API tools faq
paste
Advertisement
Guest User

Untitled

a guest
Mar 14th, 2016
125
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 20.04 KB | None | 0 0
raw download clone embed print report
  1. root@pine64plus:~/hpl# for i in 1056000 1104000 1152000 1200000 1248000 1296000 ; do
  2. > echo -e "\n\n\nNow testing $i\n"
  3. > echo $i >/sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
  4. > echo performance >/sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
  5. > bin/OpenBLAS/xhpl
  6. > echo interactive >/sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
  7. > sleep 180
  8. > done
  9.  
  10.  
  11.  
  12. Now testing 1056000
  13.  
  14. ================================================================================
  15. HPLinpack 2.1 -- High-Performance Linpack benchmark -- October 26, 2012
  16. Written by A. Petitet and R. Clint Whaley, Innovative Computing Laboratory, UTK
  17. Modified by Piotr Luszczek, Innovative Computing Laboratory, UTK
  18. Modified by Julien Langou, University of Colorado Denver
  19. ================================================================================
  20.  
  21. An explanation of the input/output parameters follows:
  22. T/V : Wall time / encoded variant.
  23. N : The order of the coefficient matrix A.
  24. NB : The partitioning blocking factor.
  25. P : The number of process rows.
  26. Q : The number of process columns.
  27. Time : Time in seconds to solve the linear system.
  28. Gflops : Rate of execution for solving the linear system.
  29.  
  30. The following parameter values will be used:
  31.  
  32. N : 6000
  33. NB : 256
  34. PMAP : Row-major process mapping
  35. P : 1
  36. Q : 1
  37. PFACT : Left
  38. NBMIN : 2
  39. NDIV : 2
  40. RFACT : Right
  41. BCAST : 2ring
  42. DEPTH : 0
  43. SWAP : Mix (threshold = 64)
  44. L1 : transposed form
  45. U : transposed form
  46. EQUIL : yes
  47. ALIGN : 8 double precision words
  48.  
  49. --------------------------------------------------------------------------------
  50.  
  51. - The matrix A is randomly generated for each test.
  52. - The following scaled residual check will be computed:
  53. ||Ax-b||_oo / ( eps * ( || x ||_oo * || A ||_oo + || b ||_oo ) * N )
  54. - The relative machine precision (eps) is taken to be 1.110223e-16
  55. - Computational tests pass if scaled residuals are less than 16.0
  56.  
  57. ================================================================================
  58. T/V N NB P Q Time Gflops
  59. --------------------------------------------------------------------------------
  60. WR02R2L2 6000 256 1 1 50.11 2.875e+00
  61. HPL_pdgesv() start time Mon Mar 14 09:01:20 2016
  62.  
  63. HPL_pdgesv() end time Mon Mar 14 09:02:10 2016
  64.  
  65. --------------------------------------------------------------------------------
  66. ||Ax-b||_oo/(eps*(||A||_oo*||x||_oo+||b||_oo)*N)= 0.0048903 ...... PASSED
  67. ================================================================================
  68.  
  69. Finished 1 tests with the following results:
  70. 1 tests completed and passed residual checks,
  71. 0 tests completed and failed residual checks,
  72. 0 tests skipped because of illegal input values.
  73. --------------------------------------------------------------------------------
  74.  
  75. End of Tests.
  76. ================================================================================
  77.  
  78.  
  79.  
  80. Now testing 1104000
  81.  
  82. ================================================================================
  83. HPLinpack 2.1 -- High-Performance Linpack benchmark -- October 26, 2012
  84. Written by A. Petitet and R. Clint Whaley, Innovative Computing Laboratory, UTK
  85. Modified by Piotr Luszczek, Innovative Computing Laboratory, UTK
  86. Modified by Julien Langou, University of Colorado Denver
  87. ================================================================================
  88.  
  89. An explanation of the input/output parameters follows:
  90. T/V : Wall time / encoded variant.
  91. N : The order of the coefficient matrix A.
  92. NB : The partitioning blocking factor.
  93. P : The number of process rows.
  94. Q : The number of process columns.
  95. Time : Time in seconds to solve the linear system.
  96. Gflops : Rate of execution for solving the linear system.
  97.  
  98. The following parameter values will be used:
  99.  
  100. N : 6000
  101. NB : 256
  102. PMAP : Row-major process mapping
  103. P : 1
  104. Q : 1
  105. PFACT : Left
  106. NBMIN : 2
  107. NDIV : 2
  108. RFACT : Right
  109. BCAST : 2ring
  110. DEPTH : 0
  111. SWAP : Mix (threshold = 64)
  112. L1 : transposed form
  113. U : transposed form
  114. EQUIL : yes
  115. ALIGN : 8 double precision words
  116.  
  117. --------------------------------------------------------------------------------
  118.  
  119. - The matrix A is randomly generated for each test.
  120. - The following scaled residual check will be computed:
  121. ||Ax-b||_oo / ( eps * ( || x ||_oo * || A ||_oo + || b ||_oo ) * N )
  122. - The relative machine precision (eps) is taken to be 1.110223e-16
  123. - Computational tests pass if scaled residuals are less than 16.0
  124.  
  125. ================================================================================
  126. T/V N NB P Q Time Gflops
  127. --------------------------------------------------------------------------------
  128. WR02R2L2 6000 256 1 1 48.22 2.987e+00
  129. HPL_pdgesv() start time Mon Mar 14 09:05:21 2016
  130.  
  131. HPL_pdgesv() end time Mon Mar 14 09:06:09 2016
  132.  
  133. --------------------------------------------------------------------------------
  134. ||Ax-b||_oo/(eps*(||A||_oo*||x||_oo+||b||_oo)*N)= 0.0048903 ...... PASSED
  135. ================================================================================
  136.  
  137. Finished 1 tests with the following results:
  138. 1 tests completed and passed residual checks,
  139. 0 tests completed and failed residual checks,
  140. 0 tests skipped because of illegal input values.
  141. --------------------------------------------------------------------------------
  142.  
  143. End of Tests.
  144. ================================================================================
  145.  
  146.  
  147.  
  148. Now testing 1152000
  149.  
  150. ================================================================================
  151. HPLinpack 2.1 -- High-Performance Linpack benchmark -- October 26, 2012
  152. Written by A. Petitet and R. Clint Whaley, Innovative Computing Laboratory, UTK
  153. Modified by Piotr Luszczek, Innovative Computing Laboratory, UTK
  154. Modified by Julien Langou, University of Colorado Denver
  155. ================================================================================
  156.  
  157. An explanation of the input/output parameters follows:
  158. T/V : Wall time / encoded variant.
  159. N : The order of the coefficient matrix A.
  160. NB : The partitioning blocking factor.
  161. P : The number of process rows.
  162. Q : The number of process columns.
  163. Time : Time in seconds to solve the linear system.
  164. Gflops : Rate of execution for solving the linear system.
  165.  
  166. The following parameter values will be used:
  167.  
  168. N : 6000
  169. NB : 256
  170. PMAP : Row-major process mapping
  171. P : 1
  172. Q : 1
  173. PFACT : Left
  174. NBMIN : 2
  175. NDIV : 2
  176. RFACT : Right
  177. BCAST : 2ring
  178. DEPTH : 0
  179. SWAP : Mix (threshold = 64)
  180. L1 : transposed form
  181. U : transposed form
  182. EQUIL : yes
  183. ALIGN : 8 double precision words
  184.  
  185. --------------------------------------------------------------------------------
  186.  
  187. - The matrix A is randomly generated for each test.
  188. - The following scaled residual check will be computed:
  189. ||Ax-b||_oo / ( eps * ( || x ||_oo * || A ||_oo + || b ||_oo ) * N )
  190. - The relative machine precision (eps) is taken to be 1.110223e-16
  191. - Computational tests pass if scaled residuals are less than 16.0
  192.  
  193. ================================================================================
  194. T/V N NB P Q Time Gflops
  195. --------------------------------------------------------------------------------
  196. WR02R2L2 6000 256 1 1 45.95 3.135e+00
  197. HPL_pdgesv() start time Mon Mar 14 09:09:20 2016
  198.  
  199. HPL_pdgesv() end time Mon Mar 14 09:10:06 2016
  200.  
  201. --------------------------------------------------------------------------------
  202. ||Ax-b||_oo/(eps*(||A||_oo*||x||_oo+||b||_oo)*N)= 0.0048903 ...... PASSED
  203. ================================================================================
  204.  
  205. Finished 1 tests with the following results:
  206. 1 tests completed and passed residual checks,
  207. 0 tests completed and failed residual checks,
  208. 0 tests skipped because of illegal input values.
  209. --------------------------------------------------------------------------------
  210.  
  211. End of Tests.
  212. ================================================================================
  213.  
  214.  
  215.  
  216. Now testing 1200000
  217.  
  218. ================================================================================
  219. HPLinpack 2.1 -- High-Performance Linpack benchmark -- October 26, 2012
  220. Written by A. Petitet and R. Clint Whaley, Innovative Computing Laboratory, UTK
  221. Modified by Piotr Luszczek, Innovative Computing Laboratory, UTK
  222. Modified by Julien Langou, University of Colorado Denver
  223. ================================================================================
  224.  
  225. An explanation of the input/output parameters follows:
  226. T/V : Wall time / encoded variant.
  227. N : The order of the coefficient matrix A.
  228. NB : The partitioning blocking factor.
  229. P : The number of process rows.
  230. Q : The number of process columns.
  231. Time : Time in seconds to solve the linear system.
  232. Gflops : Rate of execution for solving the linear system.
  233.  
  234. The following parameter values will be used:
  235.  
  236. N : 6000
  237. NB : 256
  238. PMAP : Row-major process mapping
  239. P : 1
  240. Q : 1
  241. PFACT : Left
  242. NBMIN : 2
  243. NDIV : 2
  244. RFACT : Right
  245. BCAST : 2ring
  246. DEPTH : 0
  247. SWAP : Mix (threshold = 64)
  248. L1 : transposed form
  249. U : transposed form
  250. EQUIL : yes
  251. ALIGN : 8 double precision words
  252.  
  253. --------------------------------------------------------------------------------
  254.  
  255. - The matrix A is randomly generated for each test.
  256. - The following scaled residual check will be computed:
  257. ||Ax-b||_oo / ( eps * ( || x ||_oo * || A ||_oo + || b ||_oo ) * N )
  258. - The relative machine precision (eps) is taken to be 1.110223e-16
  259. - Computational tests pass if scaled residuals are less than 16.0
  260.  
  261. ================================================================================
  262. T/V N NB P Q Time Gflops
  263. --------------------------------------------------------------------------------
  264. WR02R2L2 6000 256 1 1 44.78 3.217e+00
  265. HPL_pdgesv() start time Mon Mar 14 09:13:16 2016
  266.  
  267. HPL_pdgesv() end time Mon Mar 14 09:14:01 2016
  268.  
  269. --------------------------------------------------------------------------------
  270. ||Ax-b||_oo/(eps*(||A||_oo*||x||_oo+||b||_oo)*N)= 0.0048903 ...... PASSED
  271. ================================================================================
  272.  
  273. Finished 1 tests with the following results:
  274. 1 tests completed and passed residual checks,
  275. 0 tests completed and failed residual checks,
  276. 0 tests skipped because of illegal input values.
  277. --------------------------------------------------------------------------------
  278.  
  279. End of Tests.
  280. ================================================================================
  281.  
  282.  
  283.  
  284. Now testing 1248000
  285.  
  286. ================================================================================
  287. HPLinpack 2.1 -- High-Performance Linpack benchmark -- October 26, 2012
  288. Written by A. Petitet and R. Clint Whaley, Innovative Computing Laboratory, UTK
  289. Modified by Piotr Luszczek, Innovative Computing Laboratory, UTK
  290. Modified by Julien Langou, University of Colorado Denver
  291. ================================================================================
  292.  
  293. An explanation of the input/output parameters follows:
  294. T/V : Wall time / encoded variant.
  295. N : The order of the coefficient matrix A.
  296. NB : The partitioning blocking factor.
  297. P : The number of process rows.
  298. Q : The number of process columns.
  299. Time : Time in seconds to solve the linear system.
  300. Gflops : Rate of execution for solving the linear system.
  301.  
  302. The following parameter values will be used:
  303.  
  304. N : 6000
  305. NB : 256
  306. PMAP : Row-major process mapping
  307. P : 1
  308. Q : 1
  309. PFACT : Left
  310. NBMIN : 2
  311. NDIV : 2
  312. RFACT : Right
  313. BCAST : 2ring
  314. DEPTH : 0
  315. SWAP : Mix (threshold = 64)
  316. L1 : transposed form
  317. U : transposed form
  318. EQUIL : yes
  319. ALIGN : 8 double precision words
  320.  
  321. --------------------------------------------------------------------------------
  322.  
  323. - The matrix A is randomly generated for each test.
  324. - The following scaled residual check will be computed:
  325. ||Ax-b||_oo / ( eps * ( || x ||_oo * || A ||_oo + || b ||_oo ) * N )
  326. - The relative machine precision (eps) is taken to be 1.110223e-16
  327. - Computational tests pass if scaled residuals are less than 16.0
  328.  
  329. ================================================================================
  330. T/V N NB P Q Time Gflops
  331. --------------------------------------------------------------------------------
  332. WR02R2L2 6000 256 1 1 43.44 3.316e+00
  333. HPL_pdgesv() start time Mon Mar 14 09:17:10 2016
  334.  
  335. HPL_pdgesv() end time Mon Mar 14 09:17:54 2016
  336.  
  337. --------------------------------------------------------------------------------
  338. ||Ax-b||_oo/(eps*(||A||_oo*||x||_oo+||b||_oo)*N)= 0.0048903 ...... PASSED
  339. ================================================================================
  340.  
  341. Finished 1 tests with the following results:
  342. 1 tests completed and passed residual checks,
  343. 0 tests completed and failed residual checks,
  344. 0 tests skipped because of illegal input values.
  345. --------------------------------------------------------------------------------
  346.  
  347. End of Tests.
  348. ================================================================================
  349.  
  350.  
  351.  
  352. Now testing 1296000
  353.  
  354. ================================================================================
  355. HPLinpack 2.1 -- High-Performance Linpack benchmark -- October 26, 2012
  356. Written by A. Petitet and R. Clint Whaley, Innovative Computing Laboratory, UTK
  357. Modified by Piotr Luszczek, Innovative Computing Laboratory, UTK
  358. Modified by Julien Langou, University of Colorado Denver
  359. ================================================================================
  360.  
  361. An explanation of the input/output parameters follows:
  362. T/V : Wall time / encoded variant.
  363. N : The order of the coefficient matrix A.
  364. NB : The partitioning blocking factor.
  365. P : The number of process rows.
  366. Q : The number of process columns.
  367. Time : Time in seconds to solve the linear system.
  368. Gflops : Rate of execution for solving the linear system.
  369.  
  370. The following parameter values will be used:
  371.  
  372. N : 6000
  373. NB : 256
  374. PMAP : Row-major process mapping
  375. P : 1
  376. Q : 1
  377. PFACT : Left
  378. NBMIN : 2
  379. NDIV : 2
  380. RFACT : Right
  381. BCAST : 2ring
  382. DEPTH : 0
  383. SWAP : Mix (threshold = 64)
  384. L1 : transposed form
  385. U : transposed form
  386. EQUIL : yes
  387. ALIGN : 8 double precision words
  388.  
  389. --------------------------------------------------------------------------------
  390.  
  391. - The matrix A is randomly generated for each test.
  392. - The following scaled residual check will be computed:
  393. ||Ax-b||_oo / ( eps * ( || x ||_oo * || A ||_oo + || b ||_oo ) * N )
  394. - The relative machine precision (eps) is taken to be 1.110223e-16
  395. - Computational tests pass if scaled residuals are less than 16.0
  396.  
  397. ================================================================================
  398. T/V N NB P Q Time Gflops
  399. --------------------------------------------------------------------------------
  400. WR02R2L2 6000 256 1 1 42.33 3.403e+00
  401. HPL_pdgesv() start time Mon Mar 14 09:21:03 2016
  402.  
  403. HPL_pdgesv() end time Mon Mar 14 09:21:45 2016
  404.  
  405. --------------------------------------------------------------------------------
  406. ||Ax-b||_oo/(eps*(||A||_oo*||x||_oo+||b||_oo)*N)= 0.0048903 ...... PASSED
  407. ================================================================================
  408.  
  409. Finished 1 tests with the following results:
  410. 1 tests completed and passed residual checks,
  411. 0 tests completed and failed residual checks,
  412. 0 tests skipped because of illegal input values.
  413. --------------------------------------------------------------------------------
  414.  
  415. End of Tests.
  416. ================================================================================
  417. ^C
  418. root@pine64plus:~/hpl# echo 1344000 >/sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
  419. root@pine64plus:~/hpl# bin/OpenBLAS/xhpl
  420. ================================================================================
  421. HPLinpack 2.1 -- High-Performance Linpack benchmark -- October 26, 2012
  422. Written by A. Petitet and R. Clint Whaley, Innovative Computing Laboratory, UTK
  423. Modified by Piotr Luszczek, Innovative Computing Laboratory, UTK
  424. Modified by Julien Langou, University of Colorado Denver
  425. ================================================================================
  426.  
  427. An explanation of the input/output parameters follows:
  428. T/V : Wall time / encoded variant.
  429. N : The order of the coefficient matrix A.
  430. NB : The partitioning blocking factor.
  431. P : The number of process rows.
  432. Q : The number of process columns.
  433. Time : Time in seconds to solve the linear system.
  434. Gflops : Rate of execution for solving the linear system.
  435.  
  436. The following parameter values will be used:
  437.  
  438. N : 6000
  439. NB : 256
  440. PMAP : Row-major process mapping
  441. P : 1
  442. Q : 1
  443. PFACT : Left
  444. NBMIN : 2
  445. NDIV : 2
  446. RFACT : Right
  447. BCAST : 2ring
  448. DEPTH : 0
  449. SWAP : Mix (threshold = 64)
  450. L1 : transposed form
  451. U : transposed form
  452. EQUIL : yes
  453. ALIGN : 8 double precision words
  454.  
  455. --------------------------------------------------------------------------------
  456.  
  457. - The matrix A is randomly generated for each test.
  458. - The following scaled residual check will be computed:
  459. ||Ax-b||_oo / ( eps * ( || x ||_oo * || A ||_oo + || b ||_oo ) * N )
  460. - The relative machine precision (eps) is taken to be 1.110223e-16
  461. - Computational tests pass if scaled residuals are less than 16.0
  462.  
  463. ================================================================================
  464. T/V N NB P Q Time Gflops
  465. --------------------------------------------------------------------------------
  466. WR02R2L2 6000 256 1 1 41.73 3.452e+00
  467. HPL_pdgesv() start time Mon Mar 14 09:23:49 2016
  468.  
  469. HPL_pdgesv() end time Mon Mar 14 09:24:31 2016
  470.  
  471. --------------------------------------------------------------------------------
  472. ||Ax-b||_oo/(eps*(||A||_oo*||x||_oo+||b||_oo)*N)= 2994406363.6726804 ...... FAILED
  473. ||Ax-b||_oo . . . . . . . . . . . . . . . . . = 9.799723
  474. ||A||_oo . . . . . . . . . . . . . . . . . . . = 1540.184305
  475. ||A||_1 . . . . . . . . . . . . . . . . . . . = 1543.558216
  476. ||x||_oo . . . . . . . . . . . . . . . . . . . = 3.189515
  477. ||x||_1 . . . . . . . . . . . . . . . . . . . = 3877.362391
  478. ||b||_oo . . . . . . . . . . . . . . . . . . . = 0.499944
  479. ================================================================================
  480.  
  481. Finished 1 tests with the following results:
  482. 0 tests completed and passed residual checks,
  483. 1 tests completed and failed residual checks,
  484. 0 tests skipped because of illegal input values.
  485. --------------------------------------------------------------------------------
  486.  
  487. End of Tests.
  488. ================================================================================
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement
Public Pastes
Advertisement
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!