grok-4.3 · popcorn · pop_l123_all

L1 P1 1_Conv1d_Depthwise.py

correct 1.58x 0.8μs

L1 P2 2_GraphEdgeSoftmaxCSR.py

compile_fail —

L1 P3 3_CSRSpMMMessagePassing.py

compile_fail —

L1 P4 4_GJKRobotEnvironmentIntersectio

correct 153.46x 0.0μs

L1 P5 5_Conv1d_Causal.py

incorrect —

L1 P6 6_EdgeSoftmaxMultiHeadCSR.py

correct 366.52x 0.0μs

L1 P7 7_Conv1d_FFT.py

correct 1.00x 1.3μs

L1 P8 8_SegmentTopKCSR.py

incorrect —

L1 P9 9_AssociativeScan.py

compile_fail —

L1 P10 10_SampledDenseDenseMatmulEdges.

correct 1.02x 0.1μs

L1 P11 11_DegreeNormalizedAggregation.p

incorrect —

L1 P12 12_DeepSeekMoEDispatchPermute.py

correct 628.24x 0.1μs

L1 P13 13_FarthestPointSampling3D.py

correct 70.44x 0.2μs

L1 P14 14_GaussianProcessRBFKernel.py

correct 2.15x 0.0μs

L1 P15 15_COOScatterAddNodeFeatures.py

correct 0.60x 0.0μs

L1 P16 16_DeepSeekMoECombineScatter.py

correct 1469.88x 0.1μs

L1 P17 17_CSRMaxAggregation.py

compile_fail —

L1 P18 18_CSRMultiHeadSpMM.py

compile_fail —

L1 P19 19_PairwiseDistanceMatrix.py

correct 6.60x 0.0μs

L1 P20 20_RotaryPositionEmbeddingBio.py

correct 2.83x 0.0μs

L1 P21 21_RadialBasisFunctionExpansion.

correct 3.45x 0.2μs

L1 P22 22_SE3InvariantLinear.py

correct 0.63x 0.4μs

L1 P23 23_virtual_all_reduce_mean.py

correct 1.18x 0.0μs

L1 P24 24_VoxelGridPooling.py

correct 92.09x 0.1μs

L1 P25 25_virtual_reduce_scatter_masked

compile_fail —

L1 P26 26_ImportanceSampling.py

correct 2.38x 0.2μs

L1 P27 27_permutation_all_to_all.py

correct 4.86x 0.0μs

L1 P28 28_DirichletMultinomialLogLikeli

correct 0.43x 0.3μs

L1 P29 29_logsumexp_across_virtual_rank

correct 5.45x 0.0μs

L1 P30 30_broadcast_masked_from_row0.py

correct 1.05x 0.0μs

L1 P31 31_bitmask_or_popcount.py

incorrect —

L1 P32 32_argmax_tiebreak_smallest_rank

compile_fail —

L2 P1 1_DeepSeekMLALoRAExpansion.py

correct 0.52x 0.1μs

L2 P3 3_FusedCrossAttentionRoPEDropout

incorrect —

L2 P4 4_GFVLA_FusedSceneGraphUpdate.py

incorrect —

L2 P5 5_LayerNorm_GELU_DilatedConv1d_R

correct 0.96x 2.9μs

L2 P6 6_TriangularAttention.py

correct 0.99x 0.5μs

L2 P7 7_Conv1d_Depthwise_Separable.py

incorrect —

L2 P8 8_DeepSeekMoEGroundedTop2Routing

correct 0.27x 0.3μs

L2 P9 9_MetropolisHastingsStep.py

compile_fail —

L2 P10 10_GFVLA_FusedPairwiseNodeAggreg

correct 0.01x 21.1μs

L2 P12 12_BatchNorm_GELU_DilatedConv1d_

compile_fail —

L2 P13 13_QuantVLA_SelectiveLayoutFusio

correct 1.18x 0.1μs

L2 P14 14_TriangularMultiplicativeUpdat

correct 1.23x 0.3μs

L2 P15 15_GQAKVHeadExpansionAttention.p

compile_fail —

L2 P16 16_HamiltonianMonteCarloStep.py

compile_fail —

L2 P17 17_FP8ScaledAttention.py

incorrect —

L2 P19 19_DepthwiseSeparableConv1D_GELU

correct 1.00x 1.8μs

L2 P20 20_GatedDeltaNetLinearAttention.

correct 33.10x 0.3μs

L2 P21 21_TriangularMultiplicativeUpdat

correct 1.19x 0.3μs

L2 P22 22_FusedBallQuery3D.py

correct 2.53x 0.1μs

L2 P23 23_GibbsSamplingStep.py

correct 5.63x 0.6μs

L2 P24 24_KimiDeltaAttentionChannelwise

correct 23.88x 0.3μs

L2 P25 25_OuterProductMean.py

correct 0.04x 5.4μs

L2 P26 26_RoPEKVCacheUpdate.py

correct 57.36x 0.1μs

L2 P28 28_FusedGroupNormSiLU.py

correct 1.30x 0.0μs

L2 P29 29_MSARowAttention.py

compile_fail —

L2 P30 30_SelectiveScan_Mamba.py

correct 41.88x 3.9μs

L2 P31 31_FusedMLAAttention.py

correct 734.14x 0.3μs

L2 P32 32_MSAColumnAttention.py

incorrect —

L2 P33 33_VectorizedPoseTransformFused.

correct 50.69x 0.0μs

L2 P35 35_FusedConv1dGroupNorm.py

correct 1.01x 0.1μs

L2 P36 36_InvariantPointAttention.py

correct 0.93x 0.8μs

L2 P37 37_CSRFusedAttentionValue.py

correct 1344.34x 0.0μs

L2 P39 39_Fused1DTemporalConvolution.py

incorrect —

L2 P40 40_SmithWatermanDPScore.py

correct 720.67x 0.2μs

L2 P41 41_VariationalELBO.py

correct 1.21x 0.3μs

L2 P42 42_BatchedForwardKinematics.py

correct 3.11x 0.0μs

L2 P43 43_ContactMapPrediction.py

correct 0.73x 0.2μs

L2 P44 44_ReparameterizationTrick.py

correct 1.56x 0.3μs

L2 P45 45_AxialAttention.py

incorrect —

L2 P46 46_SteinVariationalGradient.py

correct 0.05x 20.5μs

L2 P47 47_virtual_all_gather_concat_lin

correct 0.82x 0.0μs

L2 P48 48_ParticleFilter.py

incorrect —

L2 P49 49_sum_gelu_across_virtual_ranks

correct 1.53x 0.0μs

L2 P50 50_HiddenMarkovForward.py

compile_fail —

L3 P1 1_DilatedConvTower_GenomicBlock.

correct 0.99x 24.9μs

L3 P2 2_FusedMultiStepDiffusionSolver.

compile_fail —

L3 P3 3_SSM2D_FusedBidirectionalCrossS

correct 6.41x 0.0μs

L3 P4 4_MambaBlock.py

correct 22.62x 7.2μs

L3 P5 5_FlowMatchingProbabilityFlowODE

correct 0.05x 73.6μs

L3 P6 6_HyenaOperator.py

incorrect —

L3 P7 7_GaussianProcessRegression.py

correct 1.07x 0.8μs

L3 P8 8_EvoformerBlock.py

incorrect —

L3 P9 9_BayesianLinearRegression.py

incorrect —

L3 P10 10_GaussianMixtureModelEM.py

correct 0.01x 181.0μs

L3 P11 11_pipeline_three_stage_mlp.py

correct 0.12x 0.2μs

L3 P12 12_NormalizingFlowPlanar.py

incorrect —

L3 P13 13_ConditionalVAELoss.py

correct 1.08x 0.3μs