mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Software

Reply
 
Thread Tools
Old 2012-05-01, 21:07   #67
harlee
 
harlee's Avatar
 
Sep 2006
Odenton, MD, USA

22·41 Posts
Default

OK, did a couple of tests on my P-4 Windows XP SP3 system to see which is faser, 26.6 or 27.6. Got mixed results. The benchmark test for 26.6 seems to be slightly faster along with P-1 using a 640K FFT, while a P-1 using 2880K FFT seems to be slightly faster on 27.6.

V26.6
Code:
[Sat Apr 28 13:55:07 2012]
Compare your results to other computers at http://www.mersenne.org/report_benchmarks
Intel(R) Pentium(R) 4 CPU 2.60GHz
CPU speed: 2593.89 MHz, with hyperthreading
CPU features: Prefetch, MMX, SSE, SSE2
L1 cache size: 8 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 128 bytes
TLBS: 64
Prime95 32-bit version 26.6, RdtscTiming=1
Best time for 64K FFT length: 1.824 ms., avg: 1.853 ms.
Best time for 80K FFT length: 2.322 ms., avg: 2.345 ms.
Best time for 96K FFT length: 2.946 ms., avg: 3.036 ms.
Best time for 112K FFT length: 3.425 ms., avg: 3.464 ms.
Best time for 128K FFT length: 3.830 ms., avg: 3.880 ms.
Best time for 160K FFT length: 4.760 ms., avg: 4.902 ms.
Best time for 192K FFT length: 5.761 ms., avg: 6.351 ms.
Best time for 224K FFT length: 7.000 ms., avg: 7.126 ms.
Best time for 256K FFT length: 7.804 ms., avg: 7.965 ms.
Best time for 320K FFT length: 9.936 ms., avg: 10.149 ms.
Best time for 384K FFT length: 12.083 ms., avg: 12.332 ms.
Best time for 448K FFT length: 14.571 ms., avg: 14.861 ms.
Best time for 512K FFT length: 16.251 ms., avg: 16.587 ms.
Best time for 640K FFT length: 20.944 ms., avg: 21.255 ms.
Best time for 768K FFT length: 25.525 ms., avg: 25.983 ms.
Best time for 896K FFT length: 30.655 ms., avg: 31.063 ms.
Best time for 1024K FFT length: 34.200 ms., avg: 34.661 ms.
Best time for 1280K FFT length: 43.777 ms., avg: 44.505 ms.
Best time for 1536K FFT length: 53.659 ms., avg: 54.313 ms.
Best time for 1792K FFT length: 64.562 ms., avg: 65.337 ms.
Best time for 2048K FFT length: 72.078 ms., avg: 72.862 ms.
Best time for 2560K FFT length: 95.528 ms., avg: 96.222 ms.
Best time for 3072K FFT length: 117.047 ms., avg: 117.908 ms.
Best time for 3584K FFT length: 148.105 ms., avg: 149.117 ms.
Best time for 4096K FFT length: 162.204 ms., avg: 163.664 ms.
Best time for 5120K FFT length: 276.511 ms., avg: 279.103 ms.
Best time for 6144K FFT length: 279.257 ms., avg: 281.584 ms.
Best time for 7168K FFT length: 391.241 ms., avg: 393.953 ms.
Best time for 8192K FFT length: 487.863 ms., avg: 491.606 ms.
Timing FFTs using 2 threads on 1 physical CPUs.
Best time for 64K FFT length: 2.609 ms., avg: 2.640 ms.
Best time for 80K FFT length: 2.277 ms., avg: 2.363 ms.
Best time for 96K FFT length: 2.743 ms., avg: 2.793 ms.
Best time for 112K FFT length: 3.333 ms., avg: 3.459 ms.
Best time for 128K FFT length: 3.666 ms., avg: 3.698 ms.
Best time for 160K FFT length: 4.810 ms., avg: 4.906 ms.
Best time for 192K FFT length: 5.704 ms., avg: 5.801 ms.
Best time for 224K FFT length: 6.879 ms., avg: 7.017 ms.
Best time for 256K FFT length: 7.631 ms., avg: 7.801 ms.
Best time for 320K FFT length: 9.589 ms., avg: 9.736 ms.
Best time for 384K FFT length: 11.789 ms., avg: 12.045 ms.
Best time for 448K FFT length: 14.399 ms., avg: 14.673 ms.
Best time for 512K FFT length: 16.240 ms., avg: 16.574 ms.
Best time for 640K FFT length: 19.836 ms., avg: 20.171 ms.
Best time for 768K FFT length: 25.125 ms., avg: 25.743 ms.
Best time for 896K FFT length: 30.853 ms., avg: 31.423 ms.
Best time for 1024K FFT length: 34.997 ms., avg: 35.517 ms.
Best time for 1280K FFT length: 44.059 ms., avg: 44.896 ms.
Best time for 1536K FFT length: 53.218 ms., avg: 54.180 ms.
Best time for 1792K FFT length: 66.262 ms., avg: 66.901 ms.
Best time for 2048K FFT length: 74.350 ms., avg: 75.048 ms.
Best time for 2560K FFT length: 99.141 ms., avg: 102.282 ms.
Best time for 3072K FFT length: 129.108 ms., avg: 130.607 ms.
Best time for 3584K FFT length: 205.542 ms., avg: 208.048 ms.
Best time for 4096K FFT length: 205.281 ms., avg: 206.790 ms.
Best time for 5120K FFT length: 284.527 ms., avg: 293.391 ms.
Best time for 6144K FFT length: 312.381 ms., avg: 323.333 ms.
Best time for 7168K FFT length: 429.406 ms., avg: 442.387 ms.
Best time for 8192K FFT length: 645.930 ms., avg: 651.123 ms.
Best time for 61 bit trial factors: 11.045 ms.
Best time for 62 bit trial factors: 11.130 ms.
Best time for 63 bit trial factors: 12.205 ms.
Best time for 64 bit trial factors: 12.198 ms.
Best time for 65 bit trial factors: 14.881 ms.
Best time for 66 bit trial factors: 14.862 ms.
Best time for 67 bit trial factors: 14.863 ms.
Best time for 75 bit trial factors: 15.190 ms.
Best time for 76 bit trial factors: 15.226 ms.
Best time for 77 bit trial factors: 15.163 ms.

[Apr 27 20:53] Worker starting
[Apr 27 20:53] Setting affinity to run worker on logical CPUs 1,2
[Apr 27 20:53] Optimal P-1 factoring of M12006409 using up to 1280MB of memory.
[Apr 27 20:53] Assuming no factors below 2^64 and 3 primality tests saved if a factor is found.
[Apr 27 20:53] Optimal bounds are B1=230000, B2=5635000
[Apr 27 20:53] Chance of finding a factor is an estimated 6.08%
[Apr 27 20:53] Using Pentium4 type-3 FFT length 640K, Pass1=320, Pass2=2K
[Apr 27 20:53] M12006409 stage 1 is 71.50% complete.
[Apr 27 20:57] M12006409 stage 1 is 74.51% complete. Time: 221.923 sec.
[Apr 27 21:00] M12006409 stage 1 is 77.52% complete. Time: 221.543 sec.
[Apr 27 21:04] M12006409 stage 1 is 80.53% complete. Time: 221.726 sec.
[Apr 27 21:08] M12006409 stage 1 is 83.54% complete. Time: 226.453 sec.
[Apr 27 21:11] M12006409 stage 1 is 86.55% complete. Time: 222.828 sec.
[Apr 27 21:15] M12006409 stage 1 is 89.57% complete. Time: 221.781 sec.
[Apr 27 21:19] M12006409 stage 1 is 92.58% complete. Time: 223.572 sec.
[Apr 27 21:23] M12006409 stage 1 is 95.59% complete. Time: 221.867 sec.
[Apr 27 21:26] M12006409 stage 1 is 98.60% complete. Time: 223.108 sec.
[Apr 27 21:28] M12006409 stage 1 complete. 189276 transforms. Time: 2107.934 sec.
[Apr 27 21:28] Starting stage 1 GCD - please be patient.
[Apr 27 21:29] Stage 1 GCD complete. Time: 39.914 sec.
[Apr 27 21:29] Using 1276MB of memory.  Processing 242 relative primes (0 of 480 already processed).
[Apr 27 21:34] M12006409 stage 2 is 1.71% complete. Time: 297.292 sec.
[Apr 27 21:39] M12006409 stage 2 is 4.88% complete. Time: 291.127 sec.
[Apr 27 21:43] M12006409 stage 2 is 8.04% complete. Time: 291.062 sec.
[Apr 27 21:48] M12006409 stage 2 is 11.20% complete. Time: 291.518 sec.
[Apr 27 21:53] M12006409 stage 2 is 14.35% complete. Time: 291.901 sec.
[Apr 27 21:58] M12006409 stage 2 is 17.51% complete. Time: 291.081 sec.
[Apr 27 21:58] Worker stopped.

[Apr 28 16:29] Worker starting
[Apr 28 16:29] Setting affinity to run worker on logical CPUs 1,2
[Apr 28 16:29] Trying 1000 iterations for exponent 54933283 using 2880K FFT.
[Apr 28 16:29] If average roundoff error is above 0.24275, then a larger FFT will be used.
[Apr 28 16:29] After 100 iterations average roundoff error is 0.21141.
[Apr 28 16:29] After 200 iterations average roundoff error is 0.2115.
[Apr 28 16:29] After 300 iterations average roundoff error is 0.21239.
[Apr 28 16:30] After 400 iterations average roundoff error is 0.21186.
[Apr 28 16:30] After 500 iterations average roundoff error is 0.21181.
[Apr 28 16:30] After 600 iterations average roundoff error is 0.21182.
[Apr 28 16:30] After 700 iterations average roundoff error is 0.21179.
[Apr 28 16:30] After 800 iterations average roundoff error is 0.21184.
[Apr 28 16:31] After 900 iterations average roundoff error is 0.21174.
[Apr 28 16:31] Final average roundoff error is 0.21167, using 2880K FFT for exponent 54933283.
[Apr 28 16:31] Optimal P-1 factoring of M54933283 using up to 1280MB of memory.
[Apr 28 16:31] Assuming no factors below 2^71 and 2 primality tests saved if a factor is found.
[Apr 28 16:31] Optimal bounds are B1=580000, B2=12470000
[Apr 28 16:31] Chance of finding a factor is an estimated 4.69%
[Apr 28 16:31] Using Pentium4 type-3 FFT length 2880K, Pass1=640, Pass2=4608
[Apr 28 16:31] M54933283 stage 1 is 88.07% complete.
[Apr 28 16:51] M54933283 stage 1 is 89.26% complete. Time: 1209.709 sec.
[Apr 28 17:11] M54933283 stage 1 is 90.46% complete. Time: 1210.954 sec.
[Apr 28 17:32] M54933283 stage 1 is 91.65% complete. Time: 1233.806 sec.
[Apr 28 17:52] M54933283 stage 1 is 92.85% complete. Time: 1209.989 sec.
[Apr 28 18:12] M54933283 stage 1 is 94.04% complete. Time: 1212.068 sec.
[Apr 28 18:32] M54933283 stage 1 is 95.24% complete. Time: 1220.105 sec.
[Apr 28 18:53] M54933283 stage 1 is 96.43% complete. Time: 1208.332 sec.
[Apr 28 19:13] M54933283 stage 1 is 97.63% complete. Time: 1231.327 sec.
[Apr 28 19:34] M54933283 stage 1 is 98.82% complete. Time: 1253.163 sec.
[Apr 28 19:54] M54933283 stage 1 complete. 199598 transforms. Time: 12173.833 sec.
[Apr 28 19:54] Starting stage 1 GCD - please be patient.
[Apr 28 19:59] Stage 1 GCD complete. Time: 298.472 sec.
[Apr 28 19:59] Using 1268MB of memory.  Processing 50 relative primes (0 of 480 already processed).
[Apr 28 20:25] M54933283 stage 2 is 1.40% complete. Time: 1557.714 sec.
[Apr 28 20:50] M54933283 stage 2 is 2.85% complete. Time: 1543.487 sec.
[Apr 28 21:16] M54933283 stage 2 is 4.30% complete. Time: 1545.235 sec.
[Apr 28 21:42] M54933283 stage 2 is 5.74% complete. Time: 1548.738 sec.
[Apr 28 22:08] M54933283 stage 2 is 7.19% complete. Time: 1543.258 sec.
[Apr 28 22:34] M54933283 stage 2 is 8.64% complete. Time: 1553.157 sec.
[Apr 28 22:59] M54933283 stage 2 is 10.09% complete. Time: 1543.699 sec.
[Apr 28 23:05] Using 1268MB of memory.  Processing 50 relative primes (50 of 480 already processed).
[Apr 28 23:25] M54933283 stage 2 is 11.49% complete. Time: 1539.700 sec.
[Apr 28 23:39] Worker stopped.
harlee is offline   Reply With Quote
Old 2012-05-01, 21:09   #68
harlee
 
harlee's Avatar
 
Sep 2006
Odenton, MD, USA

2448 Posts
Default

Here are the results for 27.6 (had to break it up into two posts):

Code:
[Sat Apr 28 14:00:49 2012]
Compare your results to other computers at http://www.mersenne.org/report_benchmarks
Intel(R) Pentium(R) 4 CPU 2.60GHz
CPU speed: 2593.63 MHz, with hyperthreading
CPU features: Prefetch, SSE, SSE2
L1 cache size: 8 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 128 bytes
TLBS: 64
Prime95 32-bit version 27.6, RdtscTiming=1
Best time for 64K FFT length: 1.806 ms., avg: 1.846 ms.
Best time for 80K FFT length: 2.348 ms., avg: 2.397 ms.
Best time for 96K FFT length: 2.954 ms., avg: 3.028 ms.
Best time for 112K FFT length: 3.424 ms., avg: 3.485 ms.
Best time for 128K FFT length: 3.816 ms., avg: 3.862 ms.
Best time for 160K FFT length: 4.743 ms., avg: 4.806 ms.
Best time for 192K FFT length: 5.752 ms., avg: 5.840 ms.
Best time for 224K FFT length: 6.973 ms., avg: 7.142 ms.
Best time for 256K FFT length: 7.784 ms., avg: 8.309 ms.
Best time for 320K FFT length: 9.933 ms., avg: 10.070 ms.
Best time for 384K FFT length: 12.077 ms., avg: 12.646 ms.
Best time for 448K FFT length: 14.542 ms., avg: 15.207 ms.
Best time for 512K FFT length: 16.236 ms., avg: 16.989 ms.
Best time for 640K FFT length: 21.130 ms., avg: 21.378 ms.
Best time for 768K FFT length: 25.505 ms., avg: 27.002 ms.
Best time for 896K FFT length: 30.553 ms., avg: 30.874 ms.
Best time for 1024K FFT length: 34.095 ms., avg: 34.595 ms.
Best time for 1280K FFT length: 44.195 ms., avg: 44.695 ms.
Best time for 1536K FFT length: 53.599 ms., avg: 54.485 ms.
Best time for 1792K FFT length: 64.209 ms., avg: 64.789 ms.
Best time for 2048K FFT length: 71.856 ms., avg: 72.589 ms.
Best time for 2560K FFT length: 95.654 ms., avg: 97.045 ms.
Best time for 3072K FFT length: 116.127 ms., avg: 117.338 ms.
Best time for 3584K FFT length: 148.204 ms., avg: 149.209 ms.
Best time for 4096K FFT length: 162.536 ms., avg: 164.507 ms.
Best time for 5120K FFT length: 276.500 ms., avg: 279.067 ms.
Best time for 6144K FFT length: 278.405 ms., avg: 281.440 ms.
Best time for 7168K FFT length: 391.011 ms., avg: 393.316 ms.
Best time for 8192K FFT length: 488.250 ms., avg: 498.163 ms.
Timing FFTs using 2 threads on 1 physical CPUs.
Best time for 64K FFT length: 1.766 ms., avg: 1.791 ms.
Best time for 80K FFT length: 2.188 ms., avg: 2.259 ms.
Best time for 96K FFT length: 2.635 ms., avg: 2.713 ms.
Best time for 112K FFT length: 3.242 ms., avg: 3.372 ms.
Best time for 128K FFT length: 3.466 ms., avg: 3.531 ms.
Best time for 160K FFT length: 4.646 ms., avg: 4.744 ms.
Best time for 192K FFT length: 5.570 ms., avg: 5.699 ms.
Best time for 224K FFT length: 6.739 ms., avg: 6.850 ms.
Best time for 256K FFT length: 7.546 ms., avg: 7.645 ms.
Best time for 320K FFT length: 9.517 ms., avg: 10.094 ms.
Best time for 384K FFT length: 11.526 ms., avg: 12.217 ms.
Best time for 448K FFT length: 14.316 ms., avg: 14.724 ms.
Best time for 512K FFT length: 16.176 ms., avg: 17.562 ms.
Best time for 640K FFT length: 19.389 ms., avg: 20.192 ms.
Best time for 768K FFT length: 24.424 ms., avg: 26.029 ms.
Best time for 896K FFT length: 30.149 ms., avg: 30.902 ms.
Best time for 1024K FFT length: 34.038 ms., avg: 35.071 ms.
Best time for 1280K FFT length: 43.128 ms., avg: 44.387 ms.
Best time for 1536K FFT length: 53.148 ms., avg: 53.933 ms.
Best time for 1792K FFT length: 65.590 ms., avg: 66.984 ms.
Best time for 2048K FFT length: 77.040 ms., avg: 79.179 ms.
Best time for 2560K FFT length: 100.860 ms., avg: 103.859 ms.
Best time for 3072K FFT length: 126.063 ms., avg: 126.922 ms.
Best time for 3584K FFT length: 202.221 ms., avg: 205.251 ms.
Best time for 4096K FFT length: 205.710 ms., avg: 210.142 ms.
Best time for 5120K FFT length: 292.021 ms., avg: 295.893 ms.
Best time for 6144K FFT length: 334.573 ms., avg: 336.751 ms.
Best time for 7168K FFT length: 424.872 ms., avg: 436.276 ms.
Best time for 8192K FFT length: 660.347 ms., avg: 667.462 ms.
Best time for 61 bit trial factors: 11.063 ms.
Best time for 62 bit trial factors: 11.097 ms.
Best time for 63 bit trial factors: 12.199 ms.
Best time for 64 bit trial factors: 12.201 ms.
Best time for 65 bit trial factors: 14.952 ms.
Best time for 66 bit trial factors: 14.935 ms.
Best time for 67 bit trial factors: 14.962 ms.
Best time for 75 bit trial factors: 15.281 ms.
Best time for 76 bit trial factors: 15.225 ms.
Best time for 77 bit trial factors: 15.241 ms.

[Apr 27 23:22] Worker starting
[Apr 27 23:22] Setting affinity to run worker on logical CPUs 1,2
[Apr 27 23:22] Optimal P-1 factoring of M12006409 using up to 1280MB of memory.
[Apr 27 23:22] Assuming no factors below 2^64 and 3 primality tests saved if a factor is found.
[Apr 27 23:22] Optimal bounds are B1=230000, B2=5635000
[Apr 27 23:22] Chance of finding a factor is an estimated 6.08%
[Apr 27 23:22] Using Pentium4 FFT length 640K, Pass1=320, Pass2=2K
[Apr 27 23:22] M12006409 stage 1 is 71.50% complete.
[Apr 27 23:26] M12006409 stage 1 is 74.51% complete. Time: 225.557 sec.
[Apr 27 23:30] M12006409 stage 1 is 77.52% complete. Time: 224.732 sec.
[Apr 27 23:34] M12006409 stage 1 is 80.53% complete. Time: 224.146 sec.
[Apr 27 23:38] M12006409 stage 1 is 83.54% complete. Time: 224.090 sec.
[Apr 27 23:41] M12006409 stage 1 is 86.55% complete. Time: 224.198 sec.
[Apr 27 23:45] M12006409 stage 1 is 89.57% complete. Time: 224.215 sec.
[Apr 27 23:49] M12006409 stage 1 is 92.58% complete. Time: 224.214 sec.
[Apr 27 23:53] M12006409 stage 1 is 95.59% complete. Time: 224.714 sec.
[Apr 27 23:56] M12006409 stage 1 is 98.60% complete. Time: 224.121 sec.
[Apr 27 23:58] M12006409 stage 1 complete. 189276 transforms. Time: 2124.190 sec.
[Apr 27 23:58] Starting stage 1 GCD - please be patient.
[Apr 27 23:59] Stage 1 GCD complete. Time: 40.067 sec.
[Apr 27 23:59] Using 1276MB of memory.  Processing 242 relative primes (0 of 480 already processed).
[Apr 28 00:04] M12006409 stage 2 is 1.71% complete. Time: 298.620 sec.
[Apr 28 00:09] M12006409 stage 2 is 4.88% complete. Time: 293.318 sec.
[Apr 28 00:14] M12006409 stage 2 is 8.04% complete. Time: 296.774 sec.
[Apr 28 00:18] M12006409 stage 2 is 11.20% complete. Time: 294.098 sec.
[Apr 28 00:23] M12006409 stage 2 is 14.35% complete. Time: 294.173 sec.
[Apr 28 00:28] M12006409 stage 2 is 17.51% complete. Time: 292.268 sec.
[Apr 28 00:29] Worker stopped.

[Apr 29 15:36] Worker starting
[Apr 29 15:36] Setting affinity to run worker on logical CPUs 1,2
[Apr 29 15:36] Optimal P-1 factoring of M54933283 using up to 1280MB of memory.
[Apr 29 15:36] Assuming no factors below 2^71 and 2 primality tests saved if a factor is found.
[Apr 29 15:36] Optimal bounds are B1=580000, B2=12470000
[Apr 29 15:36] Chance of finding a factor is an estimated 4.69%
[Apr 29 15:36] Using Pentium4 FFT length 2880K, Pass1=640, Pass2=4608
[Apr 29 15:36] M54933283 stage 1 is 88.07% complete.
[Apr 29 15:56] M54933283 stage 1 is 89.26% complete. Time: 1208.631 sec.
[Apr 29 16:17] M54933283 stage 1 is 90.46% complete. Time: 1205.713 sec.
[Apr 29 16:37] M54933283 stage 1 is 91.65% complete. Time: 1207.840 sec.
[Apr 29 16:57] M54933283 stage 1 is 92.85% complete. Time: 1208.129 sec.
[Apr 29 17:17] M54933283 stage 1 is 94.04% complete. Time: 1205.610 sec.
[Apr 29 17:37] M54933283 stage 1 is 95.24% complete. Time: 1209.711 sec.
[Apr 29 17:57] M54933283 stage 1 is 96.43% complete. Time: 1207.936 sec.
[Apr 29 18:17] M54933283 stage 1 is 97.63% complete. Time: 1204.945 sec.
[Apr 29 18:37] M54933283 stage 1 is 98.82% complete. Time: 1208.444 sec.
[Apr 29 18:57] M54933283 stage 1 complete. 199598 transforms. Time: 12051.143 sec.
[Apr 29 18:57] Starting stage 1 GCD - please be patient.
[Apr 29 19:02] Stage 1 GCD complete. Time: 297.959 sec.
[Apr 29 19:02] Using 1268MB of memory.  Processing 50 relative primes (0 of 480 already processed).
[Apr 29 19:28] M54933283 stage 2 is 1.40% complete. Time: 1549.589 sec.
[Apr 29 19:54] M54933283 stage 2 is 2.85% complete. Time: 1545.956 sec.
[Apr 29 20:19] M54933283 stage 2 is 4.30% complete. Time: 1541.765 sec.
[Apr 29 20:45] M54933283 stage 2 is 5.74% complete. Time: 1549.269 sec.
[Apr 29 21:11] M54933283 stage 2 is 7.19% complete. Time: 1542.950 sec.
[Apr 29 21:37] M54933283 stage 2 is 8.64% complete. Time: 1544.795 sec.
[Apr 29 22:02] M54933283 stage 2 is 10.09% complete. Time: 1542.648 sec.
[Apr 29 22:08] Using 1268MB of memory.  Processing 50 relative primes (50 of 480 already processed).
[Apr 29 22:28] M54933283 stage 2 is 11.49% complete. Time: 1541.718 sec.
[Apr 29 22:39] Worker stopped.
harlee is offline   Reply With Quote
Old 2012-05-02, 01:16   #69
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

1D4416 Posts
Default

Quote:
Originally Posted by BrianOC View Post
Small FTT error always the same time/place.
27.6 build 4 is ready. This should fix the small FFT torture test bug. This bug only affected testing Mersenne numbers using less than 10.6 bits per FFT element (carries could propagate into more than 4 words). Normal testing would never run into this bug.
Prime95 is online now   Reply With Quote
Old 2012-05-02, 01:38   #70
BrianOC
 
Apr 2012

178 Posts
Default

Quote:
Originally Posted by Prime95 View Post
27.6 build 4 is ready. This should fix the small FFT torture test bug. This bug only affected testing Mersenne numbers using less than 10.6 bits per FFT element (carries could propagate into more than 4 words). Normal testing would never run into this bug.
Thank you! I test later.
BrianOC is offline   Reply With Quote
Old 2012-05-02, 07:06   #71
Dubslow
Basketry That Evening!
 
Dubslow's Avatar
 
"Bunslow the Bold"
Jun 2011
40<A<43 -89<O<-88

160658 Posts
Default Still got a seg fault

Running the build four torture test, I still got a seg fault. I noticed it was the same FFT length as the one I got previously, so I looked, and it was in fact also the same exponent.
Code:
[Worker #1 May 2 01:51:30] Test 24, 160000 Lucas-Lehmer iterations of M655359 using AVX FFT length 50K, Pass1=640, Pass2=80.
[Worker #2 May 2 01:51:50] Test 28, 160000 Lucas-Lehmer iterations of M622593 using AVX FFT length 50K, Pass1=640, Pass2=80.
Segmentation fault
I'll test the expo manually and see what happens.
Edit: That's not a prime exponent; I'll see if there's some way to fake the program.
Edit2: No quick way off the bat. I might recompile a version without the security code, but it'd be a pain to remove those function calls... probably not worth it. Is there anything else I can do George?
Ubuntu 11.04, 64 bit everything.
Edit3: Actually, the makefile appears to make it easy (pardon the pun). Could someone explain that please?
Code:
security:
	[ ! -e ../security.h ] && touch ../security.h || true
	[ ! -e ../security.c ] && touch ../security.c || true
	[ ! -e ../secure5.c ] && touch ../secure5.c || true
I read that as "[(not-exist-files) and (create them)] or true", but what does that actually do? Does it always attempt to make the end result true, and even so, how would that actually compile anything?

Last fiddled with by Dubslow on 2012-05-02 at 07:21
Dubslow is offline   Reply With Quote
Old 2012-05-02, 10:35   #72
TheJudger
 
TheJudger's Avatar
 
"Oliver"
Mar 2005
Germany

11×101 Posts
Default

Quote:
Originally Posted by Dubslow View Post
Code:
security:
	[ ! -e ../security.h ] && touch ../security.h || true
	[ ! -e ../security.c ] && touch ../security.c || true
	[ ! -e ../secure5.c ] && touch ../secure5.c || true
I read that as "[(not-exist-files) and (create them)] or true", but what does that actually do? Does it always attempt to make the end result true, and even so, how would that actually compile anything?
If the condition (e.g. "! -e ../security.h") is true (file does not exist) then the part after the "&&" is executed (empty file will be created). If the condition is not true than the command after the "||" is executed.

Oliver
TheJudger is offline   Reply With Quote
Old 2012-05-02, 15:42   #73
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

22×1,873 Posts
Default

Quote:
Originally Posted by Dubslow View Post
Running the build four torture test, I still got a seg fault.
Ugh. The last fix was not sufficient.
Prime95 is online now   Reply With Quote
Old 2012-05-02, 17:32   #74
diamonddave
 
diamonddave's Avatar
 
Feb 2004

25×5 Posts
Default underestimating of P-1 time

I'm using 27.6 to do P-1 on 3 different PC.

I've switched to the new version last week. Although it is faster then the previous version, it seams like it is underestimating the completion time of the exponent.

As an example on my i5-2500k I have completed 33 exponent in the last 6 days for about 5.5 per day

If I do a Manual Communication it looks like I will be done with my assignment in about 18 days

The problem is that I have 275 exponent reserved on the PC. I would expect to finish them in about 50 days at 5.5 per days.

It's like it's not taking into account the stage 2 times.

Here is my local.txt content if it helps:

Code:
OldCpuSpeed=3359
NewCpuSpeedCount=0
NewCpuSpeed=0
MaxHighMemWorkers=2
Memory=3000 during 8:30-19:00 else 4000
ThreadsPerTest=1
LastEndDatesSent=1335979556
RollingHash=-1126629492
RollingStartTime=1335946302
RollingCompleteTime=59221
RollingAverage=1621
RollingAverageIsFromV27=1
diamonddave is offline   Reply With Quote
Old 2012-05-02, 21:49   #75
Dubslow
Basketry That Evening!
 
Dubslow's Avatar
 
"Bunslow the Bold"
Jun 2011
40<A<43 -89<O<-88

3·29·83 Posts
Default Weeee compiling is never easy (or so it seems)

Okay, I worked out a few issues, but this one has me stumped:
Code:
/usr/bin/ld: cannot find -lgssapi_krb5
So I tried searching the repos for a fix. The first three were already installed, and I installed the last:
Code:
bill@Gravemind:~∰∂ sudo apt-get install libkrb5support0 libkrb5-dev libgssapi-krb5-2 libglobus-gss-assist-dev
Reading package lists... Done
Building dependency tree       
Reading state information... Done
libglobus-gss-assist-dev is already the newest version.
libgssapi-krb5-2 is already the newest version.
libkrb5-dev is already the newest version.
libkrb5support0 is already the newest version.
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
I'm really not sure how to proceed.

Also, if I were to succeed in building my own "untrused" version, would PrimeNet still accept/credit work like usual, and, more importantly, will it clear DC expos?

Last fiddled with by Dubslow on 2012-05-02 at 21:49 Reason: (or so it seems)
Dubslow is offline   Reply With Quote
Old 2012-05-03, 04:59   #76
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

22×1,873 Posts
Default

Quote:
Originally Posted by Prime95 View Post
Ugh. The last fix was not sufficient.
I've coded a fix for version 27.7. I'm running some tests now.

Why bump to version 27.7? I upgrade the minor version whenever there are changes that affect all users of the gwnum library.

In very rare cases, the bug could affect any test. How rare? In version 27.1 to 27.4, carries were propagated to 8 FFT words. In 27.6 carries were propagated to just 4 FFT words. Assuming there are 18 bits per FFT word, then a 53-bit floating point result will need to propagate carries into 2 words. But, if that second word overflows and the third word is 2^18-1 and the fourth word is 2^18-1, then 27.6 would try to propagate into the fifth word and the bug is triggered. Thus, the chance of encountering the bug was about 1 in 2^36 (1 in 2^108 in earlier v27 versions). So, don't lose any sleep over bad LL results.
Prime95 is online now   Reply With Quote
Old 2012-05-03, 07:16   #77
ckdo
 
ckdo's Avatar
 
Dec 2007
Cleves, Germany

2×5×53 Posts
Default

Quote:
Originally Posted by Prime95 View Post
Thus, the chance of encountering the bug was about 1 in 2^36 (1 in 2^108 in earlier v27 versions). So, don't lose any sleep over bad LL results.
2^36 ... what? Iterations? That's only 1000-something LL tests, or an error rate near 0.1%. (I'm still not losing more sleep than usual, though).
ckdo is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Prime95 version 27.3 Prime95 Software 148 2012-03-18 19:24
Prime95 version 26.3 Prime95 Software 76 2010-12-11 00:11
Prime95 version 25.5 Prime95 PrimeNet 369 2008-02-26 05:21
Prime95 version 25.4 Prime95 PrimeNet 143 2007-09-24 21:01
When the next prime95 version ? pacionet Software 74 2006-12-07 20:30

All times are UTC. The time now is 01:05.

Sat May 15 01:05:38 UTC 2021 up 36 days, 19:46, 0 users, load averages: 2.16, 2.08, 2.59

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.