[Python] Adds np.einsum support

[lukastruemper]

No description provided.

[daisytuner]

Daisytuner Report - mlir_torch_models (chamomile)

@@                                   Benchmarks                                   @@
=====================================================================================
  Benchmark              Time        ΔTime       Thr         Energy      ΔEnergy     
=====================================================================================
# bn_conv_bn_relu_maxpool_torch18.50 s     -1.30%      N/A         3579.25 J   -1.55%      
# bn_conv_bn_relu_maxpool_run_none3.26 s      -0.19%      N/A         646.28 J    -0.84%      
# bn_conv_bn_relu_maxpool_run_sequential3.29 s      -0.06%      N/A         657.63 J    -0.38%      
# bn_conv_bn_relu_maxpool_run_openmp3.38 s      -0.16%      N/A         686.80 J    -1.33%      
# bn_conv_bn_relu_maxpool_run_cuda3.82 s      +0.31%      N/A         743.12 J    -0.02%      

[daisytuner]

Daisytuner Report - mlir_torch_layers (chamomile)

@@                                   Benchmarks                                   @@
=====================================================================================
  Benchmark              Time        ΔTime       Thr         Energy      ΔEnergy     
=====================================================================================
# batchnorm_torch        18.94 s     +0.21%      N/A         3688.52 J   -0.35%      
# batchnorm_run_none     7.86 s      +3.75%      N/A         1487.19 J   +1.25%      
# batchnorm_run_sequential8.60 s      +1.39%      N/A         1650.77 J   +0.53%      
# batchnorm_run_openmp   4.91 s      +4.99%      N/A         1107.43 J   +5.08%      
# batchnorm_run_cuda     13.02 s     +2.25%      N/A         2486.36 J   +0.39%      
# conv2d_torch           18.93 s     +2.29%      N/A         3631.69 J   +0.13%      
# conv2d_run_openmp      4.48 s      +4.74%      N/A         1029.51 J   +1.85%      
# conv2d_run_cuda        6.43 s      +1.30%      N/A         1243.96 J   -0.19%      
# linear_torch           6.31 s      +0.07%      N/A         1511.57 J   -0.85%      
# linear_run_none        10.00 s     -0.68%      N/A         2655.21 J   -2.76%      
# linear_run_sequential  8.50 s      +0.41%      N/A         2393.68 J   -1.96%      
# linear_run_openmp      8.19 s      +1.23%      N/A         2333.59 J   -0.81%      
# linear_run_cuda        7.52 s      +1.70%      N/A         1439.79 J   -0.70%      
# matmul_torch           6.26 s      +1.49%      N/A         1489.83 J   -0.12%      
# matmul_run_none        10.07 s     +0.39%      N/A         2701.81 J   -0.69%      
# matmul_run_sequential  8.64 s      +1.55%      N/A         2455.30 J   -0.13%      
# matmul_run_openmp      8.29 s      +1.83%      N/A         2376.84 J   +0.48%      
# matmul_run_cuda        7.36 s      +1.97%      N/A         1416.67 J   +0.48%      
# pooling_torch          26.27 s     -1.54%      N/A         5002.28 J   -4.68%      
# pooling_run_none       17.45 s     +0.83%      N/A         3325.42 J   +0.24%      
# pooling_run_sequential 17.47 s     +1.60%      N/A         3257.37 J   -1.13%      
# pooling_run_openmp     10.16 s     +4.51%      N/A         2059.36 J   +3.23%      
# pooling_run_cuda       23.82 s     +1.74%      N/A         4540.77 J   -0.56%      
# relu_torch             19.26 s     +1.24%      N/A         3681.31 J   -1.06%      
# relu_run_none          4.54 s      +2.14%      N/A         865.65 J    -0.26%      
# relu_run_sequential    4.60 s      +4.78%      N/A         877.84 J    +2.58%      
# relu_run_openmp        4.04 s      +9.27%      N/A         858.70 J    +12.11%     
# relu_run_cuda          6.22 s      +3.63%      N/A         1196.56 J   +1.68%      

[daisytuner]

Daisytuner Report - python_npbench (zinnia)

@@                                   Benchmarks                                   @@
=====================================================================================
  Benchmark              Time        ΔTime       Thr         Energy      ΔEnergy     
=====================================================================================
# adi_numpy              1.33 s      +1.47%      N/A         132.20 J    +1.25%      
# adi_omp                15.02 s     -0.57%      N/A         1459.27 J   -2.43%      
# adi_cuda               4.72 s      +0.16%      N/A         457.62 J    +0.01%      
# adi_seq_tuning         15.98 s     -0.90%      N/A         1507.15 J   -0.95%      
# atax_numpy             2.16 s      -0.00%      N/A         223.93 J    -0.02%      
# atax_omp               2.95 s      -1.75%      N/A         371.03 J    -3.14%      
# atax_cuda              4.11 s      -0.82%      N/A         422.86 J    -0.78%      
# atax_seq_tuning        3.72 s      -0.97%      N/A         377.38 J    -1.32%      
# gemm_numpy             1.20 s      -1.31%      N/A         191.92 J    -1.25%      
# gemm_omp               1.12 s      -0.01%      N/A         162.34 J    +0.32%      
# gemm_cuda              10.66 s     +0.68%      N/A         1013.52 J   +0.69%      
# gemm_seq_tuning        1.11 s      -0.06%      N/A         161.79 J    -0.09%      
# gesummv_numpy          1.75 s      +0.35%      N/A         249.12 J    +0.28%      
# gesummv_omp            1.99 s      +1.14%      N/A         313.22 J    +1.64%      
# gesummv_cuda           8.29 s      -0.50%      N/A         996.46 J    -0.45%      
# gesummv_seq_tuning     6.66 s      -0.02%      N/A         813.35 J    -0.07%      
# gemver_numpy           1.08 s      +0.44%      N/A         166.11 J    +0.44%      
# gemver_omp             873.38 ms   +1.34%      N/A         114.92 J    +3.04%      
# gemver_cuda            3.85 s      -1.23%      N/A         385.92 J    -0.79%      
# gemver_seq_tuning      4.49 s      -0.01%      N/A         434.54 J    -0.03%      
# k2mm_numpy             1.19 s      +0.32%      N/A         195.19 J    +0.13%      
# k2mm_omp               3.58 s      -0.62%      N/A         662.99 J    -1.34%      
# k2mm_cuda              13.53 s     -0.04%      N/A         1282.59 J   -0.10%      
# k2mm_seq_tuning        3.63 s      +0.52%      N/A         466.86 J    -0.17%      
# k3mm_numpy             1.03 s      +0.05%      N/A         181.54 J    -0.21%      
# k3mm_omp               5.54 s      -1.20%      N/A         947.29 J    -0.96%      
# k3mm_cuda              19.85 s     +0.09%      N/A         1871.36 J   +0.08%      
# k3mm_seq_tuning        5.72 s      -0.33%      N/A         790.67 J    -0.34%      
# mvt_numpy              2.44 s      +0.06%      N/A         249.54 J    +0.12%      
# mvt_omp                2.74 s      -0.11%      N/A         284.99 J    -0.03%      
# mvt_cuda               3.35 s      -0.28%      N/A         342.16 J    -0.34%      
# mvt_seq_tuning         2.74 s      -0.15%      N/A         284.15 J    -0.32%      
# symm_numpy             782.17 ms   -0.77%      N/A         80.52 J     -1.01%      
# symm_omp               6.15 s      +1.05%      N/A         608.72 J    +1.95%      
# symm_seq_tuning        8.48 s      -0.15%      N/A         807.86 J    -0.18%      
# syr2k_numpy            887.77 ms   -1.30%      N/A         90.15 J     -1.14%      
# syr2k_omp              9.86 s      -0.66%      N/A         936.99 J    -0.63%      
# syr2k_cuda             1.63 s      -1.07%      N/A         169.01 J    -0.93%      
# syr2k_seq_tuning       9.95 s      +0.48%      N/A         944.90 J    +0.48%      
# syrk_numpy             771.66 ms   +0.19%      N/A         79.50 J     +0.16%      
# syrk_omp               5.96 s      +0.07%      N/A         573.27 J    +0.08%      
# syrk_cuda              1.52 s      -1.00%      N/A         158.33 J    -0.89%      
# syrk_seq_tuning        5.96 s      -0.02%      N/A         572.68 J    -0.11%      
# trmm_numpy             882.29 ms   +0.06%      N/A         89.82 J     +0.04%      
# trmm_omp               699.43 ms   -1.39%      N/A         89.41 J     -1.49%      
# trmm_seq_tuning        3.37 s      -0.32%      N/A         322.50 J    -0.33%