| .. |
|
attention
|
[Feat][Attention] support GQA prefill fused RoPE (#1234)
|
2026-05-06 20:30:34 +08:00 |
|
__init__.py
|
[Refactor] Separate benchmark profiling from unit tests (#224)
|
2026-02-27 16:46:35 +08:00 |
|
bench_activation.py
|
[Chore][Release] merge testbed into main: op-family alignment + new ops (#1380)
|
2026-05-09 00:03:37 +08:00 |
|
bench_ada_layer_norm.py
|
[Refactor][Benchmark] route ada_layer_norm and cumulative benches through manifest (#1387)
|
2026-05-12 09:57:31 +08:00 |
|
bench_argreduce.py
|
[Feat][Ops] implement Argmax/Argmin dim=None full-tensor reduction (#1106)
|
2026-04-29 17:10:18 +08:00 |
|
bench_batch_norm.py
|
[Chore][Release] merge testbed into main: op-family alignment + new ops (#1380)
|
2026-05-09 00:03:37 +08:00 |
|
bench_binary_arith.py
|
[Chore][Release] merge testbed into main: op-family alignment + new ops (#1380)
|
2026-05-09 00:03:37 +08:00 |
|
bench_binary_elementwise.py
|
[Bench][Elementwise] preserve shape tuples in elementwise bench files (#1202)
|
2026-05-05 13:28:17 +08:00 |
|
bench_binary_strategy.py
|
[Bench][Elementwise] preserve shape tuples in elementwise bench files (#1202)
|
2026-05-05 13:28:17 +08:00 |
|
bench_convolution.py
|
[Refactor][CONV1D] Add dilation support to forward kernel (#1038)
|
2026-04-24 19:24:55 +08:00 |
|
bench_cumulative.py
|
[Refactor][Scan] migrate cumsum + cumprod to canonical static_dims pattern (#1064)
|
2026-04-26 23:51:17 +08:00 |
|
bench_deltanet.py
|
[Refactor][Benchmark] Add explicit type parameters to BenchmarkBase subclasses (#955)
|
2026-04-14 10:25:38 +08:00 |
|
bench_deltanet_recurrence.py
|
[Refactor][Benchmark] Narrow gen_inputs return types across workload protocols (#970)
|
2026-04-15 00:46:16 +08:00 |
|
bench_dropout.py
|
[Bench][Elementwise] preserve shape tuples in elementwise bench files (#1202)
|
2026-05-05 13:28:17 +08:00 |
|
bench_elementwise_fp8.py
|
[Bench][Elementwise] preserve shape tuples in elementwise bench files (#1202)
|
2026-05-05 13:28:17 +08:00 |
|
bench_engram.py
|
[Refactor][Benchmark] Add explicit type parameters to BenchmarkBase subclasses (#955)
|
2026-04-14 10:25:38 +08:00 |
|
bench_fft.py
|
[Refactor][Benchmark] Add explicit type parameters to BenchmarkBase subclasses (#955)
|
2026-04-14 10:25:38 +08:00 |
|
bench_fp8_lighting_indexer.py
|
[Refactor][Benchmark] Narrow gen_inputs return types across workload protocols (#970)
|
2026-04-15 00:46:16 +08:00 |
|
bench_fp8_quant.py
|
[Refactor][Benchmark] Narrow gen_inputs return types across workload protocols (#970)
|
2026-04-15 00:46:16 +08:00 |
|
bench_fused_add_layer_norm.py
|
[Refactor][Roofline] remove manifest-level roofline evaluator (#1024)
|
2026-04-22 18:12:49 +08:00 |
|
bench_fused_add_rms_norm.py
|
[Refactor][Roofline] remove manifest-level roofline evaluator (#1024)
|
2026-04-22 18:12:49 +08:00 |
|
bench_gated_deltanet.py
|
[Refactor][Benchmark] Add explicit type parameters to BenchmarkBase subclasses (#955)
|
2026-04-14 10:25:38 +08:00 |
|
bench_gated_deltanet_recurrence.py
|
[Refactor][Benchmark] Narrow gen_inputs return types across workload protocols (#970)
|
2026-04-15 00:46:16 +08:00 |
|
bench_gemm.py
|
[Refactor][Benchmark] Add explicit type parameters to BenchmarkBase subclasses (#955)
|
2026-04-14 10:25:38 +08:00 |
|
bench_gla_chunkwise.py
|
[Refactor][Benchmark] Add explicit type parameters to BenchmarkBase subclasses (#955)
|
2026-04-14 10:25:38 +08:00 |
|
bench_gla_recurrence.py
|
[Refactor][Benchmark] Narrow gen_inputs return types across workload protocols (#970)
|
2026-04-15 00:46:16 +08:00 |
|
bench_group_norm.py
|
[Fix][NORM] enforce required weight/bias on GroupNorm/InstanceNorm forward (#1402)
|
2026-05-12 09:57:31 +08:00 |
|
bench_grouped_gemm.py
|
[Refactor][Benchmark] Add explicit type parameters to BenchmarkBase subclasses (#955)
|
2026-04-14 10:25:38 +08:00 |
|
bench_grouped_gemm_block_m.py
|
[Refactor][Benchmark] Narrow gen_inputs return types across workload protocols (#970)
|
2026-04-15 00:46:16 +08:00 |
|
bench_independent_elementwise.py
|
[Chore][Release] merge testbed into main: op-family alignment + new ops (#1380)
|
2026-05-09 00:03:37 +08:00 |
|
bench_instance_norm.py
|
[Fix][NORM] enforce required weight/bias on GroupNorm/InstanceNorm forward (#1402)
|
2026-05-12 09:57:31 +08:00 |
|
bench_layer_norm.py
|
[Chore][Release] merge testbed into main: op-family alignment + new ops (#1380)
|
2026-05-09 00:03:37 +08:00 |
|
bench_logical_reduce.py
|
[Fix][Reduction] migrate benchmark callsites to explicit dim=-1
|
2026-05-12 20:10:23 +08:00 |
|
bench_mamba.py
|
[BugFix][Mamba] Fix da_cumsum kernel to support dt_bias, softplus, and clamp (#1118)
|
2026-05-09 18:00:38 +08:00 |
|
bench_mhc_post.py
|
[Refactor][Benchmark] Add explicit type parameters to BenchmarkBase subclasses (#955)
|
2026-04-14 10:25:38 +08:00 |
|
bench_mhc_pre.py
|
[Refactor][Benchmark] Narrow gen_inputs return types across workload protocols (#970)
|
2026-04-15 00:46:16 +08:00 |
|
bench_moe_experts_nopad.py
|
[Feat][MoE] Add modular MoE interface: ABCs + FusedMoe orchestration (#1087)
|
2026-04-28 20:51:01 +08:00 |
|
bench_moe_fused_moe.py
|
[Refactor][Benchmark] Add explicit type parameters to BenchmarkBase subclasses (#955)
|
2026-04-14 10:25:38 +08:00 |
|
bench_moe_fused_topk.py
|
[Bench] Fix nightly benchmark interface regressions (#1113)
|
2026-04-29 12:07:15 +08:00 |
|
bench_moe_kernel_breakdown.py
|
[Refactor][Naming] Add Fwd direction suffix to all elementwise ops and kernels (#978)
|
2026-04-16 22:04:56 +08:00 |
|
bench_moe_permute.py
|
[Refactor][Roofline] remove manifest-level roofline evaluator (#1024)
|
2026-04-22 18:12:49 +08:00 |
|
bench_moe_permute_align.py
|
[Refactor][Roofline] remove manifest-level roofline evaluator (#1024)
|
2026-04-22 18:12:49 +08:00 |
|
bench_moe_permute_nopad.py
|
[Refactor][Roofline] remove manifest-level roofline evaluator (#1024)
|
2026-04-22 18:12:49 +08:00 |
|
bench_moe_shared_fused_moe.py
|
[Refactor][Benchmark] Add explicit type parameters to BenchmarkBase subclasses (#955)
|
2026-04-14 10:25:38 +08:00 |
|
bench_moe_unpermute.py
|
[Refactor][Roofline] remove manifest-level roofline evaluator (#1024)
|
2026-04-22 18:12:49 +08:00 |
|
bench_pool.py
|
[Refactor][Benchmark] Narrow gen_inputs return types across workload protocols (#970)
|
2026-04-15 00:46:16 +08:00 |
|
bench_reduce.py
|
[Fix][Reduction] migrate benchmark callsites to explicit dim=-1
|
2026-05-12 20:10:23 +08:00 |
|
bench_reduce_multidim.py
|
[Refactor][Benchmark] route ada_layer_norm and cumulative benches through manifest (#1387)
|
2026-05-12 09:57:31 +08:00 |
|
bench_rms_norm.py
|
[Chore][Release] merge testbed into main: op-family alignment + new ops (#1380)
|
2026-05-09 00:03:37 +08:00 |
|
bench_rope.py
|
[Bench][Elementwise] preserve shape tuples in elementwise bench files (#1202)
|
2026-05-05 13:28:17 +08:00 |
|
bench_softmax.py
|
[Refactor][Manifest] split monolithic yaml into per-family package (#1092)
|
2026-04-28 13:41:10 +08:00 |
|
bench_topk_selector.py
|
[Refactor][Benchmark] Add explicit type parameters to BenchmarkBase subclasses (#955)
|
2026-04-14 10:25:38 +08:00 |
|
bench_unary_elementwise.py
|
[Feat][Ops] align elementwise_unary_math family to manifest spec (#1174)
|
2026-05-05 07:43:06 +08:00 |
|
bench_unary_strategy.py
|
[Bench][Elementwise] preserve shape tuples in elementwise bench files (#1202)
|
2026-05-05 13:28:17 +08:00 |
|
bench_vector_norm.py
|
[Feat] vector_norm: support empty-dim full reduction (L1/L2/Inf) (#1104)
|
2026-04-29 16:13:37 +08:00 |
64dc0b67fd