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

[James Heinrich]

Originally Posted by kriesel

30.x (not 3.x which would have been vintage ~1996, and perhaps not even released to others by George).

Correct, v4 was the "first web release". In those days version numbers advanced rapidly, more akin to build numbers used now. X.YbZ naming convention started with v25.1b1

[James Heinrich]

I think kruoli's suggestion is worth repeating: have Prime95 do an automatic PRP-CF after finding an ECM factor. Possibly configurable as to maximum exponent (although ECM isn't normally used on any exponents where the PRP-CF would take any significant amount of time). The idea being allow the ECM-F discoverer first chance at proving the PRP.

[LOBES]

I downloaded p95v3010b3.win64.zip and extracted (fresh install). Started it up, got the normal join gimps dialog box, configured it as usual, it pulled an assignment, and then immediately stops. Tried various fixes, but same result every time. As soon as I start it, it logs "Exchanging program options with server", then stops.

Fault bucket 1777693214842881550, type 4
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: prime95.exe
P2: 30.10.1.0
P3: 63d0b514
P4: prime95.exe
P5: 30.10.1.0
P6: 63d0b514
P7: c0000094
P8: 0000000002137a08
P9: 
P10:

So I kept all the files as is, downloaded p95v309b3.win64.zip and extracted to same location, overwriting files, and it starts up and works with no issues. So I'll stay at this level for now. Is there something new with v3010 I'm not aware of that would cause this?

Windows 10 Home, Intel(R) Core(TM) i5-3570K CPU @ 3.40GHz, 16.0 GB RAM

[LOBES]

I downloaded p95v3010b3.win64.zip and extracted (fresh install). Started it up, got the normal join gimps dialog box, configured it as usual, it pulled an assignment, and then immediately stops. Tried various fixes, but same result every time. As soon as I start it, it logs "Exchanging program options with server", then stops.

Fault bucket 1777693214842881550, type 4
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: prime95.exe
P2: 30.10.1.0
P3: 63d0b514
P4: prime95.exe
P5: 30.10.1.0
P6: 63d0b514
P7: c0000094
P8: 0000000002137a08
P9: 
P10:

So I kept all the files as is, downloaded p95v309b3.win64.zip and extracted to same location, overwriting files, and it starts up and works with no issues. So I'll stay at this level for now. Is there something new with v3010 I'm not aware of that would cause this?

Windows 10 Home, Intel(R) Core(TM) i5-3570K CPU @ 3.40GHz, 16.0 GB RAM

[James Heinrich]

Originally Posted by James Heinrich

I just found Prime95 disappeared on my v30.10b4 system

This is still happening randomly on my v30.10/ECM system. It's not usable as-is, I'll need to either revert it to v30.8 or have it do other work until a new build comes out.

[Prime95]

Originally Posted by LOBES

I downloaded p95v3010b3.win64.zip and extracted (fresh install). Started it up, got the normal join gimps dialog box, configured it as usual, it pulled an assignment, and then immediately stops.

Originally Posted by James Heinrich

This is still happening randomly on my v30.10/ECM system. It's not usable as-is, I'll need to either revert it to v30.8 or have it do other work until a new build comes out.

Please try version 30.11 when it becomes available. I'm testing on Linux right now.

[cmpute]

I'm trying to running P+1 tasks, and sometimes it crashes before stage 2. The assignment to reproduce is:

Pplus1=N/A,1,2,1213,-1,1000000000,221000000000,2,199,"327511,715079841864486572798550389025242773233223013545064445043124471"

The stack trace is

(gdb) backtrace
#0  0x000000000046105d in pp1_stage2_impl(pp1handle*) ()
#1  0x00000000004766a2 in pplus1 ()
#2  0x0000000000441e10 in primeContinue ()
#3  0x0000000000442039 in LauncherDispatch ()
#4  0x000000000044542f in Launcher ()
#5  0x00000000004b710a in ThreadStarter ()
#6  0x00007ffff7e22609 in start_thread (arg=<optimized out>) at pthread_create.c:477
#7  0x00007ffff7ab4133 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95

This happened to multiple machines with both v30.10b4 and v30.8b17, could you please fix this?

[Prime95]

30.11 build 1

PRE-BETA software!

Since last build:
-- ECM memory leak fixed
-- local.txt merged into prime.txt and deleted
-- PRP-CF done after new factor is found
-- save file created prior to starting stage 2
-- some gwnum bugs fixed for LLR/PFGW

Caveats? Same as previous builds:
Worker groups not yet implemented
Stage 2 time estimates and optimal B2 bounds could be off
Accurate estimates of stage 2 memory consumed may be off
Further stage 2 multithreading improvements are needed.
Stage 2 is pretty verbose, there's lots of code cleanup in my future.
P+1 apparently is broken (see previous post)
I haven't tested some of the new features as much as I'd like - I wanted to get this out before going on 10-day vacation

Back up prime.txt and local.txt before running this new version!

Windows 64-bit: https://mersenne.org/download/softwa...11b1.win64.zip
Linux 64-bit: https://mersenne.org/download/softwa...linux64.tar.gz
Source: https://mersenne.org/download/softwa...1b1.source.zip

[James Heinrich]

Originally Posted by Prime95

Please try version 30.11 when it becomes available. I'm testing on Linux right now.

So far it seems to be working -- it continued the exponent 30.10b4 was memory-leaking without apparent problem. Will let you know if that changes, of course.

[kruoli]

Resuming P-1 stage 1 files results in an access violation.

[Prime95]

Originally Posted by kruoli

Resuming P-1 stage 1 files results in an access violation.

Please send me the save file and worktodo.txt entry. Thanks.