Hypnos cluster alpaka matlab test

[erikzenker]

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