Posts by James

1) Message boards : Number crunching : Is turning on hyper threading better if you still assign number of tasks to only pysical cores? (Message 147189)
Posted 1002 days ago by

James

These things vary from system to system, from OS to OS, and from subproject to subprojects; only way to know for sure is to try it out for yourself.

That being said, there typically isn't that great of a difference IIRC.

2) Message boards : Generalized Fermat Prime Search : Genefer 3.3.5 testing (Message 146101)
Posted 1029 days ago by

James

linux64 SSE2i

I think the times are wrong for 3638450^524288+1 (maybe because it was unexpectedly terminated?)
The result seems fine.

3638450^524288+1

daohe@daohe-XPS-15-9570:~/Documents/genefer335test/s3$ ../gfn3.3.5 -q "3638450^524288+1" -x sse2i genefer 3.3.5.1 (CPU/Linux/64-bit/gcc-7.5.0/BOINC-7.17.0) Copyright 2001-2020, Yves Gallot Copyright 2009, Mark Rodenkirch, David Underbakke Copyright 2010-2012, Shoichiro Yamada, Ken Brazier Copyright 2011-2014, Michael Goetz, Ronald Schneider Copyright 2011-2018, Iain Bethune Genefer is free source code, under the MIT license. Supported transform implementations: fma avx sse4 sse2 fmai avxi sse4i sse2i Command line: ../gfn3.3.5 -q 3638450^524288+1 -x sse2i Low priority change succeeded. Testing 3638450^524288+1... Using sse2i transform Resuming 3638450^524288+1 from a checkpoint (1840338 iterations left) Estimated time remaining for 3638450^524288+1 is 5:02:00 3638450^524288+1 is a probable prime. (3439810 digits) (err = 0.0039) (time = 28:36:32) 20:46:04

11081688^262144+1

daohe@daohe-XPS-15-9570:~/Documents/genefer335test/s2$ ../gfn3.3.5 -q "11081688^262144+1" -x sse2i genefer 3.3.5.1 (CPU/Linux/64-bit/gcc-7.5.0/BOINC-7.17.0) Copyright 2001-2020, Yves Gallot Copyright 2009, Mark Rodenkirch, David Underbakke Copyright 2010-2012, Shoichiro Yamada, Ken Brazier Copyright 2011-2014, Michael Goetz, Ronald Schneider Copyright 2011-2018, Iain Bethune Genefer is free source code, under the MIT license. Supported transform implementations: fma avx sse4 sse2 fmai avxi sse4i sse2i Command line: ../gfn3.3.5 -q 11081688^262144+1 -x sse2i Low priority change succeeded. Testing 11081688^262144+1... Using sse2i transform Starting initialization... Initialization complete (0.565 seconds). Estimated time for 11081688^262144+1 is 11:20:00 11081688^262144+1 is a probable prime. (1846702 digits) (err = 0.0156) (time = 12:47:47) 04:56:53

80146408^131072+1

daohe@daohe-XPS-15-9570:~/Documents/genefer335test/s1$ ../gfn3.3.5 -q "80146408^131072+1" -x sse2i genefer 3.3.5.1 (CPU/Linux/64-bit/gcc-7.5.0/BOINC-7.17.0) Copyright 2001-2020, Yves Gallot Copyright 2009, Mark Rodenkirch, David Underbakke Copyright 2010-2012, Shoichiro Yamada, Ken Brazier Copyright 2011-2014, Michael Goetz, Ronald Schneider Copyright 2011-2018, Iain Bethune Genefer is free source code, under the MIT license. Supported transform implementations: fma avx sse4 sse2 fmai avxi sse4i sse2i Command line: ../gfn3.3.5 -q 80146408^131072+1 -x sse2i Low priority change succeeded. Testing 80146408^131072+1... Using sse2i transform Starting initialization... Initialization complete (0.211 seconds). Estimated time for 80146408^131072+1 is 2:49:00 80146408^131072+1 is a probable prime. (1035978 digits) (err = 0.1875) (time = 3:50:20) 19:58:59

98752904^65536+1

daohe@daohe-XPS-15-9570:~/Documents/genefer335test/s0$ ../gfn3.3.5 -q "98752904^65536+1" -x sse2i genefer 3.3.5.1 (CPU/Linux/64-bit/gcc-7.5.0/BOINC-7.17.0) Copyright 2001-2020, Yves Gallot Copyright 2009, Mark Rodenkirch, David Underbakke Copyright 2010-2012, Shoichiro Yamada, Ken Brazier Copyright 2011-2014, Michael Goetz, Ronald Schneider Copyright 2011-2018, Iain Bethune Genefer is free source code, under the MIT license. Supported transform implementations: fma avx sse4 sse2 fmai avxi sse4i sse2i Command line: ../gfn3.3.5 -q 98752904^65536+1 -x sse2i Low priority change succeeded. Testing 98752904^65536+1... Using sse2i transform Starting initialization... Initialization complete (0.057 seconds). Estimated time for 98752904^65536+1 is 0:37:20 98752904^65536+1 is a probable prime. (523931 digits) (err = 0.1875) (time = 0:57:28) 17:05:26

3) Message boards : Number crunching : Any way to force NOT reporting results immediately? (Message 146033)
Posted 1030 days ago by

James

Although I'm not certain this will work, could you try putting in your app_config.xml

<report_results_immediately>0</report_results_immediately>

instead of

<report_results_immediately/>

4) Message boards : Generalized Fermat Prime Search : Genefer 3.3.5 testing (Message 146014)
Posted 1031 days ago by

James

sse2i tests will be done soon

I take this back. Something in the kitchen tripped the RCD and the machine I was running the tests on was not on an UPS.

EDIT: ah genefer checkpoints. The tests will be done soon :)

5) Message boards : Generalized Fermat Prime Search : Genefer 3.3.5 testing (Message 146004)
Posted 1031 days ago by

James

linux64 sse4i. These took longer than I expected; sse2i tests will be done soon

root@thisisapi:~/genefer-3_3_5-test# ./gfn3.3.5 -q "98752904^65536+1" -x sse4i genefer 3.3.5.1 (CPU/Linux/64-bit/gcc-7.5.0/BOINC-7.17.0) Copyright 2001-2020, Yves Gallot Copyright 2009, Mark Rodenkirch, David Underbakke Copyright 2010-2012, Shoichiro Yamada, Ken Brazier Copyright 2011-2014, Michael Goetz, Ronald Schneider Copyright 2011-2018, Iain Bethune Genefer is free source code, under the MIT license. Supported transform implementations: avx sse4 sse2 avxi sse4i sse2i Command line: ./gfn3.3.5 -q 98752904^65536+1 -x sse4i Low priority change succeeded. Testing 98752904^65536+1... Using sse4i transform Starting initialization... Initialization complete (0.144 seconds). Estimated time for 98752904^65536+1 is 0:47:30 98752904^65536+1 is a probable prime. (523931 digits) (err = 0.1875) (time = 0:46:21) 12:24:06 root@thisisapi:~/genefer-3_3_5-test# ./gfn3.3.5 tests -x sse4i genefer 3.3.5.1 (CPU/Linux/64-bit/gcc-7.5.0/BOINC-7.17.0) Copyright 2001-2020, Yves Gallot Copyright 2009, Mark Rodenkirch, David Underbakke Copyright 2010-2012, Shoichiro Yamada, Ken Brazier Copyright 2011-2014, Michael Goetz, Ronald Schneider Copyright 2011-2018, Iain Bethune Genefer is free source code, under the MIT license. Supported transform implementations: avx sse4 sse2 avxi sse4i sse2i Command line: ./gfn3.3.5 tests -x sse4i Low priority change succeeded. Start test of file 'tests' - 14:09:05 Testing 80146408^131072+1... Using sse4i transform Starting initialization... Initialization complete (0.319 seconds). Estimated time for 80146408^131072+1 is 3:15:00 80146408^131072+1 is a probable prime. (1035978 digits) (err = 0.1875) (time = 3:17:10) 17:26:15 Testing 11081688^262144+1... Using sse4i transform Starting initialization... Initialization complete (0.790 seconds). Estimated time for 11081688^262144+1 is 12:00:00 11081688^262144+1 is a probable prime. (1846702 digits) (err = 0.0156) (time = 12:16:19) 05:42:34 Testing 3638450^524288+1... Using sse4i transform Starting initialization... Initialization complete (2.538 seconds). Estimated time for 3638450^524288+1 is 47:10:00 3638450^524288+1 is a probable prime. (3439810 digits) (err = 0.0039) (time = 53:54:45) 11:37:19 Test of file 'tests' completed - 11:37:19

6) Message boards : Generalized Fermat Prime Search : Genefer 3.3.5 testing (Message 145800)
Posted 1034 days ago by

James

It seems no one is running linux64 sse2i and sse4i. I'll take those.

Eventually, when the above is done, I'll contemplate genefer_linux32...

7) Message boards : Number crunching : LLR2 installed on all big LLR projects (Message 145281)
Posted 1047 days ago by

James

Thanks for this speeding up for the double check. First time I did not trust it. Where can I see how does it work?

If you're on discord: https://canary.discord.com/channels/357493752434130944/357550381431259136/745508916561051678

Otherwise, Pavel has infographics that resumes how the fast double-check works:
https://raw.githubusercontent.com/patnashev/llr2/master/demo/ProofInfo.png
https://raw.githubusercontent.com/patnashev/llr2/master/demo/ProofInfoP.png
https://raw.githubusercontent.com/patnashev/llr2/master/demo/ProofMagic.png

I do suggest you read the explanation on discord. Pavel wrote quite a bit of explanation there.

8) Message boards : Problems and Help : Genefer Tasks All Coming Back Invalid (Message 143805)
Posted 1093 days ago by

James

Apologies for the delayed response, can you try adding <ignore_intel_dev> to your cc_config.xml?

<cc_config> <options> <ignore_intel_dev>0</ignore_intel_dev> </options> </cc_config>

This should disable your Intel iGPU for BOINC

9) Message boards : Number crunching : Rtx 30 series (Message 143129)
Posted 1117 days ago by

James

They claim 2x FP32 performance among other things, which sounds really good for now...

I was looking at the lineup and was hesitant on a 3070 given that it runs on GDDR6, as opposed to the 3080 and 3090 running on GDDR6X. If I remember correctly, most genefer tasks are memory bandwidth limited on Turing already, and Turing runs on GDDR6.

10) Message boards : AP26 - AP27 Search : AP27 Update Testing (Message 141706)
Posted 1171 days ago by

James

Ran my reservations, everything looks good.

Linux avx singlethreaded