10.5.5 : The compilation
Let's compile :|
$ make -- Configuring done -- Generating done -- Build files have been written to: ExampleOptimisation/build [ 2%] Built target hadamard_product_O2 [ 4%] Built target hadamard_product_O1 [ 4%] Built target hadamard_product_vectorize [ 6%] Built target hadamard_product_O0 [ 6%] Built target hadamard_product_O3 [ 8%] Built target hadamard_product_Ofast [ 10%] Built target hadamard_product_intrinsics [ 13%] Built target asterics_hpc [ 15%] Built target saxpy_O2 [ 17%] Built target saxpy_O0 [ 17%] Built target saxpy_O3 [ 19%] Built target saxpy_O1 [ 21%] Built target saxpy_Ofast [ 23%] Built target saxpy_vectorize [ 26%] Built target saxpy_intrinsics [ 30%] Built target reduction_real_O2 [ 32%] Built target reduction_real_intrinsics_interleave8_O3 [ 34%] Built target reduction_real_O1 [ 36%] Built target reduction_real_Ofast [ 39%] Built target reduction_O0 [ 41%] Built target reduction_O1 [ 43%] Built target reduction_O2 [ 43%] Built target reduction_O3 [ 45%] Built target reduction_real_intrinsics_interleave4_O3 [ 47%] Built target reduction_real_vectorize_Ofast [ 50%] Built target reduction_real_intrinsics_interleave2_O3 [ 52%] Built target reduction_real_intrinsics_O3 [ 54%] Built target reduction_real_O3 [ 56%] Built target reduction_real_O0 [ 58%] Built target reduction_real_vectorize_O3 [ 60%] Built target barycentre_intrinsics [ 63%] Built target barycentre_base_O2 [ 65%] Built target barycentre_base_O1 [ 67%] Built target barycentre_base_O0 [ 69%] Built target barycentre_vectorizeSplit_O3 [ 71%] Built target barycentre_base_Ofast [ 73%] Built target barycentre_base_O3 [ 76%] Built target barycentre_vectorize_O3 [ 80%] Built target sgemm_vectorize_Ofast [ 82%] Built target sgemm_base_O1 [ 84%] Built target sgemm_vectorize_O3 [ 86%] Built target sgemm_base_Ofast [ 89%] Built target sgemm_base_O3 [ 91%] Built target sgemm_base_O0 Scanning dependencies of target sgemm_intrinsics_O3 [ 91%] Building CXX object 6-Sgemm/CMakeFiles/sgemm_intrinsics_O3.dir/sgemm_intrinsics.cpp.o [ 93%] Building CXX object 6-Sgemm/CMakeFiles/sgemm_intrinsics_O3.dir/main_sgemm_intrinsics.cpp.o [ 93%] Linking CXX executable sgemm_intrinsics_O3 [ 93%] Built target sgemm_intrinsics_O3 [ 95%] Built target sgemm_swap_Ofast [ 97%] Built target sgemm_swap_O3 [100%] Built target sgemm_base_O2 |
Let's get the performances :
|
$ make plot_all [ 1%] Built target asterics_hpc [ 2%] Built target sgemm_swap_O3 [ 3%] Built target sgemm_base_Ofast [ 4%] Built target sgemm_base_O3 [ 6%] Built target sgemm_swap_Ofast [ 8%] Built target plot_sgemmSwap [ 9%] Built target hadamard_product_intrinsics [ 9%] Built target hadamard_product_vectorize [ 9%] Built target hadamard_product_O3 [ 12%] Built target plot_hadamardIntrinsics [ 13%] Built target hadamard_product_Ofast [ 14%] Built target hadamard_product_O2 [ 15%] Built target hadamard_product_O1 [ 16%] Built target hadamard_product_O0 [ 19%] Built target plot_hadamardBase [ 20%] Built target plot_hadamardVectorize [ 21%] Built target saxpy_intrinsics [ 21%] Built target saxpy_O3 [ 22%] Built target saxpy_vectorize [ 25%] Built target plot_saxpyIntrinsics [ 26%] Built target plot_saxpyVectorize [ 27%] Built target saxpy_Ofast [ 28%] Built target saxpy_O2 [ 30%] Built target saxpy_O0 [ 31%] Built target saxpy_O1 [ 33%] Built target plot_saxpyBase [ 34%] Built target reduction_real_intrinsics_O3 [ 36%] Built target reduction_real_intrinsics_interleave8_O3 [ 37%] Built target reduction_real_Ofast [ 38%] Built target reduction_real_intrinsics_interleave4_O3 [ 39%] Built target reduction_real_vectorize_Ofast [ 40%] Built target reduction_real_intrinsics_interleave2_O3 [ 43%] Built target plot_reductionIntrinsicsInterleave8 [ 44%] Built target reduction_real_vectorize_O3 [ 45%] Built target reduction_real_O3 [ 48%] Built target plot_reductionVectorize [ 48%] Built target reduction_O3 [ 49%] Built target reduction_O0 [ 50%] Built target reduction_O1 [ 51%] Built target reduction_O2 [ 54%] Built target plot_reductionBase [ 55%] Built target reduction_real_O0 [ 57%] Built target reduction_real_O2 [ 59%] Built target reduction_real_O1 [ 61%] Built target plot_reductionReal [ 63%] Built target plot_reductionIntrinsicsInterleave2 [ 66%] Built target plot_reductionIntrinsicsInterleave4 [ 68%] Built target plot_reductionIntrinsics [ 69%] Built target barycentre_vectorize_O3 [ 71%] Built target barycentre_intrinsics [ 72%] Built target barycentre_vectorizeSplit_O3 [ 73%] Built target barycentre_base_O3 [ 74%] Built target plot_barycentreIntrinsics [ 75%] Built target barycentre_base_O2 [ 77%] Built target barycentre_base_O1 [ 78%] Built target barycentre_base_O0 [ 79%] Built target barycentre_base_Ofast [ 83%] Built target plot_barycentreBase [ 85%] Built target plot_barycentreVectorize [ 86%] Built target sgemm_intrinsics_O3 [ 89%] Built target sgemm_vectorize_Ofast [ 90%] Built target sgemm_vectorize_O3 Scanning dependencies of target plot_sgemmIntrinsics [ 91%] Run sgemm_intrinsics_O3 program SGEMM Intrinsics evaluateSgemm : nbElement = 16, cyclePerElement = 5.51172 cy/el, elapsedTime = 1411 cy evaluateSgemm : nbElement = 24, cyclePerElement = 6.54167 cy/el, elapsedTime = 3768 cy evaluateSgemm : nbElement = 32, cyclePerElement = 7.54492 cy/el, elapsedTime = 7726 cy evaluateSgemm : nbElement = 56, cyclePerElement = 10.11 cy/el, elapsedTime = 31705 cy evaluateSgemm : nbElement = 80, cyclePerElement = 13.63 cy/el, elapsedTime = 87232 cy evaluateSgemm : nbElement = 104, cyclePerElement = 17.518 cy/el, elapsedTime = 189475 cy [ 92%] Call gnuplot sgemmIntrinsics [ 92%] Built target plot_sgemmIntrinsics [ 93%] Built target sgemm_base_O2 [ 95%] Built target sgemm_base_O1 [ 96%] Built target sgemm_base_O0 [ 98%] Built target plot_sgemmBase [100%] Built target plot_sgemmVectorize [100%] Built target plot_all |