ECM users - version 30.9-30.12 (see post#235) - Page 17

kruoli · 2023-02-13, 07:45

Killed usually means OOM. How much memory have you allocated? Which exponent was run?

bur · 2023-02-13, 18:47

Now the whole server apparently crashed. At least I couldn't connect via SSH anymore and the KVM switch showed:

Code:

root login: [xxx.xxx] Out of memory: Killed process 5451 (mprime) total-vm:41141968kB, anon-rss:32094432kB, file-rss:0kB, shmem-rss:0kB, UID:xxx pgtables:79972kB oom_score_adj:0

Sending Ctrl-Alt-Del gracefully restarted.

Allocated memory: Memory=28672 during 7:30-23:30 else 28672, I ran B1=3M ECM curves on exponents around 20xxx with 12 workers.

bur · 2023-02-14, 14:07

Killed again (just mprime, not the server). Here's a bit more output from before being killed:

Code:

[Worker #2 Feb 14 13:24] Stage 1 complete. 77076108 transforms, 1 modular invers
es. Total time: 80.841 sec.
[Worker #2 Feb 14 13:24] Available memory is 23718MB.
[Worker #2 Feb 14 13:24] Switching to FMA3 FFT length 1280
[Worker #2 Feb 14 13:24] Optimal B2 is 2267*B1 = 6801000000.  Actual B2 will be 6803009850.
[Worker #2 Feb 14 13:24] Estimated stage 2 vs. stage 1 runtime ratio: 0.260
[Worker #2 Feb 14 13:24] Using 824MB of memory.  D: 51870, degree-5184 polynomials.  Ftree polys in memory: 12
[Worker #11 Feb 14 13:24] PolyH built.  Time: 0.370 sec.
[Worker #8 Feb 14 13:24] PolyH built.  Time: 0.369 sec.
[Worker #10 Feb 14 13:24] PolyG built.  Time: 0.776 sec.
[Worker #9 Feb 14 13:24] PolyH built.  Time: 0.366 sec.
[Worker #8 Feb 14 13:24] H(X) scaled.  Time: 0.102 sec.
[Worker #7 Feb 14 13:24] PolyH built.  Time: 0.362 sec.
[Worker #2 Feb 14 13:24] nQx complete.  Time: 0.199 sec.
[Worker #8 Feb 14 13:24] PolyF up.  Time: 0.017 sec.
[Worker #1 Feb 14 13:24] PolyG built.  Time: 0.756 sec.
[Worker #8 Feb 14 13:24] PolyF down.  Time: 0.267 sec.
[Worker #8 Feb 14 13:24] PolyF up.  Time: 0.037 sec.
[Worker #10 Feb 14 13:24] PolyH built.  Time: 0.384 sec.
[Worker #8 Feb 14 13:24] PolyF down.  Time: 0.744 sec.
[Worker #1 Feb 14 13:25] PolyH built.  Time: 22.275 sec.
Killed

When occasionally checking htop I never see more than 14-15 GB in use. I also never had this problem before when doing ECM on that range.

Mark Rose · 2023-02-14, 16:46

Quote:

Originally Posted by bur

Killed again (just mprime, not the server). Here's a bit more output from before being killed:

Code:

[Worker #2 Feb 14 13:24] Stage 1 complete. 77076108 transforms, 1 modular invers
es. Total time: 80.841 sec.
[Worker #2 Feb 14 13:24] Available memory is 23718MB.
[Worker #2 Feb 14 13:24] Switching to FMA3 FFT length 1280
[Worker #2 Feb 14 13:24] Optimal B2 is 2267*B1 = 6801000000.  Actual B2 will be 6803009850.
[Worker #2 Feb 14 13:24] Estimated stage 2 vs. stage 1 runtime ratio: 0.260
[Worker #2 Feb 14 13:24] Using 824MB of memory.  D: 51870, degree-5184 polynomials.  Ftree polys in memory: 12
[Worker #11 Feb 14 13:24] PolyH built.  Time: 0.370 sec.
[Worker #8 Feb 14 13:24] PolyH built.  Time: 0.369 sec.
[Worker #10 Feb 14 13:24] PolyG built.  Time: 0.776 sec.
[Worker #9 Feb 14 13:24] PolyH built.  Time: 0.366 sec.
[Worker #8 Feb 14 13:24] H(X) scaled.  Time: 0.102 sec.
[Worker #7 Feb 14 13:24] PolyH built.  Time: 0.362 sec.
[Worker #2 Feb 14 13:24] nQx complete.  Time: 0.199 sec.
[Worker #8 Feb 14 13:24] PolyF up.  Time: 0.017 sec.
[Worker #1 Feb 14 13:24] PolyG built.  Time: 0.756 sec.
[Worker #8 Feb 14 13:24] PolyF down.  Time: 0.267 sec.
[Worker #8 Feb 14 13:24] PolyF up.  Time: 0.037 sec.
[Worker #10 Feb 14 13:24] PolyH built.  Time: 0.384 sec.
[Worker #8 Feb 14 13:24] PolyF down.  Time: 0.744 sec.
[Worker #1 Feb 14 13:25] PolyH built.  Time: 22.275 sec.
Killed

When occasionally checking htop I never see more than 14-15 GB in use. I also never had this problem before when doing ECM on that range.

Check /var/log/dmesg and you should see a message showing how much memory it was using, like this:

Feb 13 15:44:14 foobox kernel: [459648.620241] Out of memory: Killed process 472141 (mprime) total-vm:24459956kB, anon-rss:23648420kB, file-rss:0kB, shmem-rss:0kB, UID:1000 pgtables:46532kB oom_score_adj:0

You can see in this case mprime was using 24ish GB at the time it was killed.

bur · 2023-02-15, 07:39

There's nothing like that in /var/log/dmesg, but I got this message from the crashed server (see above)
root login: [xxx.xxx] Out of memory: Killed process 5451 (mprime) total-vm:41141968kB, anon-rss:32094432kB, file-rss:0kB, shmem-rss:0kB, UID:xxx pgtables:79972kB oom_score_adj:0

So it was using 41 GB? That's beyond the limit I set in local.txt.

bur · 2023-02-17, 06:57

There is some problem with RAM usage. After about 12 hours of ECM it steadily increases until it reaches 30 GB and more and eventually mprime crashes. Individual workers still state "28000 MB free" and report using 1000 MB or so.

When I just manually quit mprime and restart it, the RAM usage is down to 4-5 GB from 29 GB. Peak usage in normal operation is around 14-15 GB which agrees well with 12 workers and each reporting a usage around 1 GB.

Is there something like a debug log I could post?

kruoli · 2023-02-17, 18:58

In the latest 30.10 build, I had a strange behaviour on start of ECM stage 2:

Code:

[Main thread Feb 17 19:55] Optimizing for CPU architecture: AMD Zen, L2 cache size: 8x512 KB, L3 cache size: 16 MB
[Main thread Feb 17 19:55] Starting workers.
[Worker #1 Feb 17 19:55] Worker starting
[Worker #3 Feb 17 19:55] Waiting 10 seconds to stagger worker starts.
[Worker #4 Feb 17 19:55] Waiting 15 seconds to stagger worker starts.
[Worker #1 Feb 17 19:55] Setting affinity to run worker on logical CPUs 2 (zero-based)
[Worker #2 Feb 17 19:55] Waiting 5 seconds to stagger worker starts.
[Worker #1 Feb 17 19:55] 
[Worker #1 Feb 17 19:55] Using FMA3 FFT length 3K
[Worker #1 Feb 17 19:55] 0.123 bits-per-word below FFT limit (more than 0.509 allows extra optimizations)
[Worker #1 Feb 17 19:55] ECM on M65563: curve #1 with s=6264378133609367, B1=3000000, B2=100069839870
[Worker #1 Feb 17 19:55] Available memory is 13168MB.
[Worker #1 Feb 17 19:55] Actual B2 will be 18446742843095154688.
[Worker #1 Feb 17 19:55] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Worker #2 Feb 17 19:55] Worker starting
[Worker #2 Feb 17 19:55] Setting affinity to run worker on logical CPUs 6 (zero-based)
[Worker #2 Feb 17 19:55] 
[Worker #2 Feb 17 19:55] Using FMA3 FFT length 4K
[Worker #2 Feb 17 19:55] 2.439 bits-per-word below FFT limit (more than 0.509 allows extra optimizations)
[Worker #2 Feb 17 19:55] ECM on M77557: curve #1 with s=1710131093015744, B1=3000000, B2=100000000000
[Worker #2 Feb 17 19:55] Exceeded limit on number of workers that can use lots of memory.
[Worker #2 Feb 17 19:55] Looking for work that uses less memory.
[Worker #2 Feb 17 19:55] 
[Worker #2 Feb 17 19:55] Using FMA3 FFT length 3K
[Worker #2 Feb 17 19:55] 0.172 bits-per-word below FFT limit (more than 0.509 allows extra optimizations)
[Worker #2 Feb 17 19:55] ECM on M65413: curve #1 with s=1982442750841053, B1=3000000, B2=100000000000
[Worker #2 Feb 17 19:55] M65413 curve 1 stage 1 at prime 731681 [24.38%].
[Worker #1 Feb 17 19:55] Available memory is 13081MB.
[Worker #1 Feb 17 19:55] Actual B2 will be 18446742843095154688.
[Worker #1 Feb 17 19:55] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Worker #3 Feb 17 19:55] Worker starting
[Worker #3 Feb 17 19:55] Setting affinity to run worker on logical CPUs 10 (zero-based)
[Worker #3 Feb 17 19:55] 
[Worker #3 Feb 17 19:55] Using FMA3 FFT length 5K
[Worker #3 Feb 17 19:55] 3.314 bits-per-word below FFT limit (more than 0.509 allows extra optimizations)
[Worker #3 Feb 17 19:55] ECM on M91921: curve #1 with s=7335134578640101, B1=3000000, B2=100000000000
[Worker #3 Feb 17 19:55] M91921 curve 1 stage 1 at prime 2653723 [88.45%].
[Worker #1 Feb 17 19:55] Available memory is 12996MB.
[Worker #1 Feb 17 19:55] Actual B2 will be 18446742843095154688.
[Worker #1 Feb 17 19:55] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Worker #4 Feb 17 19:55] Worker starting
[Worker #4 Feb 17 19:55] Setting affinity to run worker on logical CPUs 14 (zero-based)
[Worker #4 Feb 17 19:55] 
[Worker #4 Feb 17 19:55] Using FMA3 FFT length 3K
[Worker #4 Feb 17 19:55] 0.264 bits-per-word below FFT limit (more than 0.509 allows extra optimizations)
[Worker #4 Feb 17 19:55] ECM on M65129: curve #1 with s=2921619629826331, B1=3000000, B2=100000000000
[Worker #4 Feb 17 19:55] Exceeded limit on number of workers that can use lots of memory.
[Worker #4 Feb 17 19:55] Looking for work that uses less memory.
[Worker #4 Feb 17 19:55] 
[Worker #4 Feb 17 19:55] Using FMA3 FFT length 4K
[Worker #4 Feb 17 19:55] 0.739 bits-per-word below FFT limit (more than 0.509 allows extra optimizations)
[Worker #4 Feb 17 19:55] ECM on M84521: curve #1 with s=6025389633378337, B1=3000000, B2=100000000000
[Worker #1 Feb 17 19:55] Available memory is 12910MB.
[Worker #4 Feb 17 19:55] M84521 curve 1 stage 1 at prime 1059221 [35.30%].
[Worker #1 Feb 17 19:55] Actual B2 will be 18446742843095154688.
[Worker #1 Feb 17 19:55] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Worker #1 Feb 17 19:55] Available memory is 12781MB.
[Worker #1 Feb 17 19:55] Actual B2 will be 18446742843095154688.
[...]
[Worker #1 Feb 17 19:55] Available memory is 1036MB.
[Worker #1 Feb 17 19:55] Actual B2 will be 18446742843095154688.
[Worker #1 Feb 17 19:55] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Worker #1 Feb 17 19:55] Available memory is 1025MB.
[Worker #1 Feb 17 19:55] Actual B2 will be 18446742843095154688.
[Worker #1 Feb 17 19:55] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Worker #1 Feb 17 19:55] Available memory is 1015MB.
[Worker #1 Feb 17 19:55] Actual B2 will be 18446742843095154688.
[Worker #1 Feb 17 19:55] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Worker #1 Feb 17 19:55] Available memory is 1005MB.
[Worker #1 Feb 17 19:55] Actual B2 will be 18446742843095154688.
[Worker #1 Feb 17 19:55] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Worker #1 Feb 17 19:55] Available memory is 995MB.
[Worker #1 Feb 17 19:55] Switching to FMA3 FFT length 4K
[Worker #1 Feb 17 19:55] Actual B2 will be 100261321875.
[Worker #1 Feb 17 19:55] Estimated stage 2 vs. stage 1 runtime ratio: inf
[Worker #1 Feb 17 19:55] Using 948MB of memory.  D: 64350, degree-7200 polynomials.  Ftree polys in memory: 0
[Worker #1 Feb 17 19:55] nQx complete.  Time: 0.953 sec.
[Worker #1 Feb 17 19:55] Setting affinity to run polymult helper thread on logical CPUs 0 (zero-based)
[Worker #1 Feb 17 19:55] Setting affinity to run polymult helper thread on logical CPUs 1 (zero-based)
[Worker #1 Feb 17 19:55] Setting affinity to run polymult helper thread on logical CPUs 3 (zero-based)
[Worker #1 Feb 17 19:55] Setting affinity to run polymult helper thread on logical CPUs 3 (zero-based)
[Worker #1 Feb 17 19:55] PolyR built.  Time: 0.689 sec.
[Worker #1 Feb 17 19:55] Poly compress.  Time: 0.132 sec.
[Worker #1 Feb 17 19:55] Stage 2 init complete. 431567 transforms, 2 modular inverses. Time: 5.266 sec.
[Worker #1 Feb 17 19:55] PolyG built.  Time: 2.326 sec.
[Worker #1 Feb 17 19:55] M65563 stage 2 at B2=90068281875 [89.83%]
[Worker #1 Feb 17 19:55] PolyG built.  Time: 2.375 sec.
[Worker #1 Feb 17 19:55] PolyH built.  Time: 0.883 sec.
[Worker #1 Feb 17 19:55] PolyG built.  Time: 2.392 sec.
[Worker #1 Feb 17 19:55] PolyH built.  Time: 0.893 sec.

Why was it counting down few MB a time? Is the runtime ratio that off?

James Heinrich · 2023-02-18, 02:46

Quote:

Originally Posted by kruoli

Why was it counting down few MB a time? Is the runtime ratio that off?

I just found Prime95 disappeared on my v30.10 system, so I restarted it and it went away again. Looking closer at the output in the 5 seconds before it disappeared, I see something similar:

Code:

[Feb 17 21:39] Available memory is 512MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 507MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Available memory is 502MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 497MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 492MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Available memory is 487MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 482MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 477MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Available memory is 472MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 468MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 463MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 458MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 454MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 449MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 445MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 440MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 436MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 431MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Available memory is 427MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Available memory is 423MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Available memory is 419MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Available memory is 414MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 410MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 406MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 402MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 398MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Available memory is 394MB.
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Available memory is 390MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Available memory is 386MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 382MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 379MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Available memory is 375MB.
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Available memory is 371MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 367MB.
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 364MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 360MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 356MB.
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Available memory is 353MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 349MB.
[Feb 17 21:39] Actual B2 will be 18446698010340491264.
[Feb 17 21:39] Estimated stage 2 vs. stage 1 runtime ratio: 0.000
[Feb 17 21:39] Setting affinity to run helper thread 1 on CPU core #2
[Feb 17 21:39] Setting affinity to run helper thread 2 on CPU core #3
[Feb 17 21:39] Setting affinity to run helper thread 3 on CPU core #4
[Feb 17 21:39] Available memory is 346MB.

Until it got down to zero then crash.

Rubiksmath · 2023-02-18, 03:03

Looks like it is choosing the unsigned 64 bit integer limit as B2.

bur · 2023-02-19, 17:31

The "memory leak" does not appear in 3.09b3. The 12 workers at max consume 15 GB when all are in Stage 2 simultaneously.

3.09b3 uses B1=3M, B2=11G whereas 3.10b1 uses B1=3M, B2=6.8G. Should I consider the smaller B2 value more efficient and force 3.09 to use it?

kriesel · 2023-02-19, 17:42

30.x (not 3.x which would have been vintage ~1996, and perhaps not even released to others by George; oldest version listed in whatsnew.txt is v8).

2023-02-13, 18:47	#178
bur Aug 2020 796581e-4;32539e-3 2DA₁₆ Posts	Now the whole server apparently crashed. At least I couldn't connect via SSH anymore and the KVM switch showed: Code: root login: [xxx.xxx] Out of memory: Killed process 5451 (mprime) total-vm:41141968kB, anon-rss:32094432kB, file-rss:0kB, shmem-rss:0kB, UID:xxx pgtables:79972kB oom_score_adj:0 Sending Ctrl-Alt-Del gracefully restarted. Allocated memory: Memory=28672 during 7:30-23:30 else 28672, I ran B1=3M ECM curves on exponents around 20xxx with 12 workers. Last fiddled with by bur on 2023-02-13 at 18:49

2023-02-17, 06:57	#182
bur Aug 2020 796581e-4;32539e-3 2×5×73 Posts	There is some problem with RAM usage. After about 12 hours of ECM it steadily increases until it reaches 30 GB and more and eventually mprime crashes. Individual workers still state "28000 MB free" and report using 1000 MB or so. When I just manually quit mprime and restart it, the RAM usage is down to 4-5 GB from 29 GB. Peak usage in normal operation is around 14-15 GB which agrees well with 12 workers and each reporting a usage around 1 GB. Is there something like a debug log I could post? Last fiddled with by bur on 2023-02-17 at 06:58

2023-02-19, 17:42	#187
kriesel "TF79LL86GIMPS96gpu17" Mar 2017 US midwest 7,717 Posts	30.x (not 3.x which would have been vintage ~1996, and perhaps not even released to others by George; oldest version listed in whatsnew.txt is v8). Last fiddled with by kriesel on 2023-02-19 at 17:44

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
That's a Lot of Users!!!	jinydu	Lounge	9	2006-11-10 00:14
Beta version 24.6 - Athlon users wanted	Prime95	Software	139	2005-03-30 12:13
For Old Users	Citrix	Prime Sierpinski Project	15	2004-08-22 16:43
Opportunity! Retaining new users post-M40	GP2	Lounge	55	2003-11-21 21:08
AMD USERS	ET_	Lounge	3	2003-10-11 16:52

2023-02-13, 07:45	#177
kruoli "Oliver" Sep 2017 Porta Westfalica, DE 1,559 Posts	Killed usually means OOM. How much memory have you allocated? Which exponent was run?

2023-02-15, 07:39	#181
bur Aug 2020 796581e-4;32539e-3 2×5×73 Posts	There's nothing like that in /var/log/dmesg, but I got this message from the crashed server (see above) root login: [xxx.xxx] Out of memory: Killed process 5451 (mprime) total-vm:41141968kB, anon-rss:32094432kB, file-rss:0kB, shmem-rss:0kB, UID:xxx pgtables:79972kB oom_score_adj:0 So it was using 41 GB? That's beyond the limit I set in local.txt.

2023-02-18, 03:03	#185
Rubiksmath Sep 2022 2⁴·7 Posts	Looks like it is choosing the unsigned 64 bit integer limit as B2.

2023-02-19, 17:31	#186
bur Aug 2020 796581e-4;32539e-3 2·5·73 Posts	The "memory leak" does not appear in 3.09b3. The 12 workers at max consume 15 GB when all are in Stage 2 simultaneously. 3.09b3 uses B1=3M, B2=11G whereas 3.10b1 uses B1=3M, B2=6.8G. Should I consider the smaller B2 value more efficient and force 3.09 to use it?