Endpoints — github.com/ByteDance-Seed/Triton-distributed

hub / github.com/ByteDance-Seed/Triton-distributed / endpoints

Endpoints149 in github.com/ByteDance-Seed/Triton-distributed

⨍ Functions 5,103◇ Types & classes 521↳ Endpoints 149

Route__syncthreads

_extra_module.dispatch

python/triton_dist/language/extra/language_extra.py:None

Route_forward_pull_kernel

triton_dist.jit(do_not_specialize=["rank", "signal_target"])

python/triton_dist/kernels/nvidia/low_latency_allgather.py:None

Route_forward_push_2d_kernel

triton_dist.jit(do_not_specialize=["rank", "signal_target"])

python/triton_dist/kernels/nvidia/low_latency_allgather.py:None

Route_forward_push_2d_ll_kernel

triton_dist.jit(do_not_specialize=["rank", "signal_target"])

python/triton_dist/kernels/nvidia/low_latency_allgather.py:None

Route_forward_push_2d_ll_multimem_kernel

triton_dist.jit(do_not_specialize=["rank", "signal_target"])

python/triton_dist/kernels/nvidia/low_latency_allgather.py:None

Route_forward_push_3d_kernel

triton_dist.jit(do_not_specialize=["rank", "signal_target"])

python/triton_dist/kernels/nvidia/low_latency_allgather.py:None

Route_forward_push_numa_2d_kernel

triton_dist.jit(do_not_specialize=["rank", "signal_target"])

python/triton_dist/kernels/nvidia/low_latency_allgather.py:None

Route_forward_push_numa_2d_ll_kernel

triton_dist.jit(do_not_specialize=["rank", "signal_target"])

python/triton_dist/kernels/nvidia/low_latency_allgather.py:None

Route_forward_push_numa_2d_ll_multinode_kernel

triton_dist.jit(do_not_specialize=["rank", "signal_target"])

python/triton_dist/kernels/nvidia/low_latency_allgather.py:None

Routeag_gemm_intra_node_op

triton_dist.tune.autotune( config_space=[{"gemm_config": c} for c in matmul_get_configs()], key_fn=key

python/triton_dist/kernels/amd/allgather_gemm.py:None

Routeatomic_add

_extra_module.dispatch

python/triton_dist/language/extra/language_extra.py:None

Routeatomic_cas

_extra_module.dispatch

python/triton_dist/language/extra/language_extra.py:None

Routebarrier

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebarrier_all

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebarrier_all_block

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebarrier_all_intra_node_non_atomic

triton_dist.jit(do_not_specialize=["local_rank", "rank", "num_ranks", "target_value"])

python/triton_dist/kernels/nvidia/common_ops.py:None

Routebarrier_all_intra_node_non_atomic_block

triton_dist.jit(do_not_specialize=["local_rank", "rank", "num_ranks", "target_value"])

python/triton_dist/kernels/nvidia/common_ops.py:None

Routebarrier_all_vec

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebarrier_all_warp

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebarrier_all_wave

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebarrier_all_wg

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebarrier_block

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebarrier_warp

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebroadcast

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebroadcast_block

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebroadcast_warp

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebroadcastmem

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebroadcastmem_block

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routebroadcastmem_warp

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routechunk_fwd_kernel_o

aot_compile_spaces({ "chunk_fwd_kernel_o": { "signature": chunk_fwd_o_signature.format(k_d

python/triton_dist/kernels/nvidia/gdn.py:None

Routechunk_gated_delta_rule_fwd_kernel_h_blockdim64

aot_compile_spaces({ "chunk_gated_delta_rule_fwd_kernel_h_blockdim64": { "signature": chun

python/triton_dist/kernels/nvidia/gdn.py:None

Routechunk_kkt_inv_ut_fused_kernel

aot_compile_spaces({ "chunk_kkt_inv_ut_fused_kernel": { "signature": chunk_kkt_inv_ut_fuse

python/triton_dist/kernels/nvidia/gdn.py:None

Routeclock64

builtin(eval_return_type=uint64, codegen_func=codegen_get_clock)

python/little_kernel/language/intrin/time_utils.py:None

Routefcollect

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routefcollect_block

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routefcollect_warp

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routefcollectmem

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routefcollectmem_block

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routefcollectmem_warp

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routefence

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routegemm_allreduce_op

triton_dist.tune.autotune( config_space=[{"gemm_config": c} for c in get_hip_autotune_config()], key_f

python/triton_dist/kernels/amd/gemm_allreduce.py:None

Routegemm_non_persistent

triton_dist.tune.autotune( config_space=[{"gemm_config": x} for x in get_config_space(False)], key_fn=

python/triton_dist/kernels/nvidia/allgather_gemm.py:None

Routegemm_only

triton_dist.tune.autotune( config_space=[{"gemm_config": c} for c in matmul_get_configs()], key_fn=key

python/triton_dist/kernels/amd/allgather_gemm.py:None

Routegemm_persistent

triton_dist.tune.autotune( config_space=[{"gemm_config": x} for x in get_config_space(True)], key_fn=k

python/triton_dist/kernels/nvidia/allgather_gemm.py:None

Routegemm_rs

triton_dist.tune.autotune( config_space=get_gemm_rs_config_space(), key_fn=key_fn, prune_fn=prune_

python/triton_dist/kernels/nvidia/gemm_reduce_scatter.py:None

Routegemm_rs_intra_node_op

triton_dist.tune.autotune(config_space=[{"gemm_config": c} for c in get_hip_autotune_config()], key_fn=key_fn,

python/triton_dist/kernels/amd/gemm_reduce_scatter.py:None

Routeget_lane_idx

builtin(eval_return_type=int32, codegen_func=codegen_get_lane_idx)

python/little_kernel/language/intrin/simt.py:None

Routegetmem

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routegetmem_block

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routegetmem_nbi

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routegetmem_nbi_block

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routegetmem_nbi_warp

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routegetmem_nbi_wave

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routegetmem_nbi_wg

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routegetmem_warp

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routegetmem_wave

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routegetmem_wg

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routeint_p

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routekernel_combine_token_intra_node

triton_dist.jit(do_not_specialize=["num_dispatch_tokens"])

python/triton_dist/kernels/amd/ep_a2a_intra_node.py:None

Routekernel_dispatch_token

aot_compile_spaces({ "kernel_dispatch_token_bf16_weight_fp32": { "signature": kernel_dispa

python/triton_dist/kernels/nvidia/ep_a2a.py:None

Routekernel_dispatch_token_intra_node

triton_dist.jit(do_not_specialize=["dispatch_recv_token_num"])

python/triton_dist/kernels/amd/ep_a2a_intra_node.py:None

Routekernel_dispatch_token_intra_node

triton_dist.jit(do_not_specialize=["dispatch_recv_token_num"])

python/triton_dist/kernels/nvidia/ep_a2a_intra_node.py:None

Routekernel_get_ag_splits_and_recv_offset

aot_compile_spaces({ "kernel_get_ag_splits_and_recv_offset": { "signature": kernel_get_ag_

python/triton_dist/kernels/nvidia/ep_a2a.py:None

Routekernel_get_dispatch_send_reqs

aot_compile_spaces({ "get_dispatch_send_reqs_i32": { "signature": "*i32:16, *i32:16, *i32,

python/triton_dist/kernels/nvidia/ep_a2a.py:None

Routekernel_gqa_fwd_batch_decode_combine_kv

aot_compile_spaces({ "gqa_fwd_batch_decode_combine_kv_fp32_fp16": { "signature": combine_k

python/triton_dist/kernels/nvidia/flash_decode.py:None

Routekernel_gqa_fwd_batch_decode_split_kv

aot_compile_spaces({ "gqa_fwd_batch_decode_split_kv_fp16_fp16_fp32": { "signature": split_

python/triton_dist/kernels/nvidia/flash_decode.py:None

Routekernel_gqa_fwd_batch_decode_split_kv_persistent

aot_compile_spaces({ "gqa_fwd_batch_decode_split_kv_persistent_fp16_fp16_fp16": { "signature":

python/triton_dist/kernels/nvidia/flash_decode.py:None

Routekernel_inter_rank_gqa_fwd_batch_decode_combine_kv

aot_compile_spaces({ "inter_rank_gqa_fwd_batch_decode_combine_kv_fp32_fp16": { "signature":

python/triton_dist/kernels/nvidia/flash_decode.py:None

Routekernel_intra_rank_gqa_fwd_batch_decode_combine_kv

aot_compile_spaces({ "intra_rank_gqa_fwd_batch_decode_combine_kv_fp32_fp16": { "signature":

python/triton_dist/kernels/nvidia/flash_decode.py:None

Routekernel_skipped_token_local_dispatch_intra_node

triton_dist.jit(do_not_specialize=["dispatch_recv_token_num"])

python/triton_dist/kernels/amd/ep_a2a_intra_node.py:None

Routekernel_skipped_token_local_dispatch_intra_node

triton_dist.jit(do_not_specialize=["dispatch_recv_token_num"])

python/triton_dist/kernels/nvidia/ep_a2a_intra_node.py:None

Routeld

_extra_module.dispatch

python/triton_dist/language/extra/language_extra.py:None

Routeld_vector

_extra_module.dispatch

python/triton_dist/language/extra/language_extra.py:None

Routematmul

triton_dist.tune.autotune( config_space=[{"config": c} for c in get_config_space(False)], key_fn=lambd

python/triton_dist/kernels/nvidia/gemm.py:None

Routematmul_descriptor_persistent

triton_dist.tune.autotune( config_space=[{"config": c} for c in get_config_space(persistent=True)], ke

python/triton_dist/kernels/nvidia/gemm.py:None

Routematmul_kernel

triton.autotune( configs=get_autotune_config(), key=['M', 'N', 'K'], )

python/triton_dist/test/amd/test_matmul_amd.py:None

Routematmul_kernel_descriptor_persistent

aot_compile_spaces({ f"matmul_kernel_descriptor_persistent_{ty}": { "signature": f"*{ty}, *{ty}, *

python/triton_dist/test/nvidia/test_compile_aot.py:None

Routematmul_persistent

triton_dist.tune.autotune( config_space=[{"config": c} for c in get_config_space()], key_fn=lambda *ar

python/triton_dist/kernels/nvidia/gemm.py:None

Routematmul_persistent_triton

triton_dist.tune.autotune(config_space=[{"config": c} for c in get_config_space_persistent()], key_fn=key_fn,

python/triton_dist/kernels/amd/gemm.py:None

Routematmul_split_k_persistent_triton

triton_dist.tune.autotune(config_space=[{"config": c} for c in get_config_space_persistent()], key_fn=key_fn_s

python/triton_dist/kernels/amd/gemm.py:None

Routematmul_splitk_chunked_triton

triton_dist.tune.autotune(config_space=[{"config": c} for c in get_splitk_config_space()], key_fn=key_fn_split

python/triton_dist/kernels/amd/gemm.py:None

Routematmul_splitk_triton

triton_dist.tune.autotune(config_space=[{"config": c} for c in get_splitk_config_space()], key_fn=key_fn_split

python/triton_dist/kernels/amd/gemm.py:None

Routematmul_tma

triton_dist.tune.autotune( config_space=[{"config": c} for c in get_config_space()], key_fn=lambda *ar

python/triton_dist/kernels/nvidia/gemm.py:None

Routematmul_tma_persistent

triton_dist.tune.autotune( config_space=[{"config": c} for c in get_config_space(persistent=True)], ke

python/triton_dist/kernels/nvidia/gemm.py:None

Routematmul_triton

triton_dist.tune.autotune(config_space=[{"config": c} for c in get_config_space()], key_fn=key_fn, prune_fn=pr

python/triton_dist/kernels/amd/gemm.py:None

Routemy_pe

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routen_pes

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routepack

_extra_module.dispatch

python/triton_dist/language/extra/language_extra.py:None

Routeputmem

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routeputmem_block

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routeputmem_nbi

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routeputmem_nbi_block

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routeputmem_nbi_warp

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routeputmem_nbi_wave

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routeputmem_nbi_wg

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routeputmem_rma

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routeputmem_rma_block

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routeputmem_rma_nbi

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routeputmem_rma_nbi_block

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

Routeputmem_rma_nbi_warp

_shmem_module.dispatch

python/triton_dist/language/extra/libshmem_device.py:None

next →1–100 of 149, ranked by callers