Reservations - Page 19

Dylan14 · 2020-12-16, 04:51

Quote:

Originally Posted by Dylan14

Taking the following gaps, to test with gfndsieve + pfgw with the -gx flag:
n: 80001-80999, k: 25e4-30e4
n: 90001-90999, k: 20e4-25e4

The first range is 66.4% complete, and the second range is 59.1% complete. ETA 2-3 weeks for both.

matzetoni · 2021-01-02, 17:40

I'd like to reserve n=350-359, k=40e12-50e12 for Fermat factoring

Does FeromantCUDA do checkpoints or do I need to split the range manually?

ET_ · 2021-01-02, 18:25

Quote:

Originally Posted by matzetoni

I'd like to reserve n=350-359, k=40e12-50e12 for Fermat factoring

Does FeromantCUDA do checkpoints or do I need to split the range manually?

You can insert the whole string in the worktodo.txt file:

FermatFactor=350,359,40000000000000,50000000000000

and the program will do its checkpoints and split the ranges.

matzetoni · 2021-01-02, 22:24

Ah, now I see the checkpoint files, thanks!

Are there any tricks to optimize throughput? Like running two instances of FeromantCUDA? I have a GTX1060.

ET_ · 2021-01-03, 15:10

Quote:

Originally Posted by matzetoni

Ah, now I see the checkpoint files, thanks!

Are there any tricks to optimize throughput? Like running two instances of FeromantCUDA? I have a GTX1060.

I noticed that FeromantCUDA uses less GPU than (say) mmff or mfaktc: With FeromantCUDA and mmff run side by side, mmff takes 67% of GPU and Feromant only 34%.
And the CPU time required by FeromantCUDA is much higher.

You may try to run 2 or 3 instances of FeromantCUDA from different directories and evaluate the gain in throughput.

matzetoni · 2021-01-03, 16:07

Okay, it looks like I can run one instance more alongside the first one and it almost doubles throughput. I'll reserve and let run

FermatFactor=360,369,40000000000000,50000000000000

as well then.

matzetoni · 2021-01-08, 20:39

reserving FermatFactor=330,339,50e12,60e12

Dylan14 · 2021-01-22, 14:26

Quote:

Originally Posted by Dylan14

Taking the following gaps, to test with gfndsieve + pfgw with the -gx flag:
n: 80001-80999, k: 25e4-30e4
n: 90001-90999, k: 20e4-25e4

First range is 98.7% complete, ETA <1 day. Second range is 88% complete, ETA 4 days.

ATH · 2021-01-23, 06:22

Reserving:

FermatFactor=170,190,1100e12,1400e12

ET_ · 2021-01-23, 13:46

Quote:

Originally Posted by ATH

Reserving:

FermatFactor=170,189,1100e12,1400e12

Would you mind taking N to 190? we modified the reservation rules to adopt the same reservation strategy with other groups.

ATH · 2021-01-23, 20:32

Ok, I edited the post.

2021-01-03, 16:07	#204
matzetoni Feb 2019 97 Posts	Okay, it looks like I can run one instance more alongside the first one and it almost doubles throughput. I'll reserve and let run FermatFactor=360,369,40000000000000,50000000000000 as well then. Last fiddled with by matzetoni on 2021-01-03 at 16:07

2021-01-23, 06:22	#207
ATH Einyen Dec 2003 Denmark 2³×431 Posts	Reserving: FermatFactor=170,190,1100e12,1400e12 Last fiddled with by ATH on 2021-01-23 at 20:31

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Reservations	ET_	Operazione Doppi Mersennes	551	2023-03-28 15:37
Reservations	kar_bon	Riesel Prime Data Collecting (k*2^n-1)	136	2023-02-14 06:58
Reservations?	R.D. Silverman	NFS@Home	15	2015-11-29 23:18
Reservations	R.D. Silverman	Cunningham Tables	15	2011-03-04 21:01
4-5M Reservations	paulunderwood	3*2^n-1 Search	15	2008-06-08 03:29

2021-01-02, 17:40	#200
matzetoni Feb 2019 97 Posts	I'd like to reserve n=350-359, k=40e12-50e12 for Fermat factoring Does FeromantCUDA do checkpoints or do I need to split the range manually?

2021-01-02, 22:24	#202
matzetoni Feb 2019 97 Posts	Ah, now I see the checkpoint files, thanks! Are there any tricks to optimize throughput? Like running two instances of FeromantCUDA? I have a GTX1060.

2021-01-08, 20:39	#205
matzetoni Feb 2019 97 Posts	reserving FermatFactor=330,339,50e12,60e12

2021-01-23, 20:32	#209
ATH Einyen Dec 2003 Denmark 6570₈ Posts	Ok, I edited the post.