Norm, Conv, Pool & Other

45 ops, 414 workloads — Normalization 23 · Convolution 13 · Pooling 8 · Top-k 1.

One table per op, one row per workload. Ratio is the baseline device time divided by ours in the same measurement regime, so green is faster than it, plain is level with it, red is slower. Times are in ms. How these numbers are taken.

Normalization

AdaLayerNormFwd

• W1x: [1024, 1152]dtype=bf16dit-xl-2
• W2x: [1024, 1152]dtype=f16dit-xl-2
• W3x: [1, 4096]dtype=bf16llama-8b-decode
• W4x: [2048, 4096]dtype=bf16llama-8b-prefill
• W5x: [2048, 4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	2.73×	0.0222	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.0615 0.0521	·	·	·
W2	2.09×	0.0255	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.0540 0.0503	·	·	·
W3	3.14×	0.0035	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis tilelang-mlir-ascend:high_perf tilelang-mlir-ascend ⓘ eager · n=5 ⓘ	0.0110 0.00962 0.00389	·	·	·
W4	1.34×	0.1030	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.1393 0.1192	·	·	·
W5	1.52×	0.1005	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.1625 0.1522	·	·	·

BatchNormBwd

dtype=f16

• W1x: [4, 128, 1024, 1024]large-spatial
• W2x: [32, 64]resnet50-fc
• W3x: [8, 64, 32, 32]resnet50-stage1
• W4x: [4, 128, 32, 32]resnet50-stage2
• W5x: [4, 256, 28, 28]resnet50-stage3

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.17×	11.4718	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	13.4796 13.4795 13.4889	·	·	·
W2	1.64×	0.0377	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0511 0.0508 0.0506	·	·	·
W3	2.06×	0.0318	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0732 0.0644 0.0650	·	·	·
W4	2.15×	0.0297	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0712 0.0630 0.0616	·	·	·
W5	1.64×	0.0525	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0935 0.0855 0.0825	·	·	·

L1NormFwd

• W1x: [4, 128, 4096]dtype=f16dim=[0,2]3d-multidim-reduce
• W2x: [2048, 4096]dtype=bf16hidden-state-l1
• W3x: [2048, 4096]dtype=f16hidden-state-l1
• W4x: [64, 32768]dtype=bf16long-seq-l1

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.93×	0.0357	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0334 0.0318 0.0198	·	·	·
W2	1.96×	0.0403	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0798 0.0695 0.0508	·	·	·
W3	1.98×	0.0399	torch_npu eager vendor library basis	0.0790	·	·	·
W4	2.05×	0.0227	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0481 0.0362 0.0200	·	·	·

L2NormFwd

• W1x: [4, 128, 4096]dtype=f16dim=[0,2]3d-multidim-reduce
• W2x: [2048, 4096]dtype=bf16hidden-state-l2
• W3x: [2048, 4096]dtype=f16hidden-state-l2
• W4x: [64, 32768]dtype=bf16long-seq-l2

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.14×	0.0355	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0401 0.0208	·	·	·
W2	1.76×	0.0420	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0742 0.0678	·	·	·
W3	1.79×	0.0411	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0747 0.0670	·	·	·
W4	1.88×	0.0220	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0400 0.0350	·	·	·

FusedAddLayerNormFwd

• W1x: [1, 8192]dtype=bf16llama-70b-decode
• W2x: [2048, 8192]dtype=bf16llama-70b-prefill
• W3x: [2048, 8192]dtype=f16llama-70b-prefill
• W4x: [1, 4096]dtype=bf16llama-8b-decode
• W5x: [2048, 4096]dtype=bf16llama-8b-prefill
• W6x: [2048, 4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	2.38×	0.00525	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0125 0.0103	·	·	·
W2	1.26×	0.2077	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2644 0.2492	·	·	·
W3	1.30×	0.1886	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2459 0.2288	·	·	·
W4	2.09×	0.004	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0085 0.00725	·	·	·
W5	1.16×	0.1185	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1309 0.1222	·	·	·
W6	1.39×	0.1135	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1584 0.1405	·	·	·

RMSNormQuantFwd

• W1x: [1, 16384]normalized: [16384]dtype=bf16llama-405b-decode
• W2x: [2048, 16384]normalized: [16384]dtype=bf16llama-405b-prefill
• W3x: [2048, 16384]normalized: [16384]dtype=f16llama-405b-prefill
• W4x: [1, 8192]normalized: [8192]dtype=bf16llama-70b-decode
• W5x: [2048, 8192]normalized: [8192]dtype=bf16llama-70b-prefill
• W6x: [2048, 8192]normalized: [8192]dtype=f16llama-70b-prefill
• W7x: [1, 4096]normalized: [4096]dtype=bf16llama-8b-decode
• W8x: [2048, 4096]normalized: [4096]dtype=bf16llama-8b-prefill
• W9x: [2048, 4096]normalized: [4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.84×	0.0622	torch_npu eager vendor library basis	0.1143	·	·	·
W2	1.35×	4.5127	torch_npu eager vendor library basis	6.0799	·	·	·
W3	1.36×	4.6085	torch_npu eager vendor library basis	6.2891	·	·	·
W4	1.74×	0.0511	torch_npu eager vendor library basis	0.0887	·	·	·
W5	1.12×	1.9303	torch_npu eager vendor library basis	2.1546	·	·	·
W6	1.14×	1.9325	torch_npu eager vendor library basis	2.2107	·	·	·
W7	1.71×	0.0418	torch_npu eager vendor library basis	0.0712	·	·	·
W8	1.14×	0.9872	torch_npu eager vendor library basis	1.1266	·	·	·
W9	1.12×	0.9789	torch_npu eager vendor library basis	1.0958	·	·	·

LayerNormFwd

• W1x: [1, 16384]normalized: [16384]dtype=bf16llama-405b-decode
• W2x: [2048, 16384]normalized: [16384]dtype=bf16llama-405b-prefill
• W3x: [2048, 16384]normalized: [16384]dtype=f16llama-405b-prefill
• W4x: [1, 8192]normalized: [8192]dtype=bf16llama-70b-decode
• W5x: [2048, 8192]normalized: [8192]dtype=bf16llama-70b-prefill
• W6x: [2048, 8192]normalized: [8192]dtype=f16llama-70b-prefill
• W7x: [1, 4096]normalized: [4096]dtype=bf16llama-8b-decode
• W8x: [2048, 4096]normalized: [4096]dtype=bf16llama-8b-prefill
• W9x: [2048, 4096]normalized: [4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.66×	0.00725	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.0116 0.0110	·	·	·
W2	1.22×	0.2909	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.3538 0.3422	·	·	·
W3	1.20×	0.2868	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.3441 0.3369	·	·	·
W4	1.76×	0.00425	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.00725 0.00725	·	·	·
W5	0.78×	0.1628	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.1280 0.1192	·	·	·
W6	0.79×	0.1653	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.1300 0.1180	·	·	·
W7	1.29×	0.0035	ops-nn:aclnnLayerNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_npu eager vendor library tilelang-mlir-ascend:high_perf tilelang-mlir-ascend ⓘ eager · n=5 ⓘ	0.0035 0.00475 0.0035 0.0335	·	·	·
W8	0.71×	0.0775	ops-nn:aclnnLayerNorm open-source ⓘ torch_compile_aclgraph torch_compile torch_npu eager vendor library basis tilelang-mlir-ascend:high_perf tilelang-mlir-ascend ⓘ eager · n=5 ⓘ	0.0537 0.0553 0.0529 0.0935	·	·	·
W9	1.24×	0.0757	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.0931 0.0793	·	·	·

BatchNormFwd

dtype=f16

• W1x: [4, 128, 1024, 1024]large-spatial
• W2x: [32, 64]resnet50-fc
• W3x: [8, 64, 32, 32]resnet50-stage1
• W4x: [4, 128, 32, 32]resnet50-stage2
• W5x: [4, 256, 28, 28]resnet50-stage3

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.27×	2.6470	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	3.3580 3.3549 3.3563	·	·	·
W2	1.13×	0.00575	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0075 0.0065 0.0035	·	·	·
W3	1.02×	0.0150	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0152 0.0100 0.0107	·	·	·
W4	1.21×	0.0127	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0152 0.0120 0.0120	·	·	·
W5	0.95×	0.0185	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0175 0.0110 0.0136	·	·	·

InstanceNormFwd

• W1x: [8, 128, 32, 32]weight, bias: [128]dtype=bf16image-affine
• W2x: [8, 128, 32, 32]weight, bias: [128]dtype=f16image-affine
• W5x: [4, 64, 30, 30]weight, bias: [64]dtype=f16tail-spatial-affine
• W7x: [4, 256, 28, 28]weight, bias: [256]dtype=f16wider-channel-affine

• W3x: [8, 128, 32, 32]running_mean, running_var: [128], f32dtype=bf16image
• W4x: [8, 128, 32, 32]running_mean, running_var: [128], f32dtype=f16image
• W6x: [4, 64, 30, 30]running_mean, running_var: [64], f32dtype=f16tail-spatial
• W8x: [4, 256, 28, 28]running_mean, running_var: [256], f32dtype=f16wider-channel

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.21×	0.0217	ops-nn-b2:aclnnGroupNormSiluV2 open-source ⓘ torch_npu eager vendor library basis	0.0222 0.0201	·	·	·
W2	1.32×	0.0198	ops-nn-b2:aclnnGroupNormSiluV2 open-source ⓘ torch_npu eager vendor library basis	0.0231 0.0216	·	·	·
W3	0.91×	0.0532	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.0480 0.0405	·	·	·
W4	0.92×	0.0525	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.0486 0.0408	·	·	·
W5	1.13×	0.0163	ops-nn-b2:aclnnGroupNormSiluV2 open-source ⓘ basis torch_npu eager vendor library	0.0123 0.0168	·	·	·
W6	0.70×	0.0447	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.0320 0.0253	·	·	·
W7	1.51×	0.0177	ops-nn-b2:aclnnGroupNormSiluV2 open-source ⓘ torch_npu eager vendor library basis	0.0220 0.0200	·	·	·
W8	1.00×	0.0516	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.0542 0.0480	·	·	·

RMSNormFwd

• W1x: [1, 16384]normalized: [16384]dtype=bf16llama-405b-decode
• W2x: [2048, 16384]normalized: [16384]dtype=bf16llama-405b-prefill
• W3x: [2048, 16384]normalized: [16384]dtype=f16llama-405b-prefill
• W4x: [1, 8192]normalized: [8192]dtype=bf16llama-70b-decode
• W5x: [2048, 8192]normalized: [8192]dtype=bf16llama-70b-prefill
• W6x: [2048, 8192]normalized: [8192]dtype=f16llama-70b-prefill
• W7x: [1, 4096]normalized: [4096]dtype=bf16llama-8b-decode
• W8x: [2048, 4096]normalized: [4096]dtype=bf16llama-8b-prefill
• W9x: [2048, 4096]normalized: [4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.34×	0.0055	ops-nn:aclnnRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library ⓘ	0.00537 0.0240 0.0232 0.0226	·	·	·
W2	0.86×	0.2157	ops-nn:aclnnRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library ⓘ	0.1745 0.8171 0.8174 0.8177	·	·	·
W3	0.85×	0.2120	ops-nn:aclnnRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library tilelang-mlir-ascend:autotune_2d tilelang-mlir-ascend ⓘ eager · n=5 ⓘ	0.1694 0.8265 0.8249 0.8260 17.4019	·	·	·
W4	1.38×	0.00325	ops-nn:aclnnRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library ⓘ	0.00275 0.0217 0.0210 0.0208	·	·	·
W5	0.82×	0.0924	ops-nn:aclnnRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library ⓘ	0.0675 0.2757 0.2769 0.2761	·	·	·
W6	0.82×	0.0907	ops-nn:aclnnRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library tilelang-mlir-ascend:plain tilelang-mlir-ascend ⓘ eager · n=5 ⓘ	0.0663 0.2767 0.3035 0.2741 0.0642	·	·	·
W7	1.45×	0.00275	ops-nn:aclnnRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library tilelang-mlir-ascend:plain tilelang-mlir-ascend ⓘ eager · n=5 ⓘ	0.0025 0.0302 0.00737 0.0195 0.00192	·	·	·
W8	0.98×	0.0545	ops-nn:aclnnRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library tilelang-mlir-ascend:plain tilelang-mlir-ascend ⓘ eager · n=5 ⓘ	0.0441 0.1700 0.1775 0.1685 0.0347	·	·	·
W9	0.93×	0.0584	ops-nn:aclnnRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library tilelang-mlir-ascend:plain tilelang-mlir-ascend ⓘ eager · n=5 ⓘ	0.0452 0.1731 0.2300 0.1726 0.0343	·	·	·

LayerNormQuantFwd

• W1x: [1, 16384]normalized: [16384]dtype=bf16llama-405b-decode
• W2x: [2048, 16384]normalized: [16384]dtype=bf16llama-405b-prefill
• W3x: [2048, 16384]normalized: [16384]dtype=f16llama-405b-prefill
• W4x: [1, 8192]normalized: [8192]dtype=bf16llama-70b-decode
• W5x: [2048, 8192]normalized: [8192]dtype=bf16llama-70b-prefill
• W6x: [2048, 8192]normalized: [8192]dtype=f16llama-70b-prefill
• W7x: [1, 4096]normalized: [4096]dtype=bf16llama-8b-decode
• W8x: [2048, 4096]normalized: [4096]dtype=bf16llama-8b-prefill
• W9x: [2048, 4096]normalized: [4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.29×	0.0777	torch_npu eager vendor library basis	0.1004	·	·	·
W2	1.00×	5.9374	torch_npu eager vendor library basis	5.9467	·	·	·
W3	1.05×	5.9260	torch_npu eager vendor library basis	6.2252	·	·	·
W4	1.13×	0.0605	torch_npu eager vendor library basis	0.0683	·	·	·
W5	0.91×	2.3194	torch_npu eager vendor library basis	2.1175	·	·	·
W6	0.89×	2.3469	torch_npu eager vendor library basis	2.0978	·	·	·
W7	1.08×	0.0506	torch_npu eager vendor library basis	0.0545	·	·	·
W8	0.87×	1.2195	torch_npu eager vendor library basis	1.0617	·	·	·
W9	0.88×	1.2189	torch_npu eager vendor library basis	1.0720	·	·	·

GroupNormFwd

• W1x: [8, 128, 32, 32]weight, bias: [128]dtype=bf16num_groups=32image-g32-affine
• W2x: [8, 128, 32, 32]weight, bias: [128]dtype=f16num_groups=32image-g32-affine
• W5x: [4, 128, 30, 30]weight, bias: [128]dtype=f16num_groups=16tail-spatial-g16-affine
• W7x: [4, 256, 28, 28]weight, bias: [256]dtype=f16num_groups=32wider-channel-g32-affine

• W3x: [8, 128, 32, 32]dtype=bf16num_groups=32image-g32
• W4x: [8, 128, 32, 32]dtype=f16num_groups=32image-g32
• W6x: [4, 128, 30, 30]dtype=f16num_groups=16tail-spatial-g16
• W8x: [4, 256, 28, 28]dtype=f16num_groups=32wider-channel-g32

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.82×	0.0262	ops-nn-b2:aclnnGroupNormSiluV2 open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library	0.0140 0.0140 0.0140 0.0140	·	·	·
W2	0.69×	0.0267	ops-nn-b2:aclnnGroupNormSiluV2 open-source ⓘ torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0132 0.0130 0.0132 0.0132	·	·	·
W3	0.98×	0.0220	ops-nn-b2:aclnnGroupNormSiluV2 open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library	0.0132 0.0235 0.0232 0.0160	·	·	·
W4	0.96×	0.0210	ops-nn-b2:aclnnGroupNormSiluV2 open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library	0.0132 0.0225 0.0195 0.0158	·	·	·
W5	0.90×	0.0190	ops-nn-b2:aclnnGroupNormSiluV2 open-source ⓘ torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0107 0.0106 0.0105 0.0105	·	·	·
W6	1.07×	0.0160	ops-nn-b2:aclnnGroupNormSiluV2 open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library	0.0115 0.0198 0.0173 0.0138	·	·	·
W7	1.00×	0.0180	ops-nn-b2:aclnnGroupNormSiluV2 open-source ⓘ torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0114 0.0112 0.0112 0.0110	·	·	·
W8	1.16×	0.0160	ops-nn-b2:aclnnGroupNormSiluV2 open-source ⓘ basis torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library	0.0118 0.0205 0.0185 0.0135	·	·	·

FusedAddRMSNormFwd

• W1x: [1, 16384]dtype=bf16llama-405b-decode
• W2x: [2048, 16384]dtype=bf16llama-405b-prefill
• W3x: [2048, 16384]dtype=f16llama-405b-prefill
• W4x: [1, 8192]dtype=bf16llama-70b-decode
• W5x: [2048, 8192]dtype=bf16llama-70b-prefill
• W6x: [2048, 8192]dtype=f16llama-70b-prefill
• W7x: [1, 4096]dtype=bf16llama-8b-decode
• W8x: [2048, 4096]dtype=bf16llama-8b-prefill
• W9x: [2048, 4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.86×	0.00725	ops-nn:aclnnAddRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_npu eager vendor library ⓘ	0.00525 0.0295 0.0288	·	·	·
W2	0.90×	0.3139	ops-nn:aclnnAddRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_npu eager vendor library tilelang-mlir-ascend:selector tilelang-mlir-ascend ⓘ eager · n=5 ⓘ	0.2755 1.1335 1.1365 ·	·	·	·
W3	0.82×	0.2913	ops-nn:aclnnAddRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_npu eager vendor library ⓘ	0.2377 1.1364 1.0870	·	·	·
W4	0.78×	0.0045	ops-nn:aclnnAddRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_npu eager vendor library ⓘ	0.00325 0.0413 0.0272	·	·	·
W5	0.74×	0.1713	ops-nn:aclnnAddRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_npu eager vendor library ⓘ	0.1235 0.4515 0.4954	·	·	·
W6	0.80×	0.1590	ops-nn:aclnnAddRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_npu eager vendor library ⓘ	0.1153 0.3962 0.4716	·	·	·
W7	0.76×	0.00363	ops-nn:aclnnAddRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_npu eager vendor library tilelang-mlir-ascend:selector tilelang-mlir-ascend ⓘ eager · n=5 ⓘ	0.0025 0.0374 0.0250 ·	·	·	·
W8	0.87×	0.0907	ops-nn:aclnnAddRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_npu eager vendor library tilelang-mlir-ascend:selector tilelang-mlir-ascend ⓘ eager · n=5 ⓘ	0.0750 0.2382 0.2285 ·	·	·	·
W9	0.86×	0.0872	ops-nn:aclnnAddRmsNorm open-source ⓘ basis torch_compile_aclgraph torch_compile torch_npu eager vendor library tilelang-mlir-ascend:selector tilelang-mlir-ascend ⓘ eager · n=5 ⓘ	0.0602 0.2209 0.2171 ·	·	·	·

InstanceNormBwd

• W1x: [8, 128, 32, 32]weight, bias: [128]dtype=bf16image-affine
• W2x: [8, 128, 32, 32]weight, bias: [128]dtype=f16image-affine
• W5x: [4, 64, 30, 30]weight, bias: [64]dtype=f16tail-spatial-affine
• W7x: [4, 256, 28, 28]weight, bias: [256]dtype=f16wider-channel-affine

• W3x: [8, 128, 32, 32]running_mean, running_var: [128], f32dtype=bf16image
• W4x: [8, 128, 32, 32]running_mean, running_var: [128], f32dtype=f16image
• W6x: [4, 64, 30, 30]running_mean, running_var: [64], f32dtype=f16tail-spatial
• W8x: [4, 256, 28, 28]running_mean, running_var: [256], f32dtype=f16wider-channel

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.56×	0.4622	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2585 0.2526	·	·	·
W2	0.55×	0.4680	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2582 0.2540	·	·	·
W3	0.43×	0.3782	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1749 0.1487	·	·	·
W4	0.43×	0.3805	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1771 0.1496	·	·	·
W5	0.41×	0.5150	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2130 0.2070	·	·	·
W6	0.36×	0.3488	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1363 0.1060	·	·	·
W7	0.28×	0.6202	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1790 0.1693	·	·	·
W8	0.21×	0.4335	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0963 0.0710	·	·	·

GroupNormBwd

• W1x: [8, 128, 32, 32]weight, bias: [128]dtype=bf16num_groups=32image-g32-affine
• W2x: [8, 128, 32, 32]weight, bias: [128]dtype=f16num_groups=32image-g32-affine
• W5x: [4, 128, 30, 30]weight, bias: [128]dtype=f16num_groups=16tail-spatial-g16-affine
• W7x: [4, 256, 28, 28]weight, bias: [256]dtype=f16num_groups=32wider-channel-g32-affine

• W3x: [8, 128, 32, 32]dtype=bf16num_groups=32image-g32
• W4x: [8, 128, 32, 32]dtype=f16num_groups=32image-g32
• W6x: [4, 128, 30, 30]dtype=f16num_groups=16tail-spatial-g16
• W8x: [4, 256, 28, 28]dtype=f16num_groups=32wider-channel-g32

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.47×	0.5555	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2587 0.2497	·	·	·
W2	0.48×	0.5400	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2619 0.2511	·	·	·
W3	0.47×	0.5058	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2396 0.2290	·	·	·
W4	0.48×	0.5104	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2411 0.2299	·	·	·
W5	0.38×	0.5814	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2177 0.2127	·	·	·
W6	0.35×	0.5711	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1968 0.1940	·	·	·
W7	0.24×	0.7140	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1779 0.1695	·	·	·
W8	0.23×	0.6837	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1540 0.1497	·	·	·

WeightNormFwd

• W1x: [1, 16384]normalized: [16384]dtype=bf16llama-405b-decode
• W2x: [2048, 16384]normalized: [16384]dtype=bf16llama-405b-prefill
• W3x: [2048, 16384]normalized: [16384]dtype=f16llama-405b-prefill
• W4x: [1, 8192]normalized: [8192]dtype=bf16llama-70b-decode
• W5x: [2048, 8192]normalized: [8192]dtype=bf16llama-70b-prefill
• W6x: [2048, 8192]normalized: [8192]dtype=f16llama-70b-prefill
• W7x: [1, 4096]normalized: [4096]dtype=bf16llama-8b-decode
• W8x: [2048, 4096]normalized: [4096]dtype=bf16llama-8b-prefill
• W9x: [2048, 4096]normalized: [4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.85×	0.0624	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0250 0.0245	·	·	·
W2	0.26×	2.6923	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.7278 0.6945	·	·	·
W3	0.26×	2.7535	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.7109 0.7335	·	·	·
W4	0.53×	0.0633	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0331 0.0210	·	·	·
W5	0.27×	1.0950	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.3050 0.2989	·	·	·
W6	0.27×	1.1033	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.3020 0.2988	·	·	·
W7	0.61×	0.0498	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0302 0.0182	·	·	·
W8	0.31×	0.5753	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1797 0.1787	·	·	·
W9	0.32×	0.5763	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1831 0.1814	·	·	·

GroupRMSNormFwd

• W1x: [1, 16384]normalized: [16384]dtype=bf16llama-405b-decode
• W2x: [2048, 16384]normalized: [16384]dtype=bf16llama-405b-prefill
• W3x: [2048, 16384]normalized: [16384]dtype=f16llama-405b-prefill
• W4x: [1, 8192]normalized: [8192]dtype=bf16llama-70b-decode
• W5x: [2048, 8192]normalized: [8192]dtype=bf16llama-70b-prefill
• W6x: [2048, 8192]normalized: [8192]dtype=f16llama-70b-prefill
• W7x: [1, 4096]normalized: [4096]dtype=bf16llama-8b-decode
• W8x: [2048, 4096]normalized: [4096]dtype=bf16llama-8b-prefill
• W9x: [2048, 4096]normalized: [4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.34×	0.1215	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0214 0.0211	·	·	·
W2	0.43×	2.6197	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	1.1327 1.1786	·	·	·
W3	0.44×	2.5910	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	1.1342 1.1487	·	·	·
W4	0.33×	0.1089	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0355 0.0226	·	·	·
W5	0.28×	1.1604	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.3225 0.3919	·	·	·
W6	0.28×	1.1596	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.3234 0.3954	·	·	·
W7	0.29×	0.1096	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0314 0.0214	·	·	·
W8	0.35×	0.5785	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.2028 0.2076	·	·	·
W9	0.35×	0.5807	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.2030 0.2094	·	·	·

GemmaRMSNormFwd

• W1x: [1, 16384]normalized: [16384]dtype=bf16llama-405b-decode
• W2x: [2048, 16384]normalized: [16384]dtype=bf16llama-405b-prefill
• W3x: [2048, 16384]normalized: [16384]dtype=f16llama-405b-prefill
• W4x: [1, 8192]normalized: [8192]dtype=bf16llama-70b-decode
• W5x: [2048, 8192]normalized: [8192]dtype=bf16llama-70b-prefill
• W6x: [2048, 8192]normalized: [8192]dtype=f16llama-70b-prefill
• W7x: [1, 4096]normalized: [4096]dtype=bf16llama-8b-decode
• W8x: [2048, 4096]normalized: [4096]dtype=bf16llama-8b-prefill
• W9x: [2048, 4096]normalized: [4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.35×	0.1289	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0272 0.0272	·	·	·
W2	0.24×	2.8287	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.7970 0.7252	·	·	·
W3	0.26×	2.8723	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.8133 0.7340	·	·	·
W4	0.32×	0.1190	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0367 0.0255	·	·	·
W5	0.24×	1.1386	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2721 0.2652	·	·	·
W6	0.24×	1.1420	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2715 0.2646	·	·	·
W7	0.27×	0.1215	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0331 0.0235	·	·	·
W8	0.30×	0.5690	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.1720 0.1747	·	·	·
W9	0.31×	0.5677	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.1745 0.1754	·	·	·

QKNormFwd

• W1x: [1, 16384]normalized: [16384]dtype=bf16llama-405b-decode
• W2x: [2048, 16384]normalized: [16384]dtype=bf16llama-405b-prefill
• W3x: [2048, 16384]normalized: [16384]dtype=f16llama-405b-prefill
• W4x: [1, 8192]normalized: [8192]dtype=bf16llama-70b-decode
• W5x: [2048, 8192]normalized: [8192]dtype=bf16llama-70b-prefill
• W6x: [2048, 8192]normalized: [8192]dtype=f16llama-70b-prefill
• W7x: [1, 4096]normalized: [4096]dtype=bf16llama-8b-decode
• W8x: [2048, 4096]normalized: [4096]dtype=bf16llama-8b-prefill
• W9x: [2048, 4096]normalized: [4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.30×	0.2245	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0455 0.0437	·	·	·
W2	0.27×	5.3171	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	1.6066 1.5517	·	·	·
W3	0.27×	5.3952	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	1.6015 1.4194	·	·	·
W4	0.28×	0.2050	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0579 0.0430	·	·	·
W5	0.22×	2.3981	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.5231 0.4988	·	·	·
W6	0.22×	2.3940	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.5114 0.4986	·	·	·
W7	0.26×	0.1995	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0505 0.0405	·	·	·
W8	0.29×	1.1075	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.3285 0.3304	·	·	·
W9	0.29×	1.1240	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.3311 0.3319	·	·	·

RMSNormBwd

• W1x: [1, 16384]normalized: [16384]dtype=bf16llama-405b-decode
• W2x: [2048, 16384]normalized: [16384]dtype=bf16llama-405b-prefill
• W3x: [2048, 16384]normalized: [16384]dtype=f16llama-405b-prefill
• W4x: [1, 8192]normalized: [8192]dtype=bf16llama-70b-decode
• W5x: [2048, 8192]normalized: [8192]dtype=bf16llama-70b-prefill
• W6x: [2048, 8192]normalized: [8192]dtype=f16llama-70b-prefill
• W7x: [1, 4096]normalized: [4096]dtype=bf16llama-8b-decode
• W8x: [2048, 4096]normalized: [4096]dtype=bf16llama-8b-prefill
• W9x: [2048, 4096]normalized: [4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.12×	0.4083	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0348 0.0339	·	·	·
W2	0.32×	6.3724	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	2.0200 2.2234	·	·	·
W3	0.31×	6.4097	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	2.0053 2.2535	·	·	·
W4	0.15×	0.2492	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0387 0.0305	·	·	·
W5	0.29×	2.7129	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.7750 0.9525	·	·	·
W6	0.29×	2.7153	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.7927 0.9451	·	·	·
W7	0.19×	0.1713	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0318 0.0267	·	·	·
W8	0.29×	1.2224	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.3574 0.3711	·	·	·
W9	0.29×	1.2143	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.3554 0.3805	·	·	·

SpectralNormPowerIterFwd

• W1x: [1, 16384]normalized: [16384]dtype=bf16llama-405b-decode
• W2x: [2048, 16384]normalized: [16384]dtype=bf16llama-405b-prefill
• W3x: [2048, 16384]normalized: [16384]dtype=f16llama-405b-prefill
• W4x: [1, 8192]normalized: [8192]dtype=bf16llama-70b-decode
• W5x: [2048, 8192]normalized: [8192]dtype=bf16llama-70b-prefill
• W6x: [2048, 8192]normalized: [8192]dtype=f16llama-70b-prefill
• W7x: [1, 4096]normalized: [4096]dtype=bf16llama-8b-decode
• W8x: [2048, 4096]normalized: [4096]dtype=bf16llama-8b-prefill
• W9x: [2048, 4096]normalized: [4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.11×	0.6388	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0725 0.0729	·	·	·
W2	0.19×	4.6319	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.8709 0.8558	·	·	·
W3	0.19×	4.6324	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.8629 0.8546	·	·	·
W4	0.14×	0.4525	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0607 0.0548	·	·	·
W5	0.21×	1.7875	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.3654 0.3498	·	·	·
W6	0.20×	1.7924	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.3663 0.3508	·	·	·
W7	0.08×	0.3643	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0285 0.0260	·	·	·
W8	0.22×	0.9865	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2136 0.2000	·	·	·
W9	0.22×	0.9859	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2130 0.2003	·	·	·

BatchNormInferenceFwd

dtype=f16

• W1x: [4, 128, 1024, 1024]large-spatial
• W2x: [32, 64]resnet50-fc
• W3x: [8, 64, 32, 32]resnet50-stage1
• W4x: [4, 128, 32, 32]resnet50-stage2
• W5x: [4, 256, 28, 28]resnet50-stage3

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.07×	42.3053	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	3.1336 3.1236	·	·	·
W2	0.08×	0.0796	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.00775 0.0035	·	·	·
W3	0.11×	0.1669	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0168 0.0110	·	·	·
W4	0.10×	0.1704	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0195 0.0120	·	·	·
W5	0.10×	0.1817	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0190 0.0135	·	·	·

LayerNormBwd

• W1x: [1, 16384]normalized: [16384]dtype=bf16llama-405b-decode
• W2x: [2048, 16384]normalized: [16384]dtype=bf16llama-405b-prefill
• W3x: [2048, 16384]normalized: [16384]dtype=f16llama-405b-prefill
• W4x: [1, 8192]normalized: [8192]dtype=bf16llama-70b-decode
• W5x: [2048, 8192]normalized: [8192]dtype=bf16llama-70b-prefill
• W6x: [2048, 8192]normalized: [8192]dtype=f16llama-70b-prefill
• W7x: [1, 4096]normalized: [4096]dtype=bf16llama-8b-decode
• W8x: [2048, 4096]normalized: [4096]dtype=bf16llama-8b-prefill
• W9x: [2048, 4096]normalized: [4096]dtype=f16llama-8b-prefill

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.06×	0.7544	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0314 0.0311	·	·	·
W2	0.05×	9.2674	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.4285 0.4290	·	·	·
W3	0.05×	9.2705	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.4310 0.4308	·	·	·
W4	0.08×	0.4495	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0222 0.0220	·	·	·
W5	0.04×	3.7468	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.1665 0.1669	·	·	·
W6	0.04×	3.7363	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.1586 0.1593	·	·	·
W7	0.09×	0.2985	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0198 0.0201	·	·	·
W8	0.06×	1.7646	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.1080 0.1084	·	·	·
W9	0.06×	1.7297	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1021 0.1020	·	·	·

Convolution

DepthwiseConv2d

input: [N, C_in, H, W]

dtype=f16N=1C_in=32H=56W=56C_out=32kH=3kW=3stride=[1,1]padding=[1,1]

• W1mobilenetv2-depthwise

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.10×	0.0602	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0668 0.0501	·	·	·

PointwiseConv2d

input: [N, C_in, H, W]bias: [C_out]

stride=[1,1]padding=[0,0]

• W1dtype=f16N=2C_in=128H=28W=28C_out=512bottleneck-expand-1x1-bias
• W3dtype=f16N=2C_in=512H=28W=28C_out=128bottleneck-reduce-1x1-bias
• W5dtype=f16N=1C_in=512H=7W=7C_out=2048classifier-1x1-bias
• W7dtype=f16N=1C_in=256H=14W=14C_out=1024late-stage-1x1-bias
• W10dtype=bf16N=2C_in=64H=56W=56C_out=256resnet-1x1-bias
• W11dtype=f16N=2C_in=64H=56W=56C_out=256resnet-1x1-bias

input: [N, C_in, H, W]

stride=[1,1]padding=[0,0]

• W2dtype=f16N=2C_in=128H=28W=28C_out=512bottleneck-expand-1x1
• W4dtype=f16N=2C_in=512H=28W=28C_out=128bottleneck-reduce-1x1
• W6dtype=f16N=1C_in=512H=7W=7C_out=2048classifier-1x1
• W8dtype=f16N=1C_in=256H=14W=14C_out=1024late-stage-1x1
• W9dtype=bf16N=2C_in=64H=56W=56C_out=256resnet-1x1
• W12dtype=f16N=2C_in=64H=56W=56C_out=256resnet-1x1

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.35×	0.1226	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0289 0.0291 0.0286	·	·	·
W2	1.33×	0.0345	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0460 0.0270 0.0304	·	·	·
W3	0.67×	0.0645	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0300 0.0299 0.0297	·	·	·
W4	1.24×	0.0396	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0441 0.0340 0.0316	·	·	·
W5	0.34×	0.2185	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0648 0.0648 0.0648	·	·	·
W6	2.04×	0.0340	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0649 0.0653 0.0649	·	·	·
W7	0.50×	0.1060	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0319 0.0320 0.0316	·	·	·
W8	2.34×	0.0213	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0331 0.0330 0.0328	·	·	·
W9	1.75×	0.0300	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0527 0.0310 0.0362	·	·	·
W10	0.66×	0.0795	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0377 0.0375 0.0377	·	·	·
W11	0.66×	0.0788	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0379 0.0376 0.0379	·	·	·
W12	1.96×	0.0310	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0535 0.0288 0.0355	·	·	·

GroupedConv2d

input: [N, C_in, H, W]

dtype=f16N=1C_in=128H=28W=28C_out=256kH=3kW=3stride=[1,1]padding=[1,1]groups=32

• W1resnext-grouped-3x3

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.57×	0.1125	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0643 0.0330 0.0478	·	·	·

Conv2dWgrad

• W1input: [2, 128, 28, 28]bias: [512]dtype=f16C_out=512kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-expand-1x1-bias
• W3input: [2, 512, 28, 28]bias: [128]dtype=f16C_out=128kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-reduce-1x1-bias
• W5input: [1, 512, 7, 7]bias: [2048]dtype=f16C_out=2048kH=1kW=1stride=[1,1]padding=[0,0]classifier-1x1-bias
• W8input: [1, 256, 112, 112]bias: [512]dtype=f16C_out=512kH=3kW=3stride=[1,1]padding=[1,1]highres-3x3-s1-bias
• W10input: [1, 256, 14, 14]bias: [1024]dtype=f16C_out=1024kH=1kW=1stride=[1,1]padding=[0,0]late-stage-1x1-bias
• W12input: [1, 64, 56, 56]bias: [128]dtype=f16C_out=128kH=5kW=5stride=[1,1]padding=[2,2]midres-5x5-s1-bias
• W16input: [2, 64, 56, 56]bias: [256]dtype=bf16C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1-bias
• W17input: [2, 64, 56, 56]bias: [256]dtype=f16C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1-bias
• W20input: [2, 64, 56, 56]bias: [64]dtype=bf16C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3-bias
• W21input: [2, 64, 56, 56]bias: [64]dtype=f16C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3-bias
• W24input: [1, 128, 56, 56]bias: [256]dtype=f16C_out=256kH=3kW=3stride=[2,2]padding=[1,1]stage-transition-3x3-s2-bias
• W26input: [1, 128, 56, 56]bias: [256]dtype=f16C_out=256kH=5kW=5stride=[2,2]padding=[2,2]stage-transition-5x5-s2-bias
• W28input: [1, 3, 112, 112]bias: [64]dtype=f16C_out=64kH=3kW=3stride=[2,2]padding=[1,1]stem-3x3-s2-bias
• W31input: [1, 128, 28, 28]bias: [128]dtype=bf16C_out=128kH=3kW=3stride=[2,2]padding=[1,1]stride2-bias

input: [N, C_in, H, W]

• W2dtype=f16N=2C_in=128H=28W=28C_out=512kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-expand-1x1
• W4dtype=f16N=2C_in=512H=28W=28C_out=128kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-reduce-1x1
• W6dtype=f16N=1C_in=512H=7W=7C_out=2048kH=1kW=1stride=[1,1]padding=[0,0]classifier-1x1
• W7dtype=f16N=1C_in=2048H=32W=32C_out=256kH=3kW=3stride=[1,1]padding=[12,12]dilation=[12,12]deeplabv3-aspp-3x3-rate12
• W9dtype=f16N=1C_in=256H=112W=112C_out=512kH=3kW=3stride=[1,1]padding=[1,1]highres-3x3-s1
• W11dtype=f16N=1C_in=256H=14W=14C_out=1024kH=1kW=1stride=[1,1]padding=[0,0]late-stage-1x1
• W13dtype=f16N=1C_in=64H=56W=56C_out=128kH=5kW=5stride=[1,1]padding=[2,2]midres-5x5-s1
• W14dtype=f16N=1C_in=32H=56W=56C_out=32kH=3kW=3stride=[1,1]padding=[1,1]groups=32mobilenetv2-depthwise
• W15dtype=bf16N=2C_in=64H=56W=56C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1
• W18dtype=f16N=2C_in=64H=56W=56C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1
• W19dtype=bf16N=2C_in=64H=56W=56C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3
• W22dtype=f16N=2C_in=64H=56W=56C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3
• W23dtype=f16N=1C_in=128H=28W=28C_out=256kH=3kW=3stride=[1,1]padding=[1,1]groups=32resnext-grouped-3x3
• W25dtype=f16N=1C_in=128H=56W=56C_out=256kH=3kW=3stride=[2,2]padding=[1,1]stage-transition-3x3-s2
• W27dtype=f16N=1C_in=128H=56W=56C_out=256kH=5kW=5stride=[2,2]padding=[2,2]stage-transition-5x5-s2
• W29dtype=f16N=1C_in=3H=112W=112C_out=64kH=3kW=3stride=[2,2]padding=[1,1]stem-3x3-s2
• W30dtype=bf16N=1C_in=128H=28W=28C_out=128kH=3kW=3stride=[2,2]padding=[1,1]stride2

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.25×	0.0668	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0767 0.0776	·	·	·
W2	1.37×	0.0620	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0780 0.0770	·	·	·
W3	0.86×	0.0890	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0622 0.0624	·	·	·
W4	0.98×	0.0849	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0626 0.0622	·	·	·
W5	1.08×	0.1025	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.1098 0.1101	·	·	·
W6	1.07×	0.1030	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.1096 0.1101	·	·	·
W7	0.03×	7.2436	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1727 0.1722	·	·	·
W8	0.41×	0.5530	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2454 0.2139	·	·	·
W9	0.40×	0.5524	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2249 0.2129	·	·	·
W10	2.21×	0.0390	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0749 0.0750	·	·	·
W11	2.26×	0.0410	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0747 0.0736	·	·	·
W12	0.39×	0.2480	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1085 0.0809	·	·	·
W13	0.38×	0.2477	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0960 0.0808	·	·	·
W14	0.28×	0.3775	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0881 0.0880	·	·	·
W15	0.99×	0.0737	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0640 0.0648	·	·	·
W16	0.93×	0.0785	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0636 0.0643	·	·	·
W17	1.08×	0.0772	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0640 0.0640	·	·	·
W18	1.01×	0.0727	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0635 0.0638	·	·	·
W19	0.42×	0.1955	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0841 0.0683	·	·	·
W20	0.45×	0.1970	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0989 0.0671	·	·	·
W21	0.45×	0.1975	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1035 0.0675	·	·	·
W22	0.45×	0.1958	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0910 0.0674	·	·	·
W23	0.43×	0.2410	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0869 0.0858	·	·	·
W24	0.03×	2.9286	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1056 0.0752	·	·	·
W25	0.03×	2.9285	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0935 0.0762	·	·	·
W26	0.01×	8.2952	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1131 0.0810	·	·	·
W27	0.01×	8.2994	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1015 0.0801	·	·	·
W28	0.24×	0.3060	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0831 0.0546	·	·	·
W29	0.24×	0.3055	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0710 0.0541	·	·	·
W30	0.11×	0.7768	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0855 0.0689	·	·	·
W31	0.11×	0.7791	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0989 0.0688	·	·	·

Conv2dFp8Fwd

input: [N, C_in, H, W]

dtype=fp8e4m3

• W1N=2C_in=64H=56W=56C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1
• W2N=2C_in=64H=56W=56C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3
• W3N=1C_in=128H=28W=28C_out=128kH=3kW=3stride=[2,2]padding=[1,1]stride2

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	2.32×	0.0415	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0973 0.1960 0.0858	·	·	·
W2	0.08×	1.2181	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0965 0.2034 0.0853	·	·	·
W3	0.05×	1.8235	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0902 0.1270 0.0810	·	·	·

Conv2dFwd

input: [N, C_in, H, W]bias: [C_out]

• W1dtype=f16N=2C_in=128H=28W=28C_out=512kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-expand-1x1-bias
• W3dtype=f16N=2C_in=512H=28W=28C_out=128kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-reduce-1x1-bias
• W5dtype=f16N=1C_in=512H=7W=7C_out=2048kH=1kW=1stride=[1,1]padding=[0,0]classifier-1x1-bias
• W14dtype=f16N=1C_in=256H=112W=112C_out=512kH=3kW=3stride=[1,1]padding=[1,1]highres-3x3-s1-bias
• W17dtype=f16N=1C_in=256H=14W=14C_out=1024kH=1kW=1stride=[1,1]padding=[0,0]late-stage-1x1-bias
• W19dtype=f16N=1C_in=64H=56W=56C_out=128kH=5kW=5stride=[1,1]padding=[2,2]midres-5x5-s1-bias
• W23dtype=bf16N=2C_in=64H=56W=56C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1-bias
• W24dtype=f16N=2C_in=64H=56W=56C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1-bias
• W27dtype=bf16N=2C_in=64H=56W=56C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3-bias
• W28dtype=f16N=2C_in=64H=56W=56C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3-bias
• W31dtype=f16N=1C_in=128H=56W=56C_out=256kH=3kW=3stride=[2,2]padding=[1,1]stage-transition-3x3-s2-bias
• W33dtype=f16N=1C_in=128H=56W=56C_out=256kH=5kW=5stride=[2,2]padding=[2,2]stage-transition-5x5-s2-bias
• W35dtype=f16N=1C_in=3H=112W=112C_out=64kH=3kW=3stride=[2,2]padding=[1,1]stem-3x3-s2-bias
• W39dtype=bf16N=1C_in=128H=28W=28C_out=128kH=3kW=3stride=[2,2]padding=[1,1]stride2-bias

input: [N, C_in, H, W]

• W2dtype=f16N=2C_in=128H=28W=28C_out=512kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-expand-1x1
• W4dtype=f16N=2C_in=512H=28W=28C_out=128kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-reduce-1x1
• W6dtype=f16N=1C_in=512H=7W=7C_out=2048kH=1kW=1stride=[1,1]padding=[0,0]classifier-1x1
• W7dtype=f16N=1C_in=2048H=32W=32C_out=256kH=3kW=3stride=[1,1]padding=[12,12]dilation=[12,12]deeplabv3-aspp-3x3-rate12
• W15dtype=f16N=1C_in=256H=112W=112C_out=512kH=3kW=3stride=[1,1]padding=[1,1]highres-3x3-s1
• W18dtype=f16N=1C_in=256H=14W=14C_out=1024kH=1kW=1stride=[1,1]padding=[0,0]late-stage-1x1
• W20dtype=f16N=1C_in=64H=56W=56C_out=128kH=5kW=5stride=[1,1]padding=[2,2]midres-5x5-s1
• W21dtype=f16N=1C_in=32H=56W=56C_out=32kH=3kW=3stride=[1,1]padding=[1,1]groups=32mobilenetv2-depthwise
• W22dtype=bf16N=2C_in=64H=56W=56C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1
• W25dtype=f16N=2C_in=64H=56W=56C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1
• W26dtype=bf16N=2C_in=64H=56W=56C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3
• W29dtype=f16N=2C_in=64H=56W=56C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3
• W30dtype=f16N=1C_in=128H=28W=28C_out=256kH=3kW=3stride=[1,1]padding=[1,1]groups=32resnext-grouped-3x3
• W32dtype=f16N=1C_in=128H=56W=56C_out=256kH=3kW=3stride=[2,2]padding=[1,1]stage-transition-3x3-s2
• W34dtype=f16N=1C_in=128H=56W=56C_out=256kH=5kW=5stride=[2,2]padding=[2,2]stage-transition-5x5-s2
• W36dtype=f16N=1C_in=3H=112W=112C_out=64kH=3kW=3stride=[2,2]padding=[1,1]stem-3x3-s2
• W38dtype=bf16N=1C_in=128H=28W=28C_out=128kH=3kW=3stride=[2,2]padding=[1,1]stride2

dtype=f16

• W8input_shape: [1, 16, 8, 8]weight_shape: [32, 16, 3, 3]default-dense-cube-fp16
• W16input_shape: [16, 256, 56, 56]weight_shape: [256, 1, 1, 1]groups=256large-grid-depthwise-fp16

• W9input_shape: [1, 16, 8, 8]padding: [1, 1]weight_shape: [16, 16, 3, 3]dtype=bf16dense-cube-bf16
• W10input_shape: [1, 4, 8, 9]padding: [1, 1]weight_shape: [4, 1, 3, 3]dtype=f16groups=4depthwise-direct-fp16
• W12input_shape: [1, 2, 5, 6]padding: [1, 1]weight_shape: [3, 2, 3, 3]dtype=f32fp32-direct
• W13input_shape: [1, 8, 9, 10]padding: [1, 1]weight_shape: [8, 2, 3, 3]dtype=f16groups=4grouped-direct-fp16

dtype=f16

• W11dilation, padding: [2, 2]input_shape: [1, 16, 9, 10]weight_shape: [16, 16, 3, 3]dilation-cube-fp16

dtype=f16

• W37input_shape: [1, 3, 10, 11]padding: [1, 1]stride: [2, 2]weight_shape: [16, 3, 3, 3]stride-padding-nondiv-cube-fp16

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.38×	0.1185	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0306 0.0305	·	·	·
W2	1.33×	0.0354	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0318 0.0314	·	·	·
W3	0.71×	0.0635	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0314 0.0309	·	·	·
W4	1.18×	0.0395	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0310 0.0309	·	·	·
W5	0.33×	0.2105	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0645 0.0640	·	·	·
W6	1.97×	0.0330	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0640 0.0643	·	·	·
W7	0.00×	79.0235	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.1179 0.1181	·	·	·
W8	0.38×	0.0881	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0267 0.0264	·	·	·
W9	0.57×	0.0540	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0219 0.0206	·	·	·
W10	0.66×	0.0840	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0444 0.0432	·	·	·
W11	0.28×	0.1180	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0214 0.0200	·	·	·
W12	0.21×	0.1457	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0262 0.0259	·	·	·
W13	0.31×	0.1595	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0372 0.0360	·	·	·
W14	0.01×	29.0843	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2147 0.2147	·	·	·
W15	0.01×	29.0799	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2286 0.2157	·	·	·
W16	0.36×	0.4460	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.1510 0.1514	·	·	·
W17	0.41×	0.1062	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0324 0.0326	·	·	·
W18	2.78×	0.0200	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0345 0.0345	·	·	·
W19	0.02×	4.2020	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0460 0.0459	·	·	·
W20	0.01×	4.2009	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0589 0.0461	·	·	·
W21	0.94×	0.0597	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0486 0.0494	·	·	·
W22	1.66×	0.0294	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0377 0.0374	·	·	·
W23	0.71×	0.0707	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0380 0.0380	·	·	·
W24	0.72×	0.0688	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0380 0.0375	·	·	·
W25	1.63×	0.0276	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0372 0.0375	·	·	·
W26	0.04×	1.2043	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0485 0.0384	·	·	·
W27	0.05×	1.2067	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0382 0.0381	·	·	·
W28	0.05×	1.2054	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0396 0.0390	·	·	·
W29	0.05×	1.2069	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0498 0.0385	·	·	·
W30	0.45×	0.1130	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0473 0.0475	·	·	·
W31	0.01×	8.3829	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0387 0.0380	·	·	·
W32	0.01×	8.3692	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0515 0.0380	·	·	·
W33	0.00×	24.0599	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0558 0.0551	·	·	·
W34	0.00×	24.0326	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0590 0.0462	·	·	·
W35	0.13×	0.4036	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0320 0.0318	·	·	·
W36	0.11×	0.3975	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0454 0.0326	·	·	·
W37	0.95×	0.0367	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0265 0.0257	·	·	·
W38	0.03×	1.8193	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0460 0.0335	·	·	·
W39	0.02×	1.8265	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0335 0.0328	·	·	·

Conv1dFwd

dtype=f16

• W1input_shape: [1, 4, 17]weight_shape: [8, 2, 3]dilation=2groups=2padding=2dilation-groups
• W10input_shape: [1, 2, 17]weight_shape: [4, 2, 3]padding=1stride=2stride-padding

input: [N, C_in, L_in]

N=1

• W2dtype=bf16C_in=128L_in=600C_out=256kW=10encodec-deep
• W5dtype=f16C_in=128L_in=600C_out=256kW=10encodec-deep
• W6dtype=bf16C_in=1L_in=24000C_out=32kW=7encodec-init
• W9dtype=f16C_in=1L_in=24000C_out=32kW=7encodec-init
• W11dtype=bf16C_in=1L_in=16000C_out=512kW=10stride=5wav2vec2-layer1
• W14dtype=f16C_in=1L_in=16000C_out=512kW=10stride=5wav2vec2-layer1
• W15dtype=bf16C_in=80L_in=3000C_out=1280kW=3whisper-large-conv1
• W18dtype=f16C_in=80L_in=3000C_out=1280kW=3whisper-large-conv1

input: [N, C_in, L_in]bias: [C_out]

N=1

• W3dtype=bf16C_in=128L_in=600C_out=256kW=10encodec-deep-bias
• W4dtype=f16C_in=128L_in=600C_out=256kW=10encodec-deep-bias
• W7dtype=bf16C_in=1L_in=24000C_out=32kW=7encodec-init-bias
• W8dtype=f16C_in=1L_in=24000C_out=32kW=7encodec-init-bias
• W12dtype=bf16C_in=1L_in=16000C_out=512kW=10stride=5wav2vec2-layer1-bias
• W13dtype=f16C_in=1L_in=16000C_out=512kW=10stride=5wav2vec2-layer1-bias
• W16dtype=bf16C_in=80L_in=3000C_out=1280kW=3whisper-large-conv1-bias
• W17dtype=f16C_in=80L_in=3000C_out=1280kW=3whisper-large-conv1-bias

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.99×	0.0527	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0424 0.0415	·	·	·
W2	0.04×	1.3505	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0470 0.0365	·	·	·
W3	0.03×	1.3445	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0384 0.0386	·	·	·
W4	0.04×	1.3594	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0377 0.0371	·	·	·
W5	0.03×	1.3445	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0473 0.0375	·	·	·
W6	1.03×	0.0320	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0328 0.0203	·	·	·
W7	1.02×	0.0323	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0213 0.0210	·	·	·
W8	1.60×	0.0304	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0328 0.0325	·	·	·
W9	1.79×	0.0269	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0478 0.0326	·	·	·
W10	1.21×	0.0320	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0302 0.0290	·	·	·
W11	0.00×	41.5343	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0350 0.0245	·	·	·
W12	0.00×	42.7620	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0257 0.0249	·	·	·
W13	0.00×	42.7612	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0425 0.0424	·	·	·
W14	0.00×	41.5373	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0532 0.0425	·	·	·
W15	0.04×	1.6562	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0721 0.0609	·	·	·
W16	0.04×	1.6618	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0622 0.0626	·	·	·
W17	0.05×	1.6615	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0607 0.0605	·	·	·
W18	0.04×	1.6499	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0715 0.0606	·	·	·

Conv2dDgrad

• W1input: [2, 128, 28, 28]bias: [512]dtype=f16C_out=512kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-expand-1x1-bias
• W3input: [2, 512, 28, 28]bias: [128]dtype=f16C_out=128kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-reduce-1x1-bias
• W5input: [1, 512, 7, 7]bias: [2048]dtype=f16C_out=2048kH=1kW=1stride=[1,1]padding=[0,0]classifier-1x1-bias
• W8input: [1, 256, 112, 112]bias: [512]dtype=f16C_out=512kH=3kW=3stride=[1,1]padding=[1,1]highres-3x3-s1-bias
• W10input: [1, 256, 14, 14]bias: [1024]dtype=f16C_out=1024kH=1kW=1stride=[1,1]padding=[0,0]late-stage-1x1-bias
• W12input: [1, 64, 56, 56]bias: [128]dtype=f16C_out=128kH=5kW=5stride=[1,1]padding=[2,2]midres-5x5-s1-bias
• W16input: [2, 64, 56, 56]bias: [256]dtype=bf16C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1-bias
• W17input: [2, 64, 56, 56]bias: [256]dtype=f16C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1-bias
• W20input: [2, 64, 56, 56]bias: [64]dtype=bf16C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3-bias
• W21input: [2, 64, 56, 56]bias: [64]dtype=f16C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3-bias
• W24input: [1, 128, 56, 56]bias: [256]dtype=f16C_out=256kH=3kW=3stride=[2,2]padding=[1,1]stage-transition-3x3-s2-bias
• W26input: [1, 128, 56, 56]bias: [256]dtype=f16C_out=256kH=5kW=5stride=[2,2]padding=[2,2]stage-transition-5x5-s2-bias
• W28input: [1, 3, 112, 112]bias: [64]dtype=f16C_out=64kH=3kW=3stride=[2,2]padding=[1,1]stem-3x3-s2-bias
• W31input: [1, 128, 28, 28]bias: [128]dtype=bf16C_out=128kH=3kW=3stride=[2,2]padding=[1,1]stride2-bias

• W2input: [2, 128, 28, 28]dtype=f16C_out=512kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-expand-1x1
• W4input: [2, 512, 28, 28]dtype=f16C_out=128kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-reduce-1x1
• W6input: [1, 512, 7, 7]dtype=f16C_out=2048kH=1kW=1stride=[1,1]padding=[0,0]classifier-1x1
• W7input: [1, 2048, 32, 32]dtype=f16C_out=256kH=3kW=3stride=[1,1]padding=[12,12]dilation=[12,12]deeplabv3-aspp-3x3-rate12
• W9input: [1, 256, 112, 112]dtype=f16C_out=512kH=3kW=3stride=[1,1]padding=[1,1]highres-3x3-s1
• W11input: [1, 256, 14, 14]dtype=f16C_out=1024kH=1kW=1stride=[1,1]padding=[0,0]late-stage-1x1
• W13input: [1, 64, 56, 56]dtype=f16C_out=128kH=5kW=5stride=[1,1]padding=[2,2]midres-5x5-s1
• W14input: [1, 32, 56, 56]dtype=f16C_out=32kH=3kW=3stride=[1,1]padding=[1,1]groups=32mobilenetv2-depthwise
• W15input: [2, 64, 56, 56]dtype=bf16C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1
• W18input: [2, 64, 56, 56]dtype=f16C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1
• W19input: [2, 64, 56, 56]dtype=bf16C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3
• W22input: [2, 64, 56, 56]dtype=f16C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3
• W23input: [1, 128, 28, 28]dtype=f16C_out=256kH=3kW=3stride=[1,1]padding=[1,1]groups=32resnext-grouped-3x3
• W25input: [1, 128, 56, 56]dtype=f16C_out=256kH=3kW=3stride=[2,2]padding=[1,1]stage-transition-3x3-s2
• W27input: [1, 128, 56, 56]dtype=f16C_out=256kH=5kW=5stride=[2,2]padding=[2,2]stage-transition-5x5-s2
• W29input: [1, 3, 112, 112]dtype=f16C_out=64kH=3kW=3stride=[2,2]padding=[1,1]stem-3x3-s2
• W30input: [1, 128, 28, 28]dtype=bf16C_out=128kH=3kW=3stride=[2,2]padding=[1,1]stride2

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.20×	0.3254	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0398 0.0393 0.0390	·	·	·
W2	0.22×	0.3274	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0399 0.0398 0.0393	·	·	·
W3	0.21×	0.3247	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0447 0.0444 0.0447	·	·	·
W4	0.23×	0.3244	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0449 0.0445 0.0445	·	·	·
W5	0.02×	4.0235	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0861 0.0865 0.0861	·	·	·
W6	0.02×	4.0235	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0865 0.0866 0.0865	·	·	·
W7	0.00×	81.0637	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.1273 0.1280 0.1276	·	·	·
W8	0.01×	37.9130	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.2029 0.2030 0.2030	·	·	·
W9	0.01×	37.9222	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.2020 0.2019 0.2023	·	·	·
W10	0.07×	1.0453	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0423 0.0423 0.0420	·	·	·
W11	0.07×	1.0428	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0421 0.0420 0.0421	·	·	·
W12	0.01×	7.8991	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0594 0.0592 0.0596	·	·	·
W13	0.01×	7.9009	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0597 0.0594 0.0594	·	·	·
W14	0.74×	0.0963	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0646 0.0646 0.0650	·	·	·
W15	0.65×	0.1133	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0548 0.0542 0.0546	·	·	·
W16	0.59×	0.1138	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0545 0.0548 0.0542	·	·	·
W17	0.61×	0.1123	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0548 0.0550 0.0548	·	·	·
W18	0.67×	0.1123	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0548 0.0548 0.0545	·	·	·
W19	0.05×	1.3177	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0689 0.0305 0.0537	·	·	·
W20	0.05×	1.3157	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0833 0.0305 0.0539	·	·	·
W21	0.05×	1.3166	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0785 0.0307 0.0540	·	·	·
W22	0.05×	1.3150	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0680 0.0310 0.0548	·	·	·
W23	0.39×	0.2032	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0626 0.0628 0.0617	·	·	·
W24	0.01×	6.9001	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0900 0.0750 0.0638	·	·	·
W25	0.01×	6.9005	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0785 0.0747 0.0639	·	·	·
W26	0.00×	20.2512	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0829 0.0830 0.0828	·	·	·
W27	0.00×	20.2491	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0828 0.0829 0.0831	·	·	·
W28	0.02×	2.7498	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0747 0.0532 0.0455	·	·	·
W29	0.02×	2.7490	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0635 0.0535 0.0457	·	·	·
W30	0.03×	2.2116	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0437 0.0430 0.0437	·	·	·
W31	0.03×	2.2049	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0437 0.0435 0.0435	·	·	·

Conv2dTranspose

input: [N, C_out, out_H, out_W]bias: [C_in]

• W1dtype=f16N=2C_out=128out_H=28out_W=28C_in=512kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-expand-1x1-bias
• W3dtype=f16N=2C_out=512out_H=28out_W=28C_in=128kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-reduce-1x1-bias
• W5dtype=f16N=1C_out=512out_H=7out_W=7C_in=2048kH=1kW=1stride=[1,1]padding=[0,0]classifier-1x1-bias
• W8dtype=f16N=1C_out=256out_H=112out_W=112C_in=512kH=3kW=3stride=[1,1]padding=[1,1]highres-3x3-s1-bias
• W10dtype=f16N=1C_out=256out_H=14out_W=14C_in=1024kH=1kW=1stride=[1,1]padding=[0,0]late-stage-1x1-bias
• W12dtype=f16N=1C_out=64out_H=56out_W=56C_in=128kH=5kW=5stride=[1,1]padding=[2,2]midres-5x5-s1-bias
• W16dtype=bf16N=2C_out=64out_H=56out_W=56C_in=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1-bias
• W17dtype=f16N=2C_out=64out_H=56out_W=56C_in=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1-bias
• W20dtype=bf16N=2C_out=64out_H=56out_W=56C_in=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3-bias
• W21dtype=f16N=2C_out=64out_H=56out_W=56C_in=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3-bias
• W24dtype=f16N=1C_out=128out_H=56out_W=56C_in=256kH=3kW=3stride=[2,2]padding=[1,1]stage-transition-3x3-s2-bias
• W26dtype=f16N=1C_out=128out_H=56out_W=56C_in=256kH=5kW=5stride=[2,2]padding=[2,2]stage-transition-5x5-s2-bias
• W28dtype=f16N=1C_out=3out_H=112out_W=112C_in=64kH=3kW=3stride=[2,2]padding=[1,1]stem-3x3-s2-bias
• W31dtype=bf16N=1C_out=128out_H=28out_W=28C_in=128kH=3kW=3stride=[2,2]padding=[1,1]stride2-bias

input: [N, C_out, out_H, out_W]

• W2dtype=f16N=2C_out=128out_H=28out_W=28kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-expand-1x1
• W4dtype=f16N=2C_out=512out_H=28out_W=28kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-reduce-1x1
• W6dtype=f16N=1C_out=512out_H=7out_W=7kH=1kW=1stride=[1,1]padding=[0,0]classifier-1x1
• W7dtype=f16N=1C_out=2048out_H=32out_W=32kH=3kW=3stride=[1,1]padding=[12,12]dilation=[12,12]deeplabv3-aspp-3x3-rate12
• W9dtype=f16N=1C_out=256out_H=112out_W=112kH=3kW=3stride=[1,1]padding=[1,1]highres-3x3-s1
• W11dtype=f16N=1C_out=256out_H=14out_W=14kH=1kW=1stride=[1,1]padding=[0,0]late-stage-1x1
• W13dtype=f16N=1C_out=64out_H=56out_W=56kH=5kW=5stride=[1,1]padding=[2,2]midres-5x5-s1
• W14dtype=f16N=1C_out=32out_H=56out_W=56kH=3kW=3stride=[1,1]padding=[1,1]groups=32mobilenetv2-depthwise
• W15dtype=bf16N=2C_out=64out_H=56out_W=56kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1
• W18dtype=f16N=2C_out=64out_H=56out_W=56kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1
• W19dtype=bf16N=2C_out=64out_H=56out_W=56kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3
• W22dtype=f16N=2C_out=64out_H=56out_W=56kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3
• W23dtype=f16N=1C_out=128out_H=28out_W=28kH=3kW=3stride=[1,1]padding=[1,1]groups=32resnext-grouped-3x3
• W25dtype=f16N=1C_out=128out_H=56out_W=56kH=3kW=3stride=[2,2]padding=[1,1]stage-transition-3x3-s2
• W27dtype=f16N=1C_out=128out_H=56out_W=56kH=5kW=5stride=[2,2]padding=[2,2]stage-transition-5x5-s2
• W29dtype=f16N=1C_out=3out_H=112out_W=112kH=3kW=3stride=[2,2]padding=[1,1]stem-3x3-s2
• W30dtype=bf16N=1C_out=128out_H=28out_W=28kH=3kW=3stride=[2,2]padding=[1,1]stride2

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.14×	0.3599	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0323 0.0324 0.0320	·	·	·
W2	0.15×	0.3319	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0307 0.0305 0.0306	·	·	·
W3	0.12×	0.4286	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0336 0.0329 0.0324	·	·	·
W4	0.15×	0.3354	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0314 0.0304 0.0301	·	·	·
W5	0.02×	4.0781	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0712 0.0712 0.0707	·	·	·
W6	0.02×	4.0293	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0740 0.0742 0.0741	·	·	·
W7	0.00×	81.0684	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.1030 0.1031 0.1035	·	·	·
W8	0.01×	37.9281	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.2189 0.2188 0.2185	·	·	·
W9	0.01×	37.9181	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.1830 0.1830 0.1832	·	·	·
W10	0.05×	1.0715	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0359 0.0360 0.0352	·	·	·
W11	0.06×	1.0448	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0360 0.0357 0.0359	·	·	·
W12	0.01×	7.9056	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0574 0.0570 0.0573	·	·	·
W13	0.01×	7.8979	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0455 0.0455 0.0457	·	·	·
W14	0.71×	0.0935	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0503 0.0498 0.0500	·	·	·
W15	0.47×	0.1206	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0398 0.0395 0.0396	·	·	·
W16	0.37×	0.1385	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0419 0.0420 0.0424	·	·	·
W17	0.41×	0.1368	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0404 0.0403 0.0400	·	·	·
W18	0.40×	0.1215	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0403 0.0403 0.0406	·	·	·
W19	0.04×	1.3161	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0564 0.0312 0.0425	·	·	·
W20	0.04×	1.3156	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0441 0.0440 0.0442	·	·	·
W21	0.04×	1.3186	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0450 0.0442 0.0447	·	·	·
W22	0.05×	1.3177	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0590 0.0307 0.0405	·	·	·
W23	0.31×	0.2030	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0539 0.0539 0.0541	·	·	·
W24	0.01×	6.9045	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0542 0.0542 0.0542	·	·	·
W25	0.01×	6.8994	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0620 0.0382 0.0461	·	·	·
W26	0.01×	20.2496	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.1381 0.1375 0.1376	·	·	·
W27	0.00×	20.2524	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0659 0.0656 0.0653	·	·	·
W28	0.02×	2.7501	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0370 0.0371 0.0371	·	·	·
W29	0.02×	2.7475	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0541 0.0297 0.0374	·	·	·
W30	0.02×	2.2070	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0375 0.0370 0.0371	·	·	·
W31	0.02×	2.2314	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0365 0.0356 0.0359	·	·	·

Conv2dBiasRelu

input: [N, C_in, H, W]bias: [C_out]

• W1dtype=f16N=2C_in=128H=28W=28C_out=512kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-expand-1x1-bias
• W2dtype=f16N=2C_in=512H=28W=28C_out=128kH=1kW=1stride=[1,1]padding=[0,0]bottleneck-reduce-1x1-bias
• W3dtype=f16N=1C_in=512H=7W=7C_out=2048kH=1kW=1stride=[1,1]padding=[0,0]classifier-1x1-bias
• W4dtype=f16N=1C_in=256H=112W=112C_out=512kH=3kW=3stride=[1,1]padding=[1,1]highres-3x3-s1-bias
• W5dtype=f16N=1C_in=256H=14W=14C_out=1024kH=1kW=1stride=[1,1]padding=[0,0]late-stage-1x1-bias
• W6dtype=f16N=1C_in=64H=56W=56C_out=128kH=5kW=5stride=[1,1]padding=[2,2]midres-5x5-s1-bias
• W7dtype=bf16N=2C_in=64H=56W=56C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1-bias
• W8dtype=f16N=2C_in=64H=56W=56C_out=256kH=1kW=1stride=[1,1]padding=[0,0]resnet-1x1-bias
• W9dtype=bf16N=2C_in=64H=56W=56C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3-bias
• W10dtype=f16N=2C_in=64H=56W=56C_out=64kH=3kW=3stride=[1,1]padding=[1,1]resnet-3x3-bias
• W11dtype=f16N=1C_in=128H=56W=56C_out=256kH=3kW=3stride=[2,2]padding=[1,1]stage-transition-3x3-s2-bias
• W12dtype=f16N=1C_in=128H=56W=56C_out=256kH=5kW=5stride=[2,2]padding=[2,2]stage-transition-5x5-s2-bias
• W13dtype=f16N=1C_in=3H=112W=112C_out=64kH=3kW=3stride=[2,2]padding=[1,1]stem-3x3-s2-bias
• W14dtype=bf16N=1C_in=128H=28W=28C_out=128kH=3kW=3stride=[2,2]padding=[1,1]stride2-bias

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.03×	1.6334	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0360 0.0356 0.0357	·	·	·
W2	0.11×	0.4725	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0357 0.0354 0.0354	·	·	·
W3	0.19×	0.4158	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0775 0.0775 0.0772	·	·	·
W4	0.01×	41.1983	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.2266 0.2264 0.2268	·	·	·
W5	0.12×	0.5000	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0441 0.0437 0.0434	·	·	·
W6	0.01×	4.9813	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0585 0.0586 0.0590	·	·	·
W7	0.02×	3.0994	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0432 0.0435 0.0434	·	·	·
W8	0.02×	3.0996	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0431 0.0430 0.0431	·	·	·
W9	0.03×	1.9804	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0690 0.0300 0.0503	·	·	·
W10	0.03×	1.9825	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0629 0.0305 0.0503	·	·	·
W11	0.01×	8.7918	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0607 0.0475 0.0495	·	·	·
W12	0.00×	24.4659	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0668 0.0665 0.0665	·	·	·
W13	0.07×	0.8106	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0563 0.0260 0.0450	·	·	·
W14	0.03×	1.9054	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0377 0.0375 0.0375	·	·	·

Conv1dCausal

input: [N, C_in, L_in]

N=1

• W1dtype=bf16C_in=128L_in=600C_out=256kW=10encodec-deep
• W4dtype=f16C_in=128L_in=600C_out=256kW=10encodec-deep
• W5dtype=bf16C_in=1L_in=24000C_out=32kW=7encodec-init
• W8dtype=f16C_in=1L_in=24000C_out=32kW=7encodec-init
• W9dtype=bf16C_in=1L_in=16000C_out=512kW=10stride=5wav2vec2-layer1
• W12dtype=f16C_in=1L_in=16000C_out=512kW=10stride=5wav2vec2-layer1
• W13dtype=bf16C_in=80L_in=3000C_out=1280kW=3whisper-large-conv1
• W16dtype=f16C_in=80L_in=3000C_out=1280kW=3whisper-large-conv1

input: [N, C_in, L_in]bias: [C_out]

N=1

• W2dtype=bf16C_in=128L_in=600C_out=256kW=10encodec-deep-bias
• W3dtype=f16C_in=128L_in=600C_out=256kW=10encodec-deep-bias
• W6dtype=bf16C_in=1L_in=24000C_out=32kW=7encodec-init-bias
• W7dtype=f16C_in=1L_in=24000C_out=32kW=7encodec-init-bias
• W10dtype=bf16C_in=1L_in=16000C_out=512kW=10stride=5wav2vec2-layer1-bias
• W11dtype=f16C_in=1L_in=16000C_out=512kW=10stride=5wav2vec2-layer1-bias
• W14dtype=bf16C_in=80L_in=3000C_out=1280kW=3whisper-large-conv1-bias
• W15dtype=f16C_in=80L_in=3000C_out=1280kW=3whisper-large-conv1-bias

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.01×	5.2867	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0537 0.0545 0.0542	·	·	·
W2	0.01×	5.3150	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0553 0.0553 0.0559	·	·	·
W3	0.01×	5.3111	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0545 0.0542 0.0545	·	·	·
W4	0.01×	5.2912	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0544 0.0548 0.0540	·	·	·
W5	0.09×	0.6435	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0437 0.0437 0.0440	·	·	·
W6	0.10×	0.6607	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0447 0.0447 0.0447	·	·	·
W7	0.12×	0.6617	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0591 0.0592 0.0586	·	·	·
W8	0.11×	0.6438	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0587 0.0589 0.0592	·	·	·
W9	0.06×	1.0870	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0703 0.0425 0.0510	·	·	·
W10	0.06×	1.0490	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0529 0.0530 0.0529	·	·	·
W11	0.08×	1.0466	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0693 0.0688 0.0689	·	·	·
W12	0.08×	1.1024	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0969 0.0452 0.0690	·	·	·
W13	0.00×	24.3873	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.1008 0.0573 0.0853	·	·	·
W14	0.00×	24.4970	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0886 0.0885 0.0885	·	·	·
W15	0.00×	24.4935	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0875 0.0872 0.0876	·	·	·
W16	0.00×	24.3831	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.1013 0.0560 0.0854	·	·	·

Conv3dFwd

input: [N, C_in, D, H, W]

N=1

• W1dtype=f16C_in=64D=8H=28W=28C_out=128stride=[1,1,1]padding=[1,1,1]groups=323d-resnext-grouped-k3
• W2dtype=f16C_in=256D=8H=16W=16C_out=256stride=[1,1,1]padding=[6,6,6]dilation=[6,6,6]3d-unet-aspp-3x3x3-rate6
• W6dtype=f16C_in=3D=16H=112W=112C_out=64stride=[1,1,1]padding=[1,1,1]r3d-stem-k3-s1
• W7dtype=bf16C_in=32D=32H=64W=64C_out=64stride=[1,1,1]padding=[1,1,1]unet-encoder-k3-s1
• W10dtype=f16C_in=64D=8H=56W=56C_out=128stride=[2,2,2]padding=[1,1,1]video-stage-downsample-k3-s2

dtype=f16

• W3input_shape: [1, 2, 4, 5, 6]weight_shape: [4, 2, 3, 3, 3]padding=1default-3d
• W4input_shape: [1, 1, 258, 256, 128]weight_shape: [1, 1, 1, 1, 1]grid-stride-sentinel

input: [N, C_in, D, H, W]bias: [C_out]

N=1padding=[1,1,1]

• W5dtype=f16C_in=3D=16H=112W=112C_out=64stride=[1,1,1]r3d-stem-k3-s1-bias
• W8dtype=bf16C_in=32D=32H=64W=64C_out=64stride=[1,1,1]unet-encoder-k3-s1-bias
• W9dtype=f16C_in=64D=8H=56W=56C_out=128stride=[2,2,2]video-stage-downsample-k3-s2-bias

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.07×	1.8140	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1353 0.1104	·	·	·
W2	0.00×	169.8936	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1353 0.1249	·	·	·
W3	0.10×	0.4829	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0447 0.0446	·	·	·
W4	4.03×	0.2365	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.9497 0.9509	·	·	·
W5	0.04×	5.1566	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2110 0.2042	·	·	·
W6	0.04×	5.1413	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.2028 0.2014	·	·	·
W7	0.00×	39.7941	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1767 0.1659	·	·	·
W8	0.00×	39.7875	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.1770 0.1681	·	·	·
W9	0.00×	2022.7671	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0886 0.0824	·	·	·
W10	0.00×	2021.8212	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0914 0.0833	·	·	·

DilatedConv2d

input: [N, C_in, H, W]

dtype=f16N=1C_in=2048H=32W=32C_out=256kH=3kW=3stride=[1,1]padding=[12,12]dilation=[12,12]

• W1deeplabv3-aspp-3x3-rate12

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.00×	78.9945	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.1340 0.1072 0.1190	·	·	·

Pooling

AvgPool3dFwd

input: [N, C, D_in, H_in, W_in]

stride=[2,2,2]

• W1dtype=f16N=2C=64D_in=8H_in=28W_in=28kernel_size=[2,3,3]padding=[1,1,1]ceil_mode=truecount_include_pad=falseceil-video
• W2dtype=bf16N=2C=24D_in=10H_in=20W_in=22kernel_size=[2,2,3]padding=[0,1,1]divisor
• W3dtype=f16N=1C=32D_in=16H_in=56W_in=56kernel_size=[2,2,2]padding=[0,0,0]video-2x2x2

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name	ms	TFLOP/s	of ceiling	by
W1	1.75×	0.0457	Torch-NPU eager reference	0.0800	·	·	·
W2	3.15×	0.0220	Torch-NPU eager reference	0.0693	·	·	·
W3	2.37×	0.0269	Torch-NPU eager reference	0.0638	·	·	·

AvgPool1dFwd

input: [N, C, L_in]

• W1dtype=f16N=4C=128L_in=4096kernel_size=3stride=2padding=1audio-downsample
• W2dtype=bf16N=2C=128L_in=2048kernel_size=4stride=2padding=1ceil_mode=truecount_include_pad=falseceil
• W3dtype=f16N=2C=256L_in=32000kernel_size=5stride=4padding=2long-temporal

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.55×	0.0333	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0515 0.0558	·	·	·
W2	1.87×	0.0213	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.0395 0.0424	·	·	·
W3	1.05×	0.1635	torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	0.1714 0.1742	·	·	·

AdaptiveMaxPool2dFwd

input: [N, C, H_in, W_in]

• W1dtype=f16N=8C=2048H_in=7W_in=7output_size=[1,1]global-1x1
• W2dtype=bf16N=2C=64H_in=55W_in=57output_size=[7,7]nondiv-7x7
• W3dtype=f16N=2C=128H_in=56W_in=56output_size=[6,6]spp-6x6

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.57×	0.0628	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.0961 0.0921	·	·	·
W2	1.24×	0.1020	ops-nn-b2:aclnnAdaptiveMaxPool3d open-source basis torch_compile_aclgraph torch_compile torch_npu eager vendor library	0.1220 0.1308 0.1227	·	·	·
W3	1.20×	0.1400	ops-nn-b2:aclnnAdaptiveMaxPool3d open-source basis torch_compile_aclgraph torch_compile torch_npu eager vendor library	0.1601 0.1727 0.1631	·	·	·

AvgPool2dFwd

input: [N, C, H_in, W_in]

stride=[2,2]

• W1dtype=bf16N=3C=96H_in=55W_in=57kernel_size=[3,5]padding=[1,2]ceil_mode=truecount_include_pad=falseceil-divisor
• W2dtype=f16N=2C=64H_in=112W_in=112kernel_size=[3,3]padding=[1,1]vision-3x3-s2
• W3dtype=f16N=2C=128H_in=56W_in=56kernel_size=[5,5]padding=[2,2]vision-5x5-s2

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	1.70×	0.0370	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0625 0.0617	·	·	·
W2	0.91×	0.0360	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0325 0.0209	·	·	·
W3	0.77×	0.0495	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0376 0.0290	·	·	·

MaxPool3dFwd

input: [N, C, D_in, H_in, W_in]

• W1dtype=f16N=8C=64D_in=16H_in=112W_in=112kernel_size=[1,2,2]stride=[1,2,2]padding=[0,0,0]c3d-pool1
• W2dtype=f16N=4C=128D_in=16H_in=56W_in=56kernel_size=[2,2,2]stride=[2,2,2]padding=[0,0,0]c3d-pool2
• W3dtype=bf16N=2C=64D_in=32H_in=112W_in=112kernel_size=[3,3,3]stride=[2,2,2]padding=[1,1,1]ceil_mode=truemedicalnet-stem

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.96×	0.5791	ops-nn:aclnnMaxPool3dWithArgmax open-source ⓘ torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.5477 0.5506 0.5475	·	·	·
W2	0.75×	0.2328	ops-nn:aclnnMaxPool3dWithArgmax open-source ⓘ basis torch_compile_aclgraph torch_compile torch_npu eager vendor library	0.1705 0.1732 0.1715	·	·	·
W3	1.45×	0.8512	ops-nn:aclnnMaxPool3dWithArgmax open-source ⓘ torch_compile_aclgraph torch_compile basis torch_npu eager vendor library	1.2424 1.2369 1.2424	·	·	·

MaxPool2dFwd

input: [N, C, H_in, W_in]

stride=[2,2]

• W1dtype=bf16N=32C=64H_in=55W_in=55kernel_size=[3,3]padding=[0,0]ceil_mode=truealexnet-ceil
• W2dtype=f16N=32C=64H_in=55W_in=55kernel_size=[3,3]padding=[0,0]ceil_mode=truealexnet-ceil
• W3dtype=f32N=32C=64H_in=55W_in=55kernel_size=[3,3]padding=[0,0]ceil_mode=truealexnet-ceil
• W4dtype=bf16N=32C=64H_in=112W_in=112kernel_size=[3,3]padding=[1,1]resnet-stem
• W5dtype=f16N=32C=64H_in=112W_in=112kernel_size=[3,3]padding=[1,1]resnet-stem
• W6dtype=f32N=32C=64H_in=112W_in=112kernel_size=[3,3]padding=[1,1]resnet-stem
• W7dtype=bf16N=16C=128H_in=56W_in=56kernel_size=[2,2]padding=[0,0]vgg-block
• W8dtype=f16N=16C=128H_in=56W_in=56kernel_size=[2,2]padding=[0,0]vgg-block
• W9dtype=f32N=16C=128H_in=56W_in=56kernel_size=[2,2]padding=[0,0]vgg-block

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.88×	0.1235	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0975 0.0970 0.0973	·	·	·
W2	0.59×	0.1245	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0670 0.0668 0.0666	·	·	·
W3	0.81×	0.1265	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0936 0.0936 0.0935	·	·	·
W4	1.10×	0.2938	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.3235 0.7558 0.3215	·	·	·
W5	1.01×	0.2940	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.3075 0.2472 0.3065	·	·	·
W6	1.33×	0.2774	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.3701 0.8113 0.3698	·	·	·
W7	1.16×	0.0737	torch_compile_aclgraph torch_compile basis torch_compile_ge torch_compile torch_npu eager vendor library	0.0803 0.0801 0.0805	·	·	·
W8	0.94×	0.0712	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0660 0.0707 0.0609	·	·	·
W9	0.91×	0.0767	torch_compile_aclgraph torch_compile torch_compile_ge torch_compile torch_npu eager vendor library basis	0.0669 0.0665 0.0665	·	·	·

AdaptiveAvgPool2dFwd

input: [N, C, H_in, W_in]

• W1dtype=bf16N=2C=64H_in=55W_in=57output_size=[7,7]nondiv-7x7
• W2dtype=f16N=8C=2048H_in=7W_in=7output_size=[1,1]resnet-global
• W3dtype=f16N=2C=128H_in=56W_in=56output_size=[6,6]spp-6x6

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.56×	0.1017	ops-nn-b2:aclnnAdaptiveAvgPool3d open-source ⓘ torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0474 0.0553 0.0473	·	·	·
W2	0.54×	0.0630	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	0.0340 0.0213	·	·	·
W3	0.36×	0.1410	ops-nn-b2:aclnnAdaptiveAvgPool3d open-source ⓘ torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0419 0.0555 0.0416	·	·	·

MaxPool1dFwd

input: [N, C, L_in]

• W1dtype=bf16N=16C=128L_in=2048kernel_size=5stride=2padding=2dilation=2ceil_mode=trueecg-cnn-dilated
• W2dtype=f16N=32C=80L_in=4096kernel_size=3stride=3sincnet-speaker-local
• W3dtype=f16N=64C=256L_in=128kernel_size=128stride=1textcnn-global

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	—	·	torch_npu eager vendor library basis	·	·	·	·
W2	0.00×	23.4675	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0910 0.0850	·	·	·
W3	0.01×	5.0949	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis	0.0779 0.0689	·	·	·

Top-k

TopkSelectorFwd

index_score: [batch, seq_len, seq_len_kv, kv_group]starts, ends: [batch, seq_len], i32

dtype=f32batch=1seq_len=32768seq_len_kv=65536kv_group=1

• W1topk=1024topk1024-s32k-kv64k
• W2topk=2048topk2048-s32k-kv64k

Workload	Ratio	Device time	Alternatives		Throughput	SOL	Bound
	alt / ours	ms	name · tier	ms	TFLOP/s	of ceiling	by
W1	0.76×	117.7379	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	89.9394 89.9271	·	·	·
W2	0.60×	164.3116	torch_compile_aclgraph torch_compile torch_npu eager vendor library basis ⓘ	98.9299 98.8865	·	·	·