Prime95 30.8 (big P-1 changes, see post #551)

[kruoli]

Leaving out the "how-far-factored" bits field and setting B1=B2 will prevent stage 2 from running. You should set B1 manually according to what your high memory machine suggests for stage 1.

[Mark Rose]

I think setting MaxHighMemWorkers=0 has done the trick.

[storm5510]

Originally Posted by Mark Rose

These are 332M exponents. I'm forcing a large B2 on a "low" memory machine to get similar work done as on the 128 GB machine, which is progressing very slowly. Yes, I know that's not optimal.

I've put in some configuration that should hopefully have the low memory machine skip Stage 2 so I can move those assignments to the other machine. The high memory machine will be in storage a little before it can finish the work I'd like to finish.

What work? Running P-1 on all the 332M+ exponents that have an LL or PRP result without having a P-1 result. Is it needed? Nope.

No, it's not optimal.

As I wrote, I know how to get Prime95 to skip Stage 2, but how would it be resumed somewhere else? You would need a new B1 and B2.

[Mark Rose]

Originally Posted by storm5510

No, it's not optimal.

As I wrote, I know how to get Prime95 to skip Stage 2, but how would it be resumed somewhere else? You would need a new B1 and B2.

I'm manually setting B1 and B2 with Pminus1 lines. I'm running stage 2 on the high memory machine. I just need to save a few extra days of work, which I can do by running B1 on the low memory machine.

[storm5510]

Originally Posted by Mark Rose

I'm manually setting B1 and B2 with Pminus1 lines. I'm running stage 2 on the high memory machine. I just need to save a few extra days of work, which I can do by running B1 on the low memory machine.

OK. Maybe I need to restate my question differently, if I can...

What is the relationship between your ending bounds on the first machine and the starting bounds on the second machine? Example:

• 1st machine: B1=1e6, B2=1e6
• 2nd machine: B1=?, B2=?

I have used the Pminus1 form myself. Prime95 seems to like to punish for using it by leaving a lot of leftovers in its folder. CUDAPm1 runs this form quite well, if a person has a decent GPU.

[Mark Rose]

Originally Posted by storm5510

OK. Maybe I need to restate my question differently, if I can...

What is the relationship between your ending bounds on the first machine and the starting bounds on the second machine? Example:

• 1st machine: B1=1e6, B2=1e6
• 2nd machine: B1=?, B2=?

I have used the Pminus1 form myself. Prime95 seems to like to punish for using it by leaving a lot of leftovers in its folder. CUDAPm1 runs this form quite well, if a person has a decent GPU.

For example, I'm putting this as a line in worktodo.txt on the low mem machine:

Pminus1=N/A,1,2,363289319,-1,1565000,354386000

I'll let it finish Stage 1, then copy the files over and move the line to the high mem machine's worktodo.txt. No adjustment in B1 or B2.

The B1 and B2 values are similar to what would be picked if I ran the exponent entirely on the high mem machine as a Pfactor assignment.

[pepi37]

Prime 30.9 build3
Ryzen 7 5700X HT is off
Win 10

my settings

Memory=25000 during 7:30-23:30 else 27000
WorkerThreads=2
CoresPerTest=4

[May 17 23:38:12] Worker starting
[May 17 23:38:12] Setting affinity to run worker on CPU core #1
[May 17 23:38:12]
[May 17 23:38:12] P-1 on 64*920^306346+1 with B1=150000, B2=4000000
[May 17 23:38:12] Setting affinity to run helper thread 1 on CPU core #2
[May 17 23:38:12] Using all-complex FMA3 FFT length 384K, Pass1=384, Pass2=1K, clm=1, 4 threads
[May 17 23:38:12] Setting affinity to run helper thread 2 on CPU core #3
[May 17 23:38:12] Setting affinity to run helper thread 3 on CPU core #4
[May 17 23:38:13] 64*920^306346+1 stage 1 is 9.90% complete.
[May 17 23:39:19] 64*920^306346+1 stage 1 complete. 390084 transforms. Total time: 66.200 sec.
[May 17 23:39:19] Round off: 0.0625
[May 17 23:39:20] Inversion of stage 1 result complete. 5 transforms, 1 modular inverse. Time: 0.791 sec.
[May 17 23:39:20] Available memory is 26995MB.
[May 17 23:39:20] Setting affinity to run helper thread 1 on CPU core #2
[May 17 23:39:20] Switching to all-complex FMA3 FFT length 400K, Pass1=1280, Pass2=320, clm=2, 4 threads
[May 17 23:39:20] Estimated stage 2 vs. stage 1 runtime ratio: 0.242
[May 17 23:39:20] Setting affinity to run helper thread 3 on CPU core #4
[May 17 23:39:20] Setting affinity to run helper thread 2 on CPU core #3
[May 17 23:39:21] Using 2483MB of memory. D: 630, 72x728 polynomial multiplication.
[May 17 23:39:21] Setting affinity to run polymult helper thread on CPU core #2
[May 17 23:39:21] Setting affinity to run polymult helper thread on CPU core #3
[May 17 23:39:21] Setting affinity to run polymult helper thread on CPU core #4
[May 17 23:39:22] Stage 2 init complete. 1784 transforms. Time: 1.965 sec.
[May 17 23:39:35] 64*920^306346+1 stage 2 complete. 47297 transforms. Total time: 12.546 sec.
[May 17 23:39:35] Round off: 0.4375
[May 17 23:39:35] Stage 2 GCD complete. Time: 0.385 sec.
[May 17 23:39:35] 64*920^306346+1 completed P-1, B1=150000, B2=4046490, Wi4: 07AC0355

I never have this problem before and I didnot change anything, but for some reason Prime95 use only 2.4GB memory of 27GB that is allowed, so B2 is very low.
Hot to fix it ?

No other process then this is active so memory is free for Prime95

[Prime95]

Originally Posted by pepi37

[May 17 23:38:12] P-1 on 64*920^306346+1 with B1=150000, B2=4000000

Did your worktodo.txt entry request these bounds?

[pepi37]

Originally Posted by Prime95

Did your worktodo.txt entry request these bounds?

I enter thise values

[Prime95]

Originally Posted by pepi37

I enter thise values

Did you also specify how much TF has been done?

[pepi37]

Originally Posted by Prime95

Did you also specify how much TF has been done?

No I never do that...