Problems running ECM on 2 workers

MatWur-S530113 · 2009-01-06, 12:02

Hello,

I'm running 2 times the same batch with ECM curves on both workers like:

Code:

 
[Worker #1]
ECM2=1,2,2213,-1,44000000,4400000000,30,"53113,7967736971962111"
.
.
.
[Worker #2]
ECM2=1,2,2213,-1,44000000,4400000000,30,"53113,7967736971962111"
.
.
.

Yesterday I interrupted prime95 while running this curves to add more batches of work for the night. Worker#1 was at curve 28 when I stopped it, Worker#2 was at curve 29 (~5%). After restarting prime95 both workers began with curve 29 at 5% and the same sigma. I think both workers stored their data in the same file on the HD, overwriting the data of the first worker.
With prime95 v24.14 (atm I'm using v25.8b5) I never had this problem as it used 2 instances and 2 different folders on the HD. Curve 30 had different sigmas again for the both workers. Then prime95 reportet 2 times '30 curves done' to the server. Indeed curve 28 of Worker#1 was not calculatet and curve 29 was calculatet twice. This could be used for another kind of the 'cheesehead-bug': Interrupt Worker#1 at curve 1 and Worker#2 at curve 30. Restart both workers at curve 30 and report 2 times 30 curves done...
I don't know how to solve this problem. I will try to run ECM on different exponents, but it is a strength of ECM to be parallelizeable and sometimes it is needed.

best regards,

Matthias

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Workers	galgrnpa	Software	5	2016-08-09 14:54
Heat problems and running less workers than cores	MacMagnus	Information & Answers	10	2013-07-14 21:29
How many workers?	Unregistered	Information & Answers	12	2013-01-22 12:09
Went from 8 workers to 4 workers on v26.6 upgrade	dmoran	Software	13	2011-05-23 12:36
Problems running Prime95 on a dual processor	WileECoyote	Hardware	11	2006-10-21 07:05

2009-01-06, 12:02	#1
MatWur-S530113 Apr 2007 Spessart/Germany 2×83 Posts	Problems running ECM on 2 workers Hello, I'm running 2 times the same batch with ECM curves on both workers like: Code: [Worker #1] ECM2=1,2,2213,-1,44000000,4400000000,30,"53113,7967736971962111" . . . [Worker #2] ECM2=1,2,2213,-1,44000000,4400000000,30,"53113,7967736971962111" . . . Yesterday I interrupted prime95 while running this curves to add more batches of work for the night. Worker#1 was at curve 28 when I stopped it, Worker#2 was at curve 29 (~5%). After restarting prime95 both workers began with curve 29 at 5% and the same sigma. I think both workers stored their data in the same file on the HD, overwriting the data of the first worker. With prime95 v24.14 (atm I'm using v25.8b5) I never had this problem as it used 2 instances and 2 different folders on the HD. Curve 30 had different sigmas again for the both workers. Then prime95 reportet 2 times '30 curves done' to the server. Indeed curve 28 of Worker#1 was not calculatet and curve 29 was calculatet twice. This could be used for another kind of the 'cheesehead-bug': Interrupt Worker#1 at curve 1 and Worker#2 at curve 30. Restart both workers at curve 30 and report 2 times 30 curves done... I don't know how to solve this problem. I will try to run ECM on different exponents, but it is a strength of ECM to be parallelizeable and sometimes it is needed. best regards, Matthias