allows cmath nodes to be removed by dead data elimination

[lukastruemper]

No description provided.

[daisytuner]

Daisytuner Report - mlir_torch_models (chamomile)

@@                                   Benchmarks                                   @@
=====================================================================================
  Benchmark              Time        ΔTime       Thr         Energy      ΔEnergy     
=====================================================================================
# bn_conv_bn_relu_maxpool_torch18.54 s     -0.67%      N/A         3568.77 J   -0.70%      
# bn_conv_bn_relu_maxpool_run_none3.25 s      -0.23%      N/A         653.03 J    -0.23%      
# bn_conv_bn_relu_maxpool_run_sequential3.28 s      -0.62%      N/A         661.81 J    -0.35%      
# bn_conv_bn_relu_maxpool_run_openmp3.24 s      -0.94%      N/A         655.95 J    -1.20%      
# bn_conv_bn_relu_maxpool_run_cuda3.66 s      -0.54%      N/A         715.60 J    -0.45%      

[daisytuner]

Daisytuner Report - mlir_torch_layers (chamomile)

@@                                   Benchmarks                                   @@
=====================================================================================
  Benchmark              Time        ΔTime       Thr         Energy      ΔEnergy     
=====================================================================================
# batchnorm_torch        19.27 s     -0.17%      N/A         3777.56 J   -0.21%      
# batchnorm_run_none     6.19 s      -2.58%      N/A         1205.34 J   -2.48%      
# batchnorm_run_sequential6.57 s      +0.27%      N/A         1281.31 J   +0.53%      
# batchnorm_run_openmp   5.70 s      -0.75%      N/A         1343.12 J   -0.75%      
# batchnorm_run_cuda     8.17 s      +1.10%      N/A         1597.30 J   +0.75%      
# conv2d_torch           18.57 s     -0.21%      N/A         3650.62 J   -0.18%      
# conv2d_run_openmp      5.02 s      +1.72%      N/A         1217.26 J   +1.37%      
# conv2d_run_cuda        7.68 s      -0.60%      N/A         1502.47 J   -0.65%      
# linear_torch           6.07 s      -0.72%      N/A         1462.06 J   -1.58%      
# linear_run_none        11.60 s     -0.08%      N/A         3168.86 J   +0.20%      
# linear_run_sequential  9.97 s      -0.32%      N/A         2773.51 J   -0.22%      
# linear_run_openmp      9.73 s      -0.82%      N/A         2857.05 J   -0.77%      
# linear_run_cuda        9.18 s      -0.52%      N/A         1801.42 J   +0.04%      
# matmul_torch           6.12 s      -1.24%      N/A         1491.80 J   -0.56%      
# matmul_run_none        11.63 s     -0.02%      N/A         3172.11 J   +0.10%      
# matmul_run_sequential  9.88 s      -0.59%      N/A         2762.86 J   -0.30%      
# matmul_run_openmp      9.79 s      -0.47%      N/A         2889.17 J   -0.03%      
# matmul_run_cuda        9.03 s      -0.19%      N/A         1769.75 J   +0.11%      
# pooling_torch          25.66 s     +0.06%      N/A         5121.09 J   +0.00%      
# pooling_run_none       25.06 s     +0.00%      N/A         4810.49 J   -0.17%      
# pooling_run_sequential 25.69 s     +0.20%      N/A         4933.67 J   +0.47%      
# pooling_run_openmp     17.06 s     -0.12%      N/A         3634.88 J   +0.04%      
# pooling_run_cuda       31.23 s     -0.13%      N/A         6086.43 J   -0.14%      
# relu_torch             18.90 s     -0.26%      N/A         3713.57 J   -0.22%      
# relu_run_none          5.21 s      -1.22%      N/A         1019.75 J   -1.20%      
+ relu_run_sequential    5.30 s      -17.10%     N/A         1034.55 J   -16.75%     
# relu_run_openmp        5.44 s      -4.89%      N/A         1266.30 J   -5.76%      
+ relu_run_cuda          5.93 s      -28.65%     N/A         1169.02 J   -28.28%     

[daisytuner]

Daisytuner Report - python_npbench (zinnia)

@@                                   Benchmarks                                   @@
=====================================================================================
  Benchmark              Time        ΔTime       Thr         Energy      ΔEnergy     
=====================================================================================
# adi_numpy              1.35 s      +3.24%      N/A         134.23 J    +2.84%      
# adi_omp                14.92 s     +1.45%      N/A         1497.27 J   +3.27%      
# adi_cuda               4.70 s      +0.65%      N/A         455.59 J    +0.63%      
# adi_seq_tuning         15.16 s     -0.32%      N/A         1403.78 J   +0.27%      
# atax_numpy             2.17 s      +0.15%      N/A         224.37 J    +0.21%      
# atax_omp               2.98 s      +0.63%      N/A         377.04 J    +1.61%      
# atax_cuda              4.14 s      +0.22%      N/A         425.65 J    +0.20%      
# atax_seq_tuning        4.14 s      -0.44%      N/A         398.81 J    -0.65%      
# gemm_numpy             1.21 s      -0.46%      N/A         193.38 J    -0.64%      
# gemm_omp               1.11 s      +0.13%      N/A         161.65 J    -0.22%      
# gemm_cuda              10.62 s     +0.34%      N/A         1009.86 J   +0.42%      
# gemm_seq_tuning        1.11 s      -0.21%      N/A         161.42 J    -0.25%      
# gesummv_numpy          1.75 s      -0.07%      N/A         248.75 J    -0.20%      
# gesummv_omp            2.01 s      +0.64%      N/A         315.27 J    +0.08%      
# gesummv_cuda           8.39 s      +1.07%      N/A         1006.06 J   +0.83%      
# gesummv_seq_tuning     8.70 s      +1.43%      N/A         977.17 J    +0.81%      
# gemver_numpy           1.09 s      +1.00%      N/A         166.57 J    +0.96%      
# gemver_omp             833.43 ms   -4.13%      N/A         105.53 J    -7.08%      
# gemver_cuda            3.87 s      -0.16%      N/A         387.05 J    -0.16%      
# gemver_seq_tuning      5.54 s      +0.97%      N/A         497.49 J    +0.82%      
# k2mm_numpy             1.20 s      +0.14%      N/A         195.82 J    +0.06%      
# k2mm_omp               3.54 s      -1.49%      N/A         663.48 J    -0.85%      
# k2mm_cuda              13.58 s     +0.05%      N/A         1287.68 J   +0.00%      
# k2mm_seq_tuning        3.06 s      +3.90%      N/A         400.66 J    +2.47%      
# k3mm_numpy             1.03 s      +0.63%      N/A         182.03 J    +0.57%      
# k3mm_omp               5.59 s      +0.72%      N/A         955.12 J    +0.01%      
# k3mm_cuda              19.81 s     -0.21%      N/A         1867.42 J   -0.41%      
# k3mm_seq_tuning        5.26 s      +0.53%      N/A         751.60 J    +0.71%      
# mvt_numpy              2.42 s      -0.60%      N/A         246.60 J    -0.96%      
# mvt_omp                2.74 s      -0.04%      N/A         284.21 J    -0.14%      
# mvt_cuda               3.36 s      +0.08%      N/A         342.77 J    -0.04%      
# mvt_seq_tuning         2.74 s      -0.53%      N/A         283.75 J    -0.68%      
# symm_numpy             787.99 ms   +0.89%      N/A         81.18 J     +0.95%      
# symm_omp               6.09 s      -0.37%      N/A         593.87 J    -2.03%      
# symm_seq_tuning        8.26 s      -0.69%      N/A         746.29 J    -0.79%      
# syr2k_numpy            900.31 ms   +0.39%      N/A         91.46 J     +0.38%      
# syr2k_omp              9.83 s      -0.63%      N/A         933.05 J    -0.70%      
# syr2k_cuda             1.65 s      +0.20%      N/A         171.18 J    +0.01%      
# syr2k_seq_tuning       9.86 s      -0.22%      N/A         935.65 J    -0.36%      
# syrk_numpy             784.02 ms   +1.04%      N/A         80.53 J     +1.04%      
# syrk_omp               5.96 s      +0.12%      N/A         572.71 J    +0.08%      
# syrk_cuda              1.51 s      -0.19%      N/A         158.13 J    -0.42%      
# syrk_seq_tuning        5.92 s      -1.47%      N/A         569.70 J    -1.37%      
# trmm_numpy             882.85 ms   +0.43%      N/A         89.84 J     +0.51%      
# trmm_omp               705.02 ms   -0.74%      N/A         89.83 J     -0.51%      
# trmm_seq_tuning        3.45 s      +0.05%      N/A         277.93 J    -0.70%