From 7321ba2fe1a43373076f253a35c2d180f07540b0 Mon Sep 17 00:00:00 2001 From: Nathalie Furmento <nathalie.furmento@labri.fr> Date: Tue, 4 Feb 2020 13:54:00 +0100 Subject: [PATCH] simucore: performance models when running potrf with default parameters --- .../sampling/codelets/45/sgemm.sirocco | 34 ++++++++++--------- .../sampling/codelets/45/splgsy.sirocco | 11 +++--- .../sampling/codelets/45/spotrf.sirocco | 26 +++++++------- .../sampling/codelets/45/ssyrk.sirocco | 34 ++++++++++--------- .../sampling/codelets/45/strsm.sirocco | 34 ++++++++++--------- 5 files changed, 75 insertions(+), 64 deletions(-) diff --git a/simucore/perfmodels/.starpu/sampling/codelets/45/sgemm.sirocco b/simucore/perfmodels/.starpu/sampling/codelets/45/sgemm.sirocco index f7e01361c..3a4a59491 100644 --- a/simucore/perfmodels/.starpu/sampling/codelets/45/sgemm.sirocco +++ b/simucore/perfmodels/.starpu/sampling/codelets/45/sgemm.sirocco @@ -7,12 +7,12 @@ # number of combinations 5 #################### -# COMB_4 +# COMB_0 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 0 #################### # DEV_0 @@ -26,9 +26,9 @@ # number of implementations 1 ##### -# Model for cpu0_impl0 (Comb4) +# Model for cpu0_impl0 (Comb0) # number of entries -5 +7 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 nan nan 0 0 0 # a b c @@ -41,14 +41,16 @@ a7cdf15b 44236800 0.000000e+00 2.633445e+05 5.249424e+04 6.32290 24c84a50 11059200 0.000000e+00 3.623402e+04 6.546678e+03 1.580528e+08 5.913839e+12 4362 c4a08f5f 2323200 0.000000e+00 3.732336e+03 6.841164e+02 3.950678e+07 1.524065e+11 10585 e6b94418 76800 0.000000e+00 3.040738e+01 3.904133e+00 2.440131e+06 7.542116e+07 80248 +a4820f8f 512000 0.000000e+00 2.424582e+02 4.362487e+01 1.672962e+04 4.187550e+06 69 +6a7210dc 1228800 0.000000e+00 1.962744e+03 4.745716e+02 4.710585e+04 9.786197e+07 24 #################### -# COMB_2 +# COMB_3 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -62,7 +64,7 @@ e6b94418 76800 0.000000e+00 3.040738e+01 3.904133e+00 2.44013 # number of implementations 1 ##### -# Model for cuda3_impl0 (Comb2) +# Model for cuda3_impl0 (Comb3) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx @@ -79,12 +81,12 @@ c4a08f5f 2323200 0.000000e+00 1.681587e+02 2.185477e+01 2.33219 e6b94418 76800 0.000000e+00 4.561876e+01 1.081980e+01 6.934052e+03 3.341172e+05 152 #################### -# COMB_0 +# COMB_1 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -98,7 +100,7 @@ e6b94418 76800 0.000000e+00 4.561876e+01 1.081980e+01 6.93405 # number of implementations 1 ##### -# Model for cuda0_impl0 (Comb0) +# Model for cuda0_impl0 (Comb1) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx @@ -115,12 +117,12 @@ c4a08f5f 2323200 0.000000e+00 1.678676e+02 2.188969e+01 2.44465 e6b94418 76800 0.000000e+00 5.712863e+01 1.016253e+01 2.182314e+04 1.286178e+06 382 #################### -# COMB_3 +# COMB_2 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -134,7 +136,7 @@ e6b94418 76800 0.000000e+00 5.712863e+01 1.016253e+01 2.18231 # number of implementations 1 ##### -# Model for cuda1_impl0 (Comb3) +# Model for cuda1_impl0 (Comb2) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx @@ -151,12 +153,12 @@ c4a08f5f 2323200 0.000000e+00 1.706067e+02 2.205696e+01 2.52293 e6b94418 76800 0.000000e+00 4.872868e+01 7.797709e+00 3.376898e+04 1.687655e+06 693 #################### -# COMB_1 +# COMB_4 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -170,7 +172,7 @@ e6b94418 76800 0.000000e+00 4.872868e+01 7.797709e+00 3.37689 # number of implementations 1 ##### -# Model for cuda2_impl0 (Comb1) +# Model for cuda2_impl0 (Comb4) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx diff --git a/simucore/perfmodels/.starpu/sampling/codelets/45/splgsy.sirocco b/simucore/perfmodels/.starpu/sampling/codelets/45/splgsy.sirocco index a7d3fadfd..bed8be833 100644 --- a/simucore/perfmodels/.starpu/sampling/codelets/45/splgsy.sirocco +++ b/simucore/perfmodels/.starpu/sampling/codelets/45/splgsy.sirocco @@ -7,12 +7,12 @@ # number of combinations 1 #################### -# COMB_4 +# COMB_0 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 0 #################### # DEV_0 @@ -26,9 +26,9 @@ # number of implementations 1 ##### -# Model for cpu0_impl0 (Comb4) +# Model for cpu0_impl0 (Comb0) # number of entries -5 +8 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 nan nan 0 0 0 # a b c @@ -41,4 +41,7 @@ nan nan nan 617e5fe6 3686400 0.000000e+00 8.939938e+03 2.153220e+02 1.231923e+07 1.101971e+11 1378 982013a8 774400 0.000000e+00 2.302314e+03 5.466484e+02 1.138264e+07 2.768379e+10 4944 ad30af9b 25600 0.000000e+00 8.731844e+01 8.227614e+00 1.389149e+06 1.223753e+08 15909 +2384211a 6400 0.000000e+00 3.342405e+01 1.257385e+01 6.684810e+02 2.550538e+04 20 +30b2a96f 51200 0.000000e+00 1.475864e+02 1.554810e+01 1.136415e+04 1.695808e+06 77 +2806844d 409600 0.000000e+00 8.902086e+02 1.436824e+02 1.370921e+05 1.252199e+08 154 diff --git a/simucore/perfmodels/.starpu/sampling/codelets/45/spotrf.sirocco b/simucore/perfmodels/.starpu/sampling/codelets/45/spotrf.sirocco index 98ef5d18d..98727a930 100644 --- a/simucore/perfmodels/.starpu/sampling/codelets/45/spotrf.sirocco +++ b/simucore/perfmodels/.starpu/sampling/codelets/45/spotrf.sirocco @@ -7,12 +7,12 @@ # number of combinations 5 #################### -# COMB_0 +# COMB_1 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -26,7 +26,7 @@ # number of implementations 1 ##### -# Model for cuda0_impl0 (Comb0) +# Model for cuda0_impl0 (Comb1) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx @@ -48,7 +48,7 @@ ad30af9b 25600 0.000000e+00 2.070017e+03 2.556262e+02 1.28341 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -84,7 +84,7 @@ ad30af9b 25600 0.000000e+00 2.102179e+03 3.123586e+02 1.30335 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -115,12 +115,12 @@ ad30af9b 25600 0.000000e+00 2.106632e+03 2.976700e+02 1.39037 617e5fe6 3686400 0.000000e+00 9.596093e+03 1.252122e+03 2.686906e+05 2.622279e+09 28 #################### -# COMB_1 +# COMB_4 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -134,7 +134,7 @@ ad30af9b 25600 0.000000e+00 2.106632e+03 2.976700e+02 1.39037 # number of implementations 1 ##### -# Model for cuda2_impl0 (Comb1) +# Model for cuda2_impl0 (Comb4) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx @@ -151,12 +151,12 @@ ad30af9b 25600 0.000000e+00 2.134159e+03 2.682985e+02 1.32317 617e5fe6 3686400 0.000000e+00 9.181625e+03 8.594056e+02 3.580834e+05 3.316592e+09 39 #################### -# COMB_4 +# COMB_0 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 0 #################### # DEV_0 @@ -170,9 +170,9 @@ ad30af9b 25600 0.000000e+00 2.134159e+03 2.682985e+02 1.32317 # number of implementations 1 ##### -# Model for cpu0_impl0 (Comb4) +# Model for cpu0_impl0 (Comb0) # number of entries -5 +7 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 nan nan 0 0 0 # a b c @@ -185,4 +185,6 @@ nan nan nan 982013a8 774400 0.000000e+00 1.554382e+03 1.710542e+02 5.595775e+04 8.803307e+07 36 ad30af9b 25600 0.000000e+00 7.450665e+01 4.146608e+00 4.097866e+03 3.062640e+05 55 617e5fe6 3686400 0.000000e+00 6.920145e+03 2.464678e+02 2.214446e+05 1.534373e+09 32 +2806844d 409600 0.000000e+00 1.036386e+03 1.971561e+02 6.011040e+04 6.455208e+07 58 +2384211a 6400 0.000000e+00 1.971042e+01 5.545559e+00 3.744980e+02 7.965824e+03 19 diff --git a/simucore/perfmodels/.starpu/sampling/codelets/45/ssyrk.sirocco b/simucore/perfmodels/.starpu/sampling/codelets/45/ssyrk.sirocco index 9892d9aaa..dcb50b092 100644 --- a/simucore/perfmodels/.starpu/sampling/codelets/45/ssyrk.sirocco +++ b/simucore/perfmodels/.starpu/sampling/codelets/45/ssyrk.sirocco @@ -7,12 +7,12 @@ # number of combinations 5 #################### -# COMB_0 +# COMB_1 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -26,7 +26,7 @@ # number of implementations 1 ##### -# Model for cuda0_impl0 (Comb0) +# Model for cuda0_impl0 (Comb1) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx @@ -43,12 +43,12 @@ ff82dda0 7372800 0.000000e+00 6.844636e+02 6.633616e+01 2.45722 5831f4e0 51200 0.000000e+00 6.011745e+01 1.013925e+01 1.322584e+03 8.177208e+04 22 #################### -# COMB_4 +# COMB_0 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 0 #################### # DEV_0 @@ -62,9 +62,9 @@ ff82dda0 7372800 0.000000e+00 6.844636e+02 6.633616e+01 2.45722 # number of implementations 1 ##### -# Model for cpu0_impl0 (Comb4) +# Model for cpu0_impl0 (Comb0) # number of entries -5 +7 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 nan nan 0 0 0 # a b c @@ -77,14 +77,16 @@ f001bd15 29491200 0.000000e+00 1.358370e+05 2.496094e+04 3.76268 ff82dda0 7372800 0.000000e+00 1.879259e+04 3.534541e+03 7.201320e+07 1.401187e+12 3832 6a2f38af 1548800 0.000000e+00 2.022561e+03 3.698244e+02 8.274299e+06 1.729480e+10 4091 5831f4e0 51200 0.000000e+00 2.572545e+01 4.086938e+00 1.642570e+05 4.332233e+06 6385 +e78115ba 57600 0.000000e+00 4.722764e+01 8.558222e+00 1.558512e+03 7.602186e+04 33 +3a47829c 819200 0.000000e+00 1.400813e+03 3.187817e+02 8.264794e+04 1.217700e+08 59 #################### -# COMB_3 +# COMB_2 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -98,7 +100,7 @@ ff82dda0 7372800 0.000000e+00 1.879259e+04 3.534541e+03 7.20132 # number of implementations 1 ##### -# Model for cuda1_impl0 (Comb3) +# Model for cuda1_impl0 (Comb2) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx @@ -115,12 +117,12 @@ ff82dda0 7372800 0.000000e+00 6.835967e+02 5.317005e+01 1.89356 5831f4e0 51200 0.000000e+00 6.087710e+01 1.044897e+01 6.087710e+02 3.815202e+04 10 #################### -# COMB_2 +# COMB_3 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -134,7 +136,7 @@ ff82dda0 7372800 0.000000e+00 6.835967e+02 5.317005e+01 1.89356 # number of implementations 1 ##### -# Model for cuda3_impl0 (Comb2) +# Model for cuda3_impl0 (Comb3) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx @@ -151,12 +153,12 @@ ff82dda0 7372800 0.000000e+00 6.738891e+02 5.051417e+01 2.45969 5831f4e0 51200 0.000000e+00 5.816600e+01 9.053481e+00 1.105154e+03 6.583973e+04 19 #################### -# COMB_1 +# COMB_4 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -170,7 +172,7 @@ ff82dda0 7372800 0.000000e+00 6.738891e+02 5.051417e+01 2.45969 # number of implementations 1 ##### -# Model for cuda2_impl0 (Comb1) +# Model for cuda2_impl0 (Comb4) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx diff --git a/simucore/perfmodels/.starpu/sampling/codelets/45/strsm.sirocco b/simucore/perfmodels/.starpu/sampling/codelets/45/strsm.sirocco index a04771f26..d9ac48768 100644 --- a/simucore/perfmodels/.starpu/sampling/codelets/45/strsm.sirocco +++ b/simucore/perfmodels/.starpu/sampling/codelets/45/strsm.sirocco @@ -7,12 +7,12 @@ # number of combinations 5 #################### -# COMB_4 +# COMB_0 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 0 #################### # DEV_0 @@ -26,9 +26,9 @@ # number of implementations 1 ##### -# Model for cpu0_impl0 (Comb4) +# Model for cpu0_impl0 (Comb0) # number of entries -5 +7 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx 0.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 nan nan 0 0 0 # a b c @@ -41,14 +41,16 @@ f001bd15 29491200 0.000000e+00 1.262354e+05 2.290170e+04 3.31241 ff82dda0 7372800 0.000000e+00 1.770446e+04 3.146495e+03 6.559501e+07 1.198005e+12 3705 6a2f38af 1548800 0.000000e+00 1.942615e+03 3.132030e+02 9.050645e+06 1.803895e+10 4659 5831f4e0 51200 0.000000e+00 3.124193e+01 7.098676e+00 1.885451e+05 6.194624e+06 6035 +22f3afbe 460800 0.000000e+00 1.461281e+02 1.995835e+01 5.114484e+03 7.613117e+05 35 +3a47829c 819200 0.000000e+00 2.006172e+03 3.121162e+02 1.263888e+05 2.596950e+08 63 #################### -# COMB_1 +# COMB_4 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -62,7 +64,7 @@ ff82dda0 7372800 0.000000e+00 1.770446e+04 3.146495e+03 6.55950 # number of implementations 1 ##### -# Model for cuda2_impl0 (Comb1) +# Model for cuda2_impl0 (Comb4) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx @@ -79,12 +81,12 @@ ff82dda0 7372800 0.000000e+00 2.363204e+03 3.410126e+02 1.07289 5831f4e0 51200 0.000000e+00 1.189339e+02 2.173609e+01 2.497612e+03 3.069724e+05 21 #################### -# COMB_3 +# COMB_2 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -98,7 +100,7 @@ ff82dda0 7372800 0.000000e+00 2.363204e+03 3.410126e+02 1.07289 # number of implementations 1 ##### -# Model for cuda1_impl0 (Comb3) +# Model for cuda1_impl0 (Comb2) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx @@ -115,12 +117,12 @@ ff82dda0 7372800 0.000000e+00 2.336819e+03 2.871770e+02 9.13696 5831f4e0 51200 0.000000e+00 6.696491e+01 1.309227e+01 2.209842e+03 1.536383e+05 33 #################### -# COMB_2 +# COMB_3 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -134,7 +136,7 @@ ff82dda0 7372800 0.000000e+00 2.336819e+03 2.871770e+02 9.13696 # number of implementations 1 ##### -# Model for cuda3_impl0 (Comb2) +# Model for cuda3_impl0 (Comb3) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx @@ -151,12 +153,12 @@ ff82dda0 7372800 0.000000e+00 2.366672e+03 3.276943e+02 8.40168 5831f4e0 51200 0.000000e+00 1.527756e+02 3.246283e+01 2.597185e+03 4.147017e+05 17 #################### -# COMB_0 +# COMB_1 # number of types devices 1 #################### # DEV_0 -# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, SCC - 4) +# device type (CPU - 0, CUDA - 1, OPENCL - 2, MIC - 3, MPI_MS - 5) 1 #################### # DEV_0 @@ -170,7 +172,7 @@ ff82dda0 7372800 0.000000e+00 2.366672e+03 3.276943e+02 8.40168 # number of implementations 1 ##### -# Model for cuda0_impl0 (Comb0) +# Model for cuda0_impl0 (Comb1) # number of entries 5 # sumlnx sumlnx2 sumlny sumlnxlny alpha beta n minx maxx -- GitLab