ECM users - version 30.9-30.13 (see post#316) - Page 23

Prime95 · 2023-04-19, 22:00

Quote:

Originally Posted by cmpute

I wonder what's the criteria of running a PRP-CF after a found factor? I recently done several PFactor tasks at around 15M, the PRP-CF was not triggered. Is it because of the PFactor task type or the exponent size?

Try setting MaxAutoPRPExponent=n in prime.txt to a larger value. Default is 5000000.

Prime95 · 2023-04-19, 22:05

Quote:

Originally Posted by ryanp

I'm curious to learn more about this. Under what circumstances would this bug manifest? Is it worth doing new builds of LLR + PFGW?

Pavel Atnashev found the bug - roundoff errors during radix conversion. Certainly worth new builds. Gerbicz error checking would catch any issues. However, IIRC, PFGW doesn't use Gerbicz checking.

WraithX · 2023-04-19, 23:52

Quote:

Originally Posted by Prime95

Pavel Atnashev found the bug - roundoff errors during radix conversion. Certainly worth new builds. Gerbicz error checking would catch any issues. However, IIRC, PFGW doesn't use Gerbicz checking.

Do we know which all versions of GWNUM were affected?

Prime95 · 2023-04-20, 00:46

Gwnum libraries dated after 2020-10-03.

cxc · 2023-05-03, 21:46

Hi!

I’m doing ECMF and while 30.8 was fine on F19 ~ F24, it repeatedly aborted on the smaller Fermats with large bounds, B1 ~ 8e8, once it reached stage 2 (i.e. same error on different hardware).

I wasn’t initially aware there was a Mac version of 30.10 – mention of it is somewhat buried on page 10 in post 107, but having found it and set it to work on F12, it works beautifully on stage 2. Thanks so much George and everyone else here helping to squash the bugs.

I have two questions, the latter of which isn’t so important if 30.10b1 works fine: with respect to ECMF, is there anything I should know about 30.10b1 that has been fixed in later betas, that I might come across? And, is there a later build for Mac OS X than 30.10b1?

Prime95 · 2023-05-03, 23:53

Quote:

Originally Posted by cxc

I have two questions, the latter of which isn’t so important if 30.10b1 works fine: with respect to ECMF, is there anything I should know about 30.10b1 that has been fixed in later betas, that I might come across? And, is there a later build for Mac OS X than 30.10b1?

I don't know and here's a link to 30.12b1: https://www.dropbox.com/s/gzdh8gnfpo...SX.tar.gz?dl=0

cxc · 2023-05-04, 00:14

That is wonderful, many, many thanks! (And I’ll be careful to backup my prime/local.txt before trying.)

harlee · 2023-05-05, 17:53

When a factor is found, 30.12 will then start a PRP on the exponent but gets the following error message as it violates the assignment rules. Could the assignment rule be update for the new software. Here is an example.

[Worker May 5 12:45] P-1 found a factor in stage #1, B1=480000.
[Worker May 5 12:45] M4578199 has a factor: 147722740303184008843879561 (P-1, B1=480000)
[Comm thread May 5 12:45] Sending result to server: UID: harlee/6917th_Scty_Sq, M4578199 has a factor: 147722740303184008843879561 (P-1, B1=480000)
[Comm thread May 5 12:45]
[Worker May 5 12:45] Starting Gerbicz error-checking PRP test of M4578199/known_factors using FMA3 FFT length 240K, Pass1=320, Pass2=768, clm=2
[Worker May 5 12:45] Preallocating disk space for the proof interim residues file p4578199.residues
[Worker May 5 12:45] PRP proof using power=7 and 64-bit hash size.
[Worker May 5 12:45] Proof requires 0.1GB of temporary disk space and uploading a 5MB proof file.
[Comm thread May 5 12:45] PrimeNet success code with additional info:
[Comm thread May 5 12:45] CPU credit is 0.0916 GHz-days.
[Comm thread May 5 12:45] Registering assignment: PRP M4578199
[Comm thread May 5 12:45] PrimeNet error 40: No assignment
[Comm thread May 5 12:45] ra: exponent 4578199 violates assignment rules.
[Comm thread May 5 12:45] Done communicating with server.

James Heinrich · 2023-05-05, 18:22

I can confirm that happens here too:

Code:

Sending result to server: UID: JamesHeinrich/i3-8100, M234029 has a factor: 30321910706821035885879083063 (ECM curve 2, B1=1000000, B2=2366257530), AID: C4A4CAC3552133D80032
PrimeNet success code with additional info:
CPU credit is 2.5243 GHz-days.
Registering assignment: PRP M234029
PrimeNet error 40: No assignment
ra: exponent 234029 violates assignment rules.
Done communicating with server.
Sending result to server: UID: JamesHeinrich/i3-8100, M234029/30321910706821035885879083063 is not prime.  Type-5 RES64: D0B187D94BB46B15. Wh4: C54208F4,6517,00000000
PrimeNet success code with additional info:
CPU credit is 0.0013 GHz-days.

Prime95 · 2023-05-05, 20:50

Quote:

Originally Posted by James Heinrich

I can confirm that happens here too

Does prime95 go ahead and run the PRP test anyway?

kruoli · 2023-05-05, 20:54

If it does, it is the behaviour I would hope for. That gives the highest probability the factor does not get spoiled.

2023-05-04, 00:14	#249
cxc "Catherine" Mar 2023 Melbourne 110111₂ Posts	That is wonderful, many, many thanks! (And I’ll be careful to backup my prime/local.txt before trying.) Last fiddled with by cxc on 2023-05-04 at 00:15

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

2023-04-20, 00:46	#246
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 17×487 Posts	Gwnum libraries dated after 2020-10-03.

2023-05-03, 21:46	#247
cxc "Catherine" Mar 2023 Melbourne 55₁₀ Posts	Hi! I’m doing ECMF and while 30.8 was fine on F19 ~ F24, it repeatedly aborted on the smaller Fermats with large bounds, B1 ~ 8e8, once it reached stage 2 (i.e. same error on different hardware). I wasn’t initially aware there was a Mac version of 30.10 – mention of it is somewhat buried on page 10 in post 107, but having found it and set it to work on F12, it works beautifully on stage 2. Thanks so much George and everyone else here helping to squash the bugs. I have two questions, the latter of which isn’t so important if 30.10b1 works fine: with respect to ECMF, is there anything I should know about 30.10b1 that has been fixed in later betas, that I might come across? And, is there a later build for Mac OS X than 30.10b1?

2023-05-05, 17:53	#250
harlee Sep 2006 Odenton, MD, USA 2⁴·13 Posts	When a factor is found, 30.12 will then start a PRP on the exponent but gets the following error message as it violates the assignment rules. Could the assignment rule be update for the new software. Here is an example. [Worker May 5 12:45] P-1 found a factor in stage #1, B1=480000. [Worker May 5 12:45] M4578199 has a factor: 147722740303184008843879561 (P-1, B1=480000) [Comm thread May 5 12:45] Sending result to server: UID: harlee/6917th_Scty_Sq, M4578199 has a factor: 147722740303184008843879561 (P-1, B1=480000) [Comm thread May 5 12:45] [Worker May 5 12:45] Starting Gerbicz error-checking PRP test of M4578199/known_factors using FMA3 FFT length 240K, Pass1=320, Pass2=768, clm=2 [Worker May 5 12:45] Preallocating disk space for the proof interim residues file p4578199.residues [Worker May 5 12:45] PRP proof using power=7 and 64-bit hash size. [Worker May 5 12:45] Proof requires 0.1GB of temporary disk space and uploading a 5MB proof file. [Comm thread May 5 12:45] PrimeNet success code with additional info: [Comm thread May 5 12:45] CPU credit is 0.0916 GHz-days. [Comm thread May 5 12:45] Registering assignment: PRP M4578199 [Comm thread May 5 12:45] PrimeNet error 40: No assignment [Comm thread May 5 12:45] ra: exponent 4578199 violates assignment rules. [Comm thread May 5 12:45] Done communicating with server.

2023-05-05, 20:54	#253
kruoli "Oliver" Sep 2017 Porta Westfalica, DE 23·71 Posts	If it does, it is the behaviour I would hope for. That gives the highest probability the factor does not get spoiled.