adds memlet simplification for contiguous memory accesses

[lukastruemper]

No description provided.

[daisytuner]

Daisytuner Report - mlir_torch_models (chamomile)

@@                                   Benchmarks                                   @@
=====================================================================================
  Benchmark              Time        ΔTime       Thr         Energy      ΔEnergy     
=====================================================================================
# bn_conv_bn_relu_maxpool_torch18.66 s     +0.12%      N/A         3648.74 J   +4.69%      
# bn_conv_bn_relu_maxpool_run_none3.26 s      -0.73%      N/A         661.98 J    +3.69%      
# bn_conv_bn_relu_maxpool_run_sequential3.29 s      -1.00%      N/A         670.81 J    +3.67%      
# bn_conv_bn_relu_maxpool_run_openmp3.40 s      +4.63%      N/A         696.97 J    +8.73%      
# bn_conv_bn_relu_maxpool_run_cuda3.70 s      +0.17%      N/A         729.45 J    +4.84%      

[daisytuner]

Daisytuner Report - mlir_torch_layers (chamomile)

@@                                   Benchmarks                                   @@
=====================================================================================
  Benchmark              Time        ΔTime       Thr         Energy      ΔEnergy     
=====================================================================================
# batchnorm_torch        19.06 s     +0.58%      N/A         3685.99 J   -0.87%      
# batchnorm_run_none     6.23 s      -3.49%      N/A         1195.00 J   -4.89%      
# batchnorm_run_sequential6.57 s      -0.33%      N/A         1256.77 J   -1.93%      
# batchnorm_run_openmp   5.78 s      -0.88%      N/A         1343.43 J   -1.98%      
# batchnorm_run_cuda     8.13 s      -0.33%      N/A         1568.42 J   -1.82%      
# conv2d_torch           18.61 s     -1.03%      N/A         3605.21 J   -2.37%      
# conv2d_run_openmp      5.04 s      +3.30%      N/A         1211.14 J   +1.59%      
# conv2d_run_cuda        7.83 s      -0.55%      N/A         1510.92 J   -2.09%      
# linear_torch           6.10 s      +0.23%      N/A         1451.16 J   -0.93%      
# linear_run_none        11.71 s     +0.36%      N/A         3176.41 J   +0.01%      
# linear_run_sequential  10.03 s     +0.55%      N/A         2761.98 J   -0.26%      
# linear_run_openmp      9.82 s      -0.50%      N/A         2867.51 J   -0.99%      
# linear_run_cuda        9.29 s      +0.21%      N/A         1799.27 J   -0.64%      
# matmul_torch           6.12 s      +1.10%      N/A         1456.12 J   -0.26%      
# matmul_run_none        11.65 s     +2.36%      N/A         3152.51 J   +1.35%      
# matmul_run_sequential  9.94 s      +0.47%      N/A         2752.06 J   -0.16%      
# matmul_run_openmp      9.89 s      +1.43%      N/A         2899.81 J   +1.02%      
# matmul_run_cuda        9.11 s      +0.44%      N/A         1760.64 J   -0.76%      
# pooling_torch          26.37 s     +2.82%      N/A         5189.56 J   +1.38%      
# pooling_run_none       25.68 s     +0.99%      N/A         4849.18 J   -0.64%      
# pooling_run_sequential 25.92 s     -0.05%      N/A         4902.93 J   -1.60%      
# pooling_run_openmp     17.45 s     +0.90%      N/A         3665.90 J   -0.66%      
# pooling_run_cuda       31.73 s     +1.18%      N/A         6101.62 J   -0.31%      
# relu_torch             18.95 s     +0.32%      N/A         3653.47 J   -1.54%      
# relu_run_none          5.26 s      -0.14%      N/A         1012.56 J   -1.62%      
# relu_run_sequential    6.38 s      +0.39%      N/A         1224.22 J   -1.01%      
# relu_run_openmp        5.70 s      -0.25%      N/A         1320.24 J   -2.17%      
# relu_run_cuda          8.40 s      +0.81%      N/A         1624.21 J   -0.43%      

[daisytuner]

Daisytuner Report - python_npbench (zinnia)

@@                                   Benchmarks                                   @@
=====================================================================================
  Benchmark              Time        ΔTime       Thr         Energy      ΔEnergy     
=====================================================================================
# adi_numpy              1.33 s      +1.32%      N/A         132.70 J    +1.32%      
# adi_omp                14.80 s     +0.41%      N/A         1451.45 J   +0.01%      
# adi_cuda               4.68 s      -0.30%      N/A         454.20 J    -0.18%      
# adi_seq_tuning         14.95 s     +0.06%      N/A         1388.08 J   +0.11%      
# atax_numpy             2.16 s      -0.58%      N/A         223.79 J    -0.71%      
# atax_omp               3.02 s      +0.29%      N/A         383.20 J    +0.89%      
# atax_cuda              4.12 s      +0.39%      N/A         424.58 J    +0.46%      
# atax_seq_tuning        4.15 s      +1.12%      N/A         401.92 J    +1.04%      
# gemm_numpy             1.21 s      +0.46%      N/A         193.55 J    +0.20%      
# gemm_omp               1.11 s      -0.08%      N/A         162.66 J    +0.18%      
# gemm_cuda              10.58 s     -0.42%      N/A         1005.09 J   -0.52%      
# gemm_seq_tuning        1.11 s      -0.30%      N/A         161.56 J    -0.08%      
# gesummv_numpy          1.73 s      -1.29%      N/A         245.71 J    -1.52%      
# gesummv_omp            1.94 s      -7.62%      N/A         305.60 J    -8.94%      
# gesummv_cuda           8.30 s      +0.64%      N/A         998.37 J    +0.56%      
# gesummv_seq_tuning     8.58 s      -3.03%      N/A         971.23 J    -1.03%      
# gemver_numpy           1.08 s      -0.08%      N/A         165.54 J    -0.07%      
# gemver_omp             867.30 ms   +0.14%      N/A         113.50 J    +0.28%      
# gemver_cuda            3.87 s      -0.02%      N/A         386.48 J    -0.12%      
# gemver_seq_tuning      5.51 s      +0.39%      N/A         496.93 J    +1.06%      
# k2mm_numpy             1.19 s      -0.32%      N/A         194.52 J    -0.53%      
# k2mm_omp               3.49 s      -0.49%      N/A         651.42 J    -0.88%      
# k2mm_cuda              13.56 s     -0.22%      N/A         1286.19 J   -0.16%      
# k2mm_seq_tuning        2.95 s      -2.62%      N/A         392.00 J    -1.18%      
# k3mm_numpy             1.02 s      -0.31%      N/A         181.42 J    +0.01%      
# k3mm_omp               5.59 s      +0.39%      N/A         962.18 J    +2.65%      
# k3mm_cuda              19.77 s     -0.30%      N/A         1866.09 J   -0.19%      
# k3mm_seq_tuning        5.28 s      -1.49%      N/A         746.64 J    -0.53%      
# mvt_numpy              2.42 s      -0.11%      N/A         246.92 J    -0.11%      
# mvt_omp                2.74 s      -0.17%      N/A         284.46 J    -0.12%      
# mvt_cuda               3.36 s      +0.15%      N/A         342.76 J    +0.12%      
# mvt_seq_tuning         2.74 s      -0.05%      N/A         284.43 J    -0.04%      
# symm_numpy             785.07 ms   -1.54%      N/A         80.78 J     -1.45%      
# symm_omp               6.09 s      +0.23%      N/A         603.02 J    +1.40%      
# symm_seq_tuning        8.22 s      -0.18%      N/A         742.50 J    -0.06%      
# syr2k_numpy            890.55 ms   +0.90%      N/A         90.42 J     +0.78%      
# syr2k_omp              9.85 s      +0.11%      N/A         937.02 J    +0.20%      
# syr2k_cuda             1.63 s      -1.12%      N/A         169.47 J    -1.05%      
# syr2k_seq_tuning       9.82 s      +0.02%      N/A         933.93 J    +0.16%      
# syrk_numpy             771.45 ms   +0.30%      N/A         79.36 J     +0.35%      
# syrk_omp               5.94 s      -1.43%      N/A         571.41 J    -1.30%      
# syrk_cuda              1.53 s      +0.17%      N/A         159.95 J    +0.31%      
# syrk_seq_tuning        5.98 s      +0.48%      N/A         575.31 J    +0.53%      
# trmm_numpy             877.73 ms   +0.33%      N/A         89.50 J     +0.24%      
# trmm_omp               707.48 ms   -1.60%      N/A         90.47 J     -0.42%      
# trmm_seq_tuning        3.39 s      -0.29%      N/A         276.74 J    -0.58%