flashinfer_qk_rmsnorm.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/norm/flashinfer_rmsnorm.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/norm/flashinfer_rmsnorm_fp4quant.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/norm/flashinfer_rmsnorm_quant.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/quantization/mxfp4_quantize.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/quantization/mxfp8_quantize.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/quantization/nvfp4_quantize.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/quantization/nvfp4_quantize_per_token.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/topk/fast_topk_clusters.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/topk/filtered_topk.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/topk/radix_topk_multi_cta.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/topk/radix_topk_single_cta.yaml
tirx_kernels/bench_suite/config/ported/flashinfer/topk/stable_sort_topk_by_value.yaml
tirx_kernels/bench_suite/config/ported/flashmla/flash_mla_sparse_fwd.yaml
tirx_kernels/bench_suite/config/ported/flashmla/sparse_flashmla_decode_head64.yaml
tirx_kernels/bench_suite/config/ported/flashmla/sparse_flashmla_prefill_head128_phase1.yaml
tirx_kernels/bench_suite/config/ported/flashmla/sparse_flashmla_prefill_head128_small_topk_phase1.yaml
tirx_kernels/bench_suite/config/ported/flashmla/sparse_flashmla_prefill_head64_phase1.yaml
tirx_kernels/bench_suite/config/ported/msa/msa_sparse_atten_fwd_combine_sm100.yaml
tirx_kernels/bench_suite/config/ported/msa/msa_sparse_atten_fwd_nvfp4_kv_sm100.yaml
tirx_kernels/bench_suite/config/ported/msa/msa_sparse_atten_fwd_sm100.yaml
tirx_kernels/bench_suite/config/ported/msa/msa_sparse_prepare_flat_schedule_sm100.yaml
tirx_kernels/bench_suite/config/ported/msa/msa_sparse_prepare_fwd_split_atomic_sm100.yaml
tirx_kernels/bench_suite/config/vsa/vsa_multishape.yaml
tirx_kernels/gemm/README.md
tirx_kernels/gemm/__init__.py
tirx_kernels/gemm/allgather_gemm.py
tirx_kernels/gemm/fp16_bf16_gemm.py
tirx_kernels/gemm/gemm_reduce_scatter.py
tirx_kernels/gemm/nvfp4_gemm.py
tirx_kernels/gemm/utils/__init__.py
tirx_kernels/gemm/utils/_baselines.py
tirx_kernels/gemm/utils/_model_shapes.py
tirx_kernels/gemm/utils/_runtime.py
tirx_kernels/gemm/utils/_specialize.py
tirx_kernels/kda/__init__.py
tirx_kernels/kda/_kda_forward_schedule.py
tirx_kernels/kda/kda_backward_packed.py
tirx_kernels/kda/kda_decode_multishape.py
tirx_kernels/kda/kda_forward_portfolio_multishape.py
tirx_kernels/mla/__init__.py
tirx_kernels/mla/mla_dsv4_multishape.py
tirx_kernels/moe/__init__.py
tirx_kernels/moe/alphamoe_fp8_blockscale_qwen3next.py
tirx_kernels/msa/__init__.py
tirx_kernels/msa/msa_decode_multishape.py
tirx_kernels/msa/msa_prefill_multishape.py
tirx_kernels/norm/__init__.py
tirx_kernels/norm/rmsnorm.py
tirx_kernels/ported/__init__.py
tirx_kernels/ported/cudnn/__init__.py
tirx_kernels/ported/cudnn/_reference.py
tirx_kernels/ported/cudnn/amax/__init__.py
tirx_kernels/ported/cudnn/amax/dense_blockscaled_gemm_persistent_amax.py
tirx_kernels/ported/cudnn/bsa/__init__.py
tirx_kernels/ported/cudnn/bsa/block_sparse_attention_backward_sm100_blk128.py
tirx_kernels/ported/cudnn/bsa/block_sparse_attention_backward_sm100_blk64.py
tirx_kernels/ported/cudnn/bsa/block_sparse_attention_forward_combine_sm100_blk64.py
tirx_kernels/ported/cudnn/bsa/block_sparse_attention_forward_sm100_blk128.py
tirx_kernels/ported/cudnn/bsa/block_sparse_attention_forward_sm100_blk64.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_backward_sm100_blk128/__init__.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_backward_sm100_blk128/data.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_backward_sm100_blk128/kernel.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_backward_sm100_blk128/reference.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_backward_sm100_blk128/spec.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_backward_sm100_blk64/__init__.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_backward_sm100_blk64/data.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_backward_sm100_blk64/kernel.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_backward_sm100_blk64/reference.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_backward_sm100_blk64/spec.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_forward_sm100_blk64/__init__.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_forward_sm100_blk64/data.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_forward_sm100_blk64/kernel.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_forward_sm100_blk64/reference.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_forward_sm100_blk64/source_kernel.py
tirx_kernels/ported/cudnn/bsa/_block_sparse_attention_forward_sm100_blk64/spec.py
tirx_kernels/ported/cudnn/csa/__init__.py
tirx_kernels/ported/cudnn/csa/compressor_fwd_sm100.py
tirx_kernels/ported/cudnn/dglu/__init__.py
tirx_kernels/ported/cudnn/dglu/moe_blockscaled_grouped_gemm_dglu_dbias.py
tirx_kernels/ported/cudnn/dglu/moe_grouped_gemm_dglu_dbias.py
tirx_kernels/ported/cudnn/dglu/_moe_blockscaled_grouped_gemm_dglu_dbias/__init__.py
tirx_kernels/ported/cudnn/dglu/_moe_blockscaled_grouped_gemm_dglu_dbias/data.py
tirx_kernels/ported/cudnn/dglu/_moe_blockscaled_grouped_gemm_dglu_dbias/kernel.py
tirx_kernels/ported/cudnn/dglu/_moe_blockscaled_grouped_gemm_dglu_dbias/spec.py
tirx_kernels/ported/cudnn/dglu/_moe_grouped_gemm_dglu_dbias/__init__.py
tirx_kernels/ported/cudnn/dglu/_moe_grouped_gemm_dglu_dbias/data.py
tirx_kernels/ported/cudnn/dglu/_moe_grouped_gemm_dglu_dbias/kernel.py
tirx_kernels/ported/cudnn/dglu/_moe_grouped_gemm_dglu_dbias/spec.py
tirx_kernels/ported/cudnn/dsa/__init__.py
tirx_kernels/ported/cudnn/dsa/sparse_attention_backward.py
tirx_kernels/ported/cudnn/dsa/_sparse_attention_backward/__init__.py
tirx_kernels/ported/cudnn/dsa/_sparse_attention_backward/data.py
tirx_kernels/ported/cudnn/dsa/_sparse_attention_backward/kernel.py
tirx_kernels/ported/cudnn/dsa/_sparse_attention_backward/reference.py
tirx_kernels/ported/cudnn/dsa/_sparse_attention_backward/spec.py
tirx_kernels/ported/cudnn/dsrelu/__init__.py
tirx_kernels/ported/cudnn/dsrelu/dense_blockscaled_gemm_persistent_dsrelu_quant.py
tirx_kernels/ported/cudnn/flex_attention/__init__.py
tirx_kernels/ported/cudnn/flex_attention/flex_attention_backward_sm100.py
tirx_kernels/ported/cudnn/flex_attention/forward_hd256_sm100.py
tirx_kernels/ported/cudnn/flex_attention/forward_sm103.py
tirx_kernels/ported/cudnn/flex_attention/_flex_attention_backward_sm100/__init__.py
tirx_kernels/ported/cudnn/flex_attention/_flex_attention_backward_sm100/data.py
tirx_kernels/ported/cudnn/flex_attention/_flex_attention_backward_sm100/kernel.py
tirx_kernels/ported/cudnn/flex_attention/_flex_attention_backward_sm100/kernel_2cta.py
tirx_kernels/ported/cudnn/linear_attention/__init__.py
tirx_kernels/ported/cudnn/linear_attention/_frost.py
tirx_kernels/ported/cudnn/linear_attention/gdn2_bprop_f16.py
tirx_kernels/ported/cudnn/linear_attention/gdn2_prefill_f16.py
tirx_kernels/ported/cudnn/linear_attention/gdn2_recompute_f16.py
tirx_kernels/ported/cudnn/linear_attention/gdn_bprop_f16.py
tirx_kernels/ported/cudnn/linear_attention/gdn_prefill_f16.py
tirx_kernels/ported/cudnn/linear_attention/gdn_recompute_f16.py
tirx_kernels/ported/cudnn/linear_attention/kda_bprop_f16.py
tirx_kernels/ported/cudnn/proj_rope_mxfp8/__init__.py
tirx_kernels/ported/cudnn/proj_rope_mxfp8/gemm_proj_rope_mxfp8_bf16in.py
tirx_kernels/ported/cudnn/proj_rope_mxfp8/gemm_proj_rope_mxfp8_mxfp8in.py
tirx_kernels/ported/cudnn/srelu/__init__.py
tirx_kernels/ported/cudnn/srelu/dense_blockscaled_gemm_persistent_srelu_quant.py
tirx_kernels/ported/cudnn/swiglu/__init__.py
tirx_kernels/ported/cudnn/swiglu/dense_blockscaled_gemm_persistent_swiglu_interleaved_quant.py
tirx_kernels/ported/cudnn/swiglu/dense_gemm_persistent_swiglu.py
tirx_kernels/ported/deepep/__init__.py
tirx_kernels/ported/deepep/combine.py
tirx_kernels/ported/deepep/dispatch.py
tirx_kernels/ported/deepep/utils/__init__.py
tirx_kernels/ported/deepep/utils/_buffer.py
tirx_kernels/ported/deepep/utils/_runtime.py
tirx_kernels/ported/deepgemm/__init__.py
tirx_kernels/ported/deepgemm/fp8_bmm.py
tirx_kernels/ported/deepgemm/fp8_gemm_1d1d.py
tirx_kernels/ported/deepgemm/k_grouped_fp8_gemm_contiguous.py
tirx_kernels/ported/deepgemm/m_grouped_fp8_gemm_contiguous.py
tirx_kernels/ported/deepgemm/m_grouped_fp8_gemm_masked.py
tirx_kernels/ported/deepgemm/mqa_logits_fp4.py
tirx_kernels/ported/deepgemm/mqa_logits_fp8.py
tirx_kernels/ported/deepgemm/paged_mqa_logits_fp4.py
tirx_kernels/ported/deepgemm/paged_mqa_logits_fp8.py
tirx_kernels/ported/deepgemm/sm100_fp8_fp4_mega_moe.py
tirx_kernels/ported/deepgemm/tf32_hc_prenorm_gemm.py
tirx_kernels/ported/deepgemm/_sglang_cutedsl/__init__.py
tirx_kernels/ported/deepgemm/_sglang_cutedsl/cutedsl_fp8_paged_mqa_logits.py
tirx_kernels/ported/deepgemm/_sglang_cutedsl/cutedsl_paged_mqa_logits.py
tirx_kernels/ported/deepgemm/_sm100_fp8_fp4_gemm_1d1d/__init__.py
tirx_kernels/ported/deepgemm/_sm100_fp8_fp4_gemm_1d1d/data.py
tirx_kernels/ported/deepgemm/_sm100_fp8_fp4_gemm_1d1d/kernel.py
tirx_kernels/ported/deepgemm/_sm100_fp8_fp4_gemm_1d1d/spec.py
tirx_kernels/ported/deepgemm/_sm100_fp8_fp4_mega_moe/__init__.py
tirx_kernels/ported/deepgemm/_sm100_fp8_fp4_mega_moe/data.py
tirx_kernels/ported/deepgemm/_sm100_fp8_fp4_mega_moe/kernel.py
tirx_kernels/ported/deepgemm/_sm100_fp8_fp4_mega_moe/spec.py
tirx_kernels/ported/fastcu/__init__.py
tirx_kernels/ported/fastcu/nvfp4_gemm_gb300.py
tirx_kernels/ported/fastcu/nvfp4_gemm_gb300_reference.cu
tirx_kernels/ported/flashattention/__init__.py
tirx_kernels/ported/flashattention/flash_attention4.py
tirx_kernels/ported/flashattention/flash_attention4_fp4.py
tirx_kernels/ported/flashattention/flash_attention_backward.py
tirx_kernels/ported/flashattention/flash_attention_backward_sm100_sketch.md
tirx_kernels/ported/flashinfer/__init__.py
tirx_kernels/ported/flashinfer/activation/__init__.py
tirx_kernels/ported/flashinfer/activation/act_and_mul.py
tirx_kernels/ported/flashinfer/activation/silu_and_mul_nvfp4_experts_quantize.py
tirx_kernels/ported/flashinfer/cascade/__init__.py
tirx_kernels/ported/flashinfer/cascade/merge_state.py
tirx_kernels/ported/flashinfer/fused_moe/__init__.py
tirx_kernels/ported/flashinfer/fused_moe/blockscaled_contiguous_gather_grouped_gemm_swiglu_fusion_rubin.py
tirx_kernels/ported/flashinfer/gdn_decode/__init__.py
tirx_kernels/ported/flashinfer/gdn_decode/gdn_decode_bf16_ilp4.py
tirx_kernels/ported/flashinfer/gdn_decode/gdn_decode_bf16_wide_vec_mtp.py
tirx_kernels/ported/flashinfer/gdn_decode/gdn_decode_bf16_wide_vec_t1.py
tirx_kernels/ported/flashinfer/gdn_decode/gdn_decode_fp32_mtp_warp.py
tirx_kernels/ported/flashinfer/gdn_prefill/__init__.py
tirx_kernels/ported/flashinfer/gdn_prefill/gdn_cp_prefill_sm100.py
tirx_kernels/ported/flashinfer/gdn_prefill/gdn_prefill_sm100.py
tirx_kernels/ported/flashinfer/gemm/__init__.py
tirx_kernels/ported/flashinfer/gemm/bmm_fp8_rubin.py
tirx_kernels/ported/flashinfer/gemm/dense_blockscaled_gemm_sm107.py
tirx_kernels/ported/flashinfer/gemm/grouped_gemm_masked_rubin.py
tirx_kernels/ported/flashinfer/kda/__init__.py
tirx_kernels/ported/flashinfer/kda/recurrent_kda_decode_grouped.py
tirx_kernels/ported/flashinfer/kda/recurrent_kda_decode_one_warp.py
tirx_kernels/ported/flashinfer/mamba/__init__.py
tirx_kernels/ported/flashinfer/mamba/selective_state_update_mtp_horizontal.py
tirx_kernels/ported/flashinfer/mamba/selective_state_update_mtp_simple.py
tirx_kernels/ported/flashinfer/mamba/selective_state_update_mtp_vertical.py
tirx_kernels/ported/flashinfer/mamba/selective_state_update_stp_horizontal.py
tirx_kernels/ported/flashinfer/mamba/selective_state_update_stp_simple.py
tirx_kernels/ported/flashinfer/mamba/selective_state_update_stp_vertical.py
tirx_kernels/ported/flashinfer/norm/__init__.py
tirx_kernels/ported/flashinfer/norm/_tirx_lite_helpers.py
tirx_kernels/ported/flashinfer/norm/add_rmsnorm_fp4quant.py
tirx_kernels/ported/flashinfer/norm/fused_add_rmsnorm.py
tirx_kernels/ported/flashinfer/norm/fused_add_rmsnorm_quant.py
tirx_kernels/ported/flashinfer/norm/fused_dit_layernorm.py
tirx_kernels/ported/flashinfer/norm/layernorm.py
tirx_kernels/ported/flashinfer/norm/qk_rmsnorm.py
tirx_kernels/ported/flashinfer/norm/rmsnorm.py
tirx_kernels/ported/flashinfer/norm/rmsnorm_fp4quant.py
tirx_kernels/ported/flashinfer/norm/rmsnorm_quant.py
tirx_kernels/ported/flashinfer/quantization/__init__.py
tirx_kernels/ported/flashinfer/quantization/mxfp4_quantize.py
tirx_kernels/ported/flashinfer/quantization/mxfp8_quantize.py
tirx_kernels/ported/flashinfer/quantization/nvfp4_quantize.py
tirx_kernels/ported/flashinfer/quantization/nvfp4_quantize_per_token.py
tirx_kernels/ported/flashinfer/topk/__init__.py
tirx_kernels/ported/flashinfer/topk/fast_topk_clusters.py
tirx_kernels/ported/flashinfer/topk/filtered_topk.py
tirx_kernels/ported/flashinfer/topk/radix_topk_multi_cta.py
tirx_kernels/ported/flashinfer/topk/radix_topk_single_cta.py
tirx_kernels/ported/flashinfer/topk/stable_sort_topk_by_value.py
tirx_kernels/ported/flashinfer/utils/__init__.py
tirx_kernels/ported/flashinfer/utils/_flashkda_bench.py
tirx_kernels/ported/flashinfer/utils/block_radix_sort.py
tirx_kernels/ported/flashinfer/utils/filtered_topk_ops.py
tirx_kernels/ported/flashinfer/utils/fp_quant.py
tirx_kernels/ported/flashinfer/utils/fp_quant_tirx_lite.py
tirx_kernels/ported/flashinfer/utils/source_checkout.py
tirx_kernels/ported/flashinfer/utils/topk_harness.py
tirx_kernels/ported/flashinfer/utils/topk_radix.py
tirx_kernels/ported/flashmla/__init__.py
tirx_kernels/ported/flashmla/flash_mla_sparse_fwd.py
tirx_kernels/ported/flashmla/sparse_decode_head64.py
tirx_kernels/ported/flashmla/sparse_prefill_head128_phase1.py
tirx_kernels/ported/flashmla/sparse_prefill_head128_small_topk_phase1.py
tirx_kernels/ported/flashmla/sparse_prefill_head64_phase1.py
tirx_kernels/ported/flashmla/utils/__init__.py
tirx_kernels/ported/flashmla/utils/_flashmla_bench.py
tirx_kernels/ported/flashmla/utils/_gemm.py
tirx_kernels/ported/flashmla/utils/_mask.py
tirx_kernels/ported/flashmla/utils/_tma.py
tirx_kernels/ported/flashmla/utils/_trtllm_gen_bench.py
tirx_kernels/ported/msa/__init__.py
tirx_kernels/ported/msa/sparse_atten_fwd.py
tirx_kernels/ported/msa/sparse_atten_fwd_combine.py
tirx_kernels/ported/msa/sparse_atten_fwd_nvfp4_kv.py
tirx_kernels/ported/msa/sparse_prepare_flat_schedule.py
tirx_kernels/ported/msa/sparse_prepare_fwd_split_atomic.py
tirx_kernels/ported/msa/utils/__init__.py
tirx_kernels/ported/msa/utils/_msa_bench.py
tirx_kernels/ported/msa/utils/_scalar_ops.py
tirx_kernels/test/__init__.py
tirx_kernels/test/__main__.py
tirx_kernels/tirx_lite/README.md
tirx_kernels/tirx_lite/__init__.py
tirx_kernels/tirx_lite/entry.py
tirx_kernels/tirx_lite/idioms.py
tirx_kernels/tirx_lite/low_level_ir.py
tirx_kernels/tirx_lite/pipeline.py
tirx_kernels/tirx_lite/ring.py
tirx_kernels/tirx_lite/scheduler.py
tirx_kernels/tirx_lite/smem.py
tirx_kernels/tirx_lite/specialize.py
tirx_kernels/vsa/__init__.py
tirx_kernels/vsa/vsa_multishape.py