Skip to content

Comments

[WIP] Left matmul refactor#149

Merged
Transurgeon merged 4 commits intomasterfrom
left_matmul_refactor
Feb 13, 2026
Merged

[WIP] Left matmul refactor#149
Transurgeon merged 4 commits intomasterfrom
left_matmul_refactor

Conversation

@dance858
Copy link
Collaborator

Adds a few extra tests to check the new left matmul logic on the python side

@github-actions
Copy link

github-actions bot commented Feb 13, 2026

Benchmarks that have stayed the same:

   before           after         ratio
 [25d65da4]       [4163920a]
      345±0ms          368±0ms     1.07  slow_pruning_1668_benchmark.SlowPruningBenchmark.time_compile_problem
      383±0ms          407±0ms     1.06  gini_portfolio.Yitzhaki.time_compile_problem
      266±0ms          279±0ms     1.05  simple_QP_benchmarks.SimpleQPBenchmark.time_compile_problem
      472±0ms          494±0ms     1.05  semidefinite_programming.SemidefiniteProgramming.time_compile_problem
      1.06±0s          1.11±0s     1.04  simple_LP_benchmarks.SimpleScalarParametrizedLPBenchmark.time_compile_problem
      1.10±0s          1.14±0s     1.04  simple_QP_benchmarks.LeastSquares.time_compile_problem
     13.8±0ms         14.2±0ms     1.03  simple_QP_benchmarks.ParametrizedQPBenchmark.time_compile_problem
      741±0ms          764±0ms     1.03  matrix_stuffing.ConeMatrixStuffingBench.time_compile_problem
      29.8±0s          30.6±0s     1.03  sdp_segfault_1132_benchmark.SDPSegfault1132Benchmark.time_compile_problem
      1.73±0s          1.77±0s     1.02  tv_inpainting.TvInpainting.time_compile_problem
      2.00±0s          2.05±0s     1.02  quantum_hilbert_matrix.QuantumHilbertMatrix.time_compile_problem
      5.14±0s          5.25±0s     1.02  optimal_advertising.OptimalAdvertising.time_compile_problem
      3.03±0s          3.09±0s     1.02  simple_QP_benchmarks.UnconstrainedQP.time_compile_problem
      273±0ms          274±0ms     1.00  matrix_stuffing.ParamSmallMatrixStuffing.time_compile_problem
      1.34±0s          1.34±0s     1.00  matrix_stuffing.ParamConeMatrixStuffing.time_compile_problem
      312±0ms          312±0ms     1.00  gini_portfolio.Murray.time_compile_problem
     34.8±0ms         34.9±0ms     1.00  matrix_stuffing.SmallMatrixStuffing.time_compile_problem
      6.54±0s          6.54±0s     1.00  svm_l1_regularization.SVMWithL1Regularization.time_compile_problem
      14.3±0s          14.3±0s     1.00  simple_LP_benchmarks.SimpleLPBenchmark.time_compile_problem
      1.68±0s          1.67±0s     1.00  finance.FactorCovarianceModel.time_compile_problem
     14.2±0ms         14.1±0ms     1.00  simple_LP_benchmarks.SimpleFullyParametrizedLPBenchmark.time_compile_problem
      5.76±0s          5.73±0s     0.99  huber_regression.HuberRegression.time_compile_problem
      1.01±0s          1.00±0s     0.99  gini_portfolio.Cajas.time_compile_problem
      188±0ms          185±0ms     0.98  high_dim_convex_plasticity.ConvexPlasticity.time_compile_problem
      18.8±0s          17.6±0s     0.93  finance.CVaRBenchmark.time_compile_problem

@dance858 dance858 changed the title Left matmul refactor [WIP] Left matmul refactor Feb 13, 2026
@dance858
Copy link
Collaborator Author

dance858 commented Feb 13, 2026

Once we have updated the sparsediff pypi package this should pass. Feel free to merge it when that's the case @Transurgeon

@Transurgeon
Copy link
Member

Once we have updated the sparsediff pypi package this should pass. Feel free to merge it when that's the case @Transurgeon

seems to work now, merging!

@Transurgeon Transurgeon reopened this Feb 13, 2026
@Transurgeon Transurgeon merged commit 25d65da into master Feb 13, 2026
38 of 47 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants