Hypnos cluster alpaka matlab test #110
Description
I repeated the tests from #58 to verify the results of the alpaka kernel ports of
commit 2f5cf96971c8 with openMP accelerator.
Setup on Hypnos Laser Nodes
- 10 nodes, each with 64 cores
- gcc 4.9.2
- openMPI 1.8.6
- MPI errors on 1.8.4
- matlab 2014a
- export OMPI_MCA_orte_precondition_transports=0099b3eaa2c1547e-afb287789133a954
- softlinks to new glibc version by admin
Test Run ALPAKA
- 10 laser nodes, each with 64 cores
- Elapsed time 6491 s
Test Run CUDA native with bug fix
- 2 k20 nodes
- Elapsed time 6101s
Result
- A small deviation in comparison to cuda is visible (~ 3 * 10^-2)
- This version contains bug fix
- Values are still in strong agreement with the experimental measurements.
- Cuda bugfix vs. cuda before bugfix has deviation of ~1 * 10^-3
| Timestep | cuda native 268b6cd | alpaka openMP 2f5cf96 | cuda native bugfix 825c6c7 |
|---|---|---|---|
| 0 | 0.7992 | 0.7992 | 0.7992 |
| 1 | 0.8296 | 0.8297 | 0.8297 |
| 2 | 0.8606 | 0.8607 | 0.8607 |
| 3 | 0.8921 | 0.8921 | 0.8921 |
| 4 | 0.9240 | 0.9241 | 0.9240 |
| 5 | 0.9562 | 0.9564 | 0.9563 |
| 6 | 0.9888 | 0.9890 | 0.9889 |
| 7 | 1.0217 | 1.0218 | 1.0217 |
| 8 | 1.0547 | 1.0549 | 1.0548 |
| 9 | 1.0879 | 1.0880 | 1.0880 |
| 10 | 1.1211 | 1.1212 | 1.1212 |
| 11 | 1.1544 | 1.1544 | 1.1544 |
| 12 | 1.1875 | 1.1874 | 1.1876 |
| 13 | 1.2205 | 1.2203 | 1.2206 |
| 14 | 1.2534 | 1.2530 | 1.2534 |
| 15 | 1.2859 | 1.2853 | 1.2860 |
| 16 | 1.3182 | 1.3173 | 1.3182 |
| 17 | 1.3500 | 1.3488 | 1.3501 |
| 18 | 1.3814 | 1.3798 | 1.3815 |
| 19 | 1.4123 | 1.4103 | 1.4124 |
| 20 | 1.4427 | 1.4402 | 1.4428 |
| 21 | 1.4725 | 1.4695 | 1.4726 |
| 22 | 1.5017 | 1.4981 | 1.5018 |
| 23 | 1.5302 | 1.5259 | 1.5303 |
| 24 | 1.5580 | 1.5530 | 1.5582 |
| 25 | 1.5851 | 1.5794 | 1.5852 |
| 26 | 1.6115 | 1.6049 | 1.6116 |
| 27 | 1.6370 | 1.6297 | 1.6371 |
| 28 | 1.6618 | 1.6535 | 1.6619 |
| 29 | 1.6857 | 1.6766 | 1.6859 |
| 30 | 1.7089 | 1.6988 | 1.7090 |
| 31 | 1.7312 | 1.7202 | 1.7313 |
| 32 | 1.7527 | 1.7407 | 1.7528 |
| 33 | 1.7733 | 1.7604 | 1.7735 |
| 34 | 1.7932 | 1.7793 | 1.7934 |
| 35 | 1.8122 | 1.7974 | 1.8125 |
| 36 | 1.8305 | 1.8146 | 1.8308 |
| 37 | 1.8479 | 1.8311 | 1.8483 |
| 38 | 1.8646 | 1.8468 | 1.8650 |
| 39 | 1.8806 | 1.8617 | 1.8809 |
| 40 | 1.8958 | 1.8759 | 1.8962 |
| 41 | 1.9103 | 1.8895 | 1.9107 |
| 42 | 1.9241 | 1.9023 | 1.9245 |
| 43 | 1.9373 | 1.9145 | 1.9376 |
| 44 | 1.9498 | 1.9260 | 1.9501 |
| 45 | 1.9617 | 1.9369 | 1.9620 |
| 46 | 1.9729 | 1.9472 | 1.9733 |
| 47 | 1.9836 | 1.9570 | 1.9839 |
| 48 | 1.9937 | 1.9662 | 1.9940 |
| 49 | 2.0032 | 1.9749 | 2.0036 |
| 50 | 2.0123 | 1.9831 | 2.0126 |
| 51 | 1.9502 | 1.9211 | 1.9506 |
| 52 | 1.8932 | 1.8644 | 1.8936 |
| 53 | 1.8407 | 1.8122 | 1.8410 |
| 54 | 1.7922 | 1.7641 | 1.7925 |
| 55 | 1.7472 | 1.7196 | 1.7476 |
| 56 | 1.7055 | 1.6783 | 1.7058 |
| 57 | 1.6667 | 1.6400 | 1.6669 |
| 58 | 1.6304 | 1.6042 | 1.6307 |
| 59 | 1.5964 | 1.5708 | 1.5967 |
| 60 | 1.5646 | 1.5396 | 1.5649 |
| 61 | 1.5347 | 1.5103 | 1.5350 |
| 62 | 1.5066 | 1.4828 | 1.5069 |
| 63 | 1.4802 | 1.4569 | 1.4805 |
| 64 | 1.4552 | 1.4324 | 1.4555 |
| 65 | 1.4316 | 1.4094 | 1.4318 |
| 66 | 1.4092 | 1.3876 | 1.4095 |
| 67 | 1.3880 | 1.3670 | 1.3883 |
| 68 | 1.3679 | 1.3474 | 1.3682 |
| 69 | 1.3488 | 1.3289 | 1.3490 |
| 70 | 1.3307 | 1.3112 | 1.3308 |
| 71 | 1.3133 | 1.2945 | 1.3135 |
| 72 | 1.2968 | 1.2785 | 1.2970 |
| 73 | 1.2811 | 1.2633 | 1.2813 |
| 74 | 1.2661 | 1.2488 | 1.2663 |
| 75 | 1.2517 | 1.2349 | 1.2519 |
| 76 | 1.2380 | 1.2216 | 1.2381 |
| 77 | 1.2248 | 1.2089 | 1.2250 |
| 78 | 1.2122 | 1.1968 | 1.2124 |
| 79 | 1.2001 | 1.1852 | 1.2003 |
| 80 | 1.1885 | 1.1740 | 1.1887 |
| 81 | 1.1774 | 1.1633 | 1.1775 |
| 82 | 1.1667 | 1.1530 | 1.1668 |
| 83 | 1.1564 | 1.1431 | 1.1566 |
| 84 | 1.1465 | 1.1336 | 1.1467 |
| 85 | 1.1370 | 1.1245 | 1.1371 |
| 86 | 1.1278 | 1.1157 | 1.1280 |
| 87 | 1.1190 | 1.1072 | 1.1191 |
| 88 | 1.1105 | 1.0990 | 1.1106 |
| 89 | 1.1023 | 1.0912 | 1.1024 |
| 90 | 1.0944 | 1.0836 | 1.0945 |
| 91 | 1.0867 | 1.0762 | 1.0868 |
| 92 | 1.0793 | 1.0691 | 1.0794 |
| 93 | 1.0722 | 1.0623 | 1.0723 |
| 94 | 1.0653 | 1.0557 | 1.0654 |
| 95 | 1.0586 | 1.0493 | 1.0587 |
| 96 | 1.0521 | 1.0431 | 1.0522 |
| 97 | 1.0459 | 1.0371 | 1.0460 |
| 98 | 1.0398 | 1.0313 | 1.0399 |
| 99 | 1.0340 | 1.0257 | 1.0341 |
| 100 | 1.0283 | 1.0203 | 1.0284 |
| 101 | 1.0228 | 1.0150 | 1.0229 |
| 102 | 1.0174 | 1.0099 | 1.0175 |
| 103 | 1.0123 | 1.0049 | 1.0123 |
| 104 | 1.0072 | 1.0001 | 1.0073 |
| 105 | 1.0024 | 0.9955 | 1.0024 |
| 106 | 0.9976 | 0.9909 | 0.9977 |
| 107 | 0.9930 | 0.9865 | 0.9931 |
| 108 | 0.9886 | 0.9823 | 0.9887 |
| 109 | 0.9842 | 0.9781 | 0.9843 |
| 110 | 0.9800 | 0.9741 | 0.9801 |
| 111 | 0.9759 | 0.9702 | 0.9760 |
| 112 | 0.9720 | 0.9664 | 0.9720 |
| 113 | 0.9681 | 0.9627 | 0.9682 |
| 114 | 0.9643 | 0.9591 | 0.9644 |
| 115 | 0.9607 | 0.9555 | 0.9607 |
| 116 | 0.9571 | 0.9521 | 0.9572 |
| 117 | 0.9536 | 0.9488 | 0.9537 |
| 118 | 0.9502 | 0.9456 | 0.9503 |
| 119 | 0.9469 | 0.9424 | 0.9470 |
| 120 | 0.9437 | 0.9394 | 0.9438 |
| 121 | 0.9406 | 0.9364 | 0.9407 |
| 122 | 0.9376 | 0.9334 | 0.9376 |
| 123 | 0.9346 | 0.9306 | 0.9347 |
| 124 | 0.9317 | 0.9278 | 0.9318 |
| 125 | 0.9289 | 0.9251 | 0.9290 |
| 126 | 0.9262 | 0.9225 | 0.9262 |
| 127 | 0.9235 | 0.9199 | 0.9235 |
| 128 | 0.9209 | 0.9174 | 0.9209 |
| 129 | 0.9183 | 0.9150 | 0.9184 |
| 130 | 0.9158 | 0.9126 | 0.9159 |
| 131 | 0.9134 | 0.9103 | 0.9135 |
| 132 | 0.9110 | 0.9080 | 0.9111 |
| 133 | 0.9087 | 0.9058 | 0.9088 |
| 134 | 0.9065 | 0.9036 | 0.9065 |
| 135 | 0.9043 | 0.9015 | 0.9043 |
| 136 | 0.9021 | 0.8995 | 0.9022 |
| 137 | 0.9000 | 0.8974 | 0.9001 |
| 138 | 0.8980 | 0.8955 | 0.8980 |
| 139 | 0.8960 | 0.8936 | 0.8960 |