FW-GLM-5-1 · popcorn · pop_l123_all

L1 P1 1_Conv1d_Depthwise.py

correct 1.38x 0.9μs

L1 P2 2_GraphEdgeSoftmaxCSR.py

correct 1784.48x 0.0μs

L1 P3 3_CSRSpMMMessagePassing.py

incorrect —

L1 P4 4_GJKRobotEnvironmentIntersectio

correct 144.88x 0.0μs

L1 P5 5_Conv1d_Causal.py

incorrect —

L1 P6 6_EdgeSoftmaxMultiHeadCSR.py

correct 1129.77x 0.0μs

L1 P7 7_Conv1d_FFT.py

correct 1.28x 1.0μs

L1 P8 8_SegmentTopKCSR.py

incorrect —

L1 P9 9_AssociativeScan.py

correct 28.33x 2.9μs

L1 P10 10_SampledDenseDenseMatmulEdges.

correct 1.99x 0.0μs

L1 P11 11_DegreeNormalizedAggregation.p

correct 143.93x 0.5μs

L1 P12 12_DeepSeekMoEDispatchPermute.py

correct 864.50x 0.1μs

L1 P13 13_FarthestPointSampling3D.py

correct 71.13x 0.1μs

L1 P14 14_GaussianProcessRBFKernel.py

incorrect —

L1 P15 15_COOScatterAddNodeFeatures.py

correct 0.65x 0.0μs

L1 P16 16_DeepSeekMoECombineScatter.py

incorrect —

L1 P17 17_CSRMaxAggregation.py

incorrect —

L1 P18 18_CSRMultiHeadSpMM.py

incorrect —

L1 P19 19_PairwiseDistanceMatrix.py

correct 5.60x 0.0μs

L1 P20 20_RotaryPositionEmbeddingBio.py

correct 5.62x 0.0μs

L1 P21 21_RadialBasisFunctionExpansion.

correct 0.66x 0.8μs

L1 P22 22_SE3InvariantLinear.py

correct 7.56x 0.1μs

L1 P23 23_virtual_all_reduce_mean.py

correct 1.23x 0.0μs

L1 P24 24_VoxelGridPooling.py

correct 261.13x 0.0μs

L1 P25 25_virtual_reduce_scatter_masked

correct 0.85x 0.0μs

L1 P26 26_ImportanceSampling.py

incorrect —

L1 P27 27_permutation_all_to_all.py

correct 5.37x 0.0μs

L1 P28 28_DirichletMultinomialLogLikeli

correct 16.46x 0.0μs

L1 P29 29_logsumexp_across_virtual_rank

correct 5.90x 0.0μs

L1 P30 30_broadcast_masked_from_row0.py

correct 1.04x 0.0μs

L1 P31 31_bitmask_or_popcount.py

correct 78.17x 0.0μs

L1 P32 32_argmax_tiebreak_smallest_rank

correct 33.16x 0.0μs

L2 P1 1_DeepSeekMLALoRAExpansion.py

correct 0.72x 0.0μs

L2 P3 3_FusedCrossAttentionRoPEDropout

correct 1.31x 0.3μs

L2 P4 4_GFVLA_FusedSceneGraphUpdate.py

incorrect —

L2 P5 5_LayerNorm_GELU_DilatedConv1d_R

correct 0.91x 3.0μs

L2 P6 6_TriangularAttention.py

correct 1.60x 0.3μs

L2 P7 7_Conv1d_Depthwise_Separable.py

correct 0.99x 0.6μs

L2 P8 8_DeepSeekMoEGroundedTop2Routing

correct 5.75x 0.0μs

L2 P9 9_MetropolisHastingsStep.py

correct 2.28x 0.1μs

L2 P10 10_GFVLA_FusedPairwiseNodeAggreg

correct 0.01x 22.9μs

L2 P12 12_BatchNorm_GELU_DilatedConv1d_

correct 1.00x 0.9μs

L2 P13 13_QuantVLA_SelectiveLayoutFusio

correct 2.26x 0.0μs

L2 P14 14_TriangularMultiplicativeUpdat

correct 1.56x 0.2μs

L2 P15 15_GQAKVHeadExpansionAttention.p

correct 536.09x 0.2μs

L2 P16 16_HamiltonianMonteCarloStep.py

correct 10.16x 0.1μs

L2 P17 17_FP8ScaledAttention.py

incorrect —

L2 P19 19_DepthwiseSeparableConv1D_GELU

incorrect —

L2 P20 20_GatedDeltaNetLinearAttention.

correct 62.30x 0.1μs

L2 P21 21_TriangularMultiplicativeUpdat

correct 0.37x 0.9μs

L2 P22 22_FusedBallQuery3D.py

correct 2.80x 0.1μs

L2 P23 23_GibbsSamplingStep.py

correct 8.43x 0.4μs

L2 P24 24_KimiDeltaAttentionChannelwise

correct 31.65x 0.2μs

L2 P25 25_OuterProductMean.py

correct 1.56x 0.1μs

L2 P26 26_RoPEKVCacheUpdate.py

correct 71.78x 0.0μs

L2 P28 28_FusedGroupNormSiLU.py

correct 1.81x 0.0μs

L2 P29 29_MSARowAttention.py

correct 3.85x 0.2μs

L2 P30 30_SelectiveScan_Mamba.py

error —

L2 P31 31_FusedMLAAttention.py

correct 14411.76x 0.0μs

L2 P32 32_MSAColumnAttention.py

incorrect —

L2 P33 33_VectorizedPoseTransformFused.

correct 42.34x 0.0μs

L2 P35 35_FusedConv1dGroupNorm.py

correct 1.28x 0.1μs

L2 P36 36_InvariantPointAttention.py

correct 0.64x 1.5μs

L2 P37 37_CSRFusedAttentionValue.py

correct 968.31x 0.0μs

L2 P39 39_Fused1DTemporalConvolution.py

incorrect —

L2 P40 40_SmithWatermanDPScore.py

correct 1087.72x 0.1μs

L2 P41 41_VariationalELBO.py

correct 1.54x 0.2μs

L2 P42 42_BatchedForwardKinematics.py

correct 3.02x 0.0μs

L2 P43 43_ContactMapPrediction.py

correct 0.77x 0.2μs

L2 P44 44_ReparameterizationTrick.py

correct 2.36x 0.2μs

L2 P45 45_AxialAttention.py

incorrect —

L2 P46 46_SteinVariationalGradient.py

correct 1.10x 0.9μs

L2 P47 47_virtual_all_gather_concat_lin

correct 0.65x 0.0μs

L2 P48 48_ParticleFilter.py

incorrect —

L2 P49 49_sum_gelu_across_virtual_ranks

correct 1.59x 0.0μs

L2 P50 50_HiddenMarkovForward.py

error —

L3 P1 1_DilatedConvTower_GenomicBlock.

incorrect —

L3 P2 2_FusedMultiStepDiffusionSolver.

correct 3.57x 0.8μs

L3 P3 3_SSM2D_FusedBidirectionalCrossS

correct 12.36x 0.0μs

L3 P4 4_MambaBlock.py

correct 1.39x 92.6μs

L3 P5 5_FlowMatchingProbabilityFlowODE

correct 2.00x 1.5μs

L3 P6 6_HyenaOperator.py

correct 1.00x 4.9μs

L3 P7 7_GaussianProcessRegression.py

correct 1.43x 0.6μs

L3 P8 8_EvoformerBlock.py

correct 0.38x 2.4μs

L3 P9 9_BayesianLinearRegression.py

correct 1.70x 0.2μs

L3 P10 10_GaussianMixtureModelEM.py

correct 3.67x 0.6μs

L3 P11 11_pipeline_three_stage_mlp.py

correct 1.75x 0.0μs

L3 P12 12_NormalizingFlowPlanar.py

incorrect —

L3 P13 13_ConditionalVAELoss.py

correct 0.99x 0.4μs