5- table

[Raman]

3,1539M seems to be an attractive target, in my opinion
besides that 3,681- 3,687- with no known non-derived algebraic factors at all.

3,1539L P232
3,1539M C232

Ok, what difficulty does that quintic 11,539M translate into
when compared to that of a sextic? with that snfs-difficulty = 255.14
Why are quintics being neglected, which is not as bad as a quartic when compared to that of a sextic?
What is actually the apparent difficulty for that of 7,341+ (snfs 261.98) 6,374+ (snfs 264.57) etc.?

10,550M factors are being due day after tomorrow at around this same time only, actually.

Just today, that I had found out a solution to run multiple jobs within that compute cluster without causing any disturbance to others. By using that crontab, digging up with all that wasted CPU cycles within those internal nodes, trying to make out with the best ever possible throughput with the help of that compute cluster. Under low priority, that would execute as long as that CPU nodes are being idle, that others jobs could be able to rather pre-empt with that of mine, at any given time.

[frmky]

Originally Posted by fivemack

Would it be possible to redo the filtering with TARGET_DENSITY=110 or even 130 to see how the matrix size goes? http://www.mersenneforum.org/showpos...45&postcount=2 was suggesting that, for something as oversieved as my M941 run (and judging from the amount of clique reduction at density 90, this one is at least that oversieved), the larger target density gives a perceptibly quicker matrix.

(I appreciate that this would be a three-day run on a 64G machine so you are likely to say no ...)

I started it with instructions to others to kill it if they needed the memory, but no one did. Target density of 130 led to a significantly smaller 24.1M matrix. I try to run a timing test on a Core 2 Quad comparing this matrix to the original matrix soon.

[Batalov]

Originally Posted by Raman

3,1539M seems to be an attractive target, in my opinion
besides that 3,681- 3,687- with no known non-derived algebraic factors at all.

No one here is considering the extension tables to be attractive targets; there's a ton of p50s in them -- I just pulled one (and a p63 but that was just luck). In fact, if you want a lot of factors, run a 2t50 ecm on all 3+/3- extension numbers. See ECMNET. This is easily distributed on a cluster.

Originally Posted by Raman

Ok, what difficulty does that quintic 11,539M translate into
when compared to that of a sextic? with that snfs-difficulty = 255.14
Why are quintics being neglected, which is not as bad as a quartic when compared to that of a sextic?
What is actually the apparent difficulty for that of 7,341+ (snfs 261.98) 6,374+ (snfs 264.57) etc.?

See here, 3rd paragraph and footnote.

[Batalov]

Originally Posted by bdodson

Not so slow, the sieving for 5,785M C211 snfs.
Other projects on the x86_64 condor cluster went idle, and also the matrix only required 48hours. So the out-of-order factors are

> prp95 factor: 11155550887239787033082696874102453277665958156925604671852423308739616133659346997785443510271
> Tue Oct 12 18:39:48 2010 prp117 cofactor

Sieving for 5, 815L C182 is more substantial; 5,785M seems to have been the last of the easier numbers.
Regards, Bruce (for) Batalov+Dodson

.

[frmky]

Originally Posted by frmky

I started it with instructions to others to kill it if they needed the memory, but no one did. Target density of 130 led to a significantly smaller 24.1M matrix. I try to run a timing test on a Core 2 Quad comparing this matrix to the original matrix soon.

I finally got a chance to run the timings. The "density 90" matrix was 27095109 x 27095291 (9909.7 MB) with weight 2843906296 (104.96/col), sparse part has weight 2326802643 (85.87/col) while the "density 130" matrix was 24148824 x 24149002 (12031.3 MB) with weight 3412291541 (141.30/col), sparse part has weight 2912446714 (120.60/col). I ran timings using 8 Core 2 quads connected with GigE. I'm sure this amplifies the time difference somewhat since with GigE, communication is a significant bottleneck, and the denser matrix has fewer total data transfers and less data transferred during each iteration. The density 90 matrix had an ETA of 897 hours, while the density 130 matrix had an ETA of 762 hours, 15% faster. I will try constructing a density 130 matrix from the outset with 3,607-.

[frmky]

NFS@Home has completed 5,895M by GNFS. This one could have used a bit more sieving, but I just let the cluster run on the 20M matrix. The log is attached.

prp80 factor: 94946879076693120072882072244060631442348888153192895617269416419532376359079371
prp101 factor: 93050795618060019993527864672451793967658975383744506887149912645295446695942469928409949311863597631

[frmky]

NFS@Home has completed 5,389- by SNFS. The factors are

prp79 factor: 9387716962151042842693323146443792205381753954656905958476063709922148736350901
prp152 factor: 25095816395292565141457536097615378979918378832607352651584897663184722057414188452946820822572335773915521260031258185798033982466691092819952150667121

[Batalov]

The update is just posted and there are two easy snfs jobs for home enthusiasts: 5^459-1 and 5^477-1. Write to Sam to see if they are still available. Good luck!

[R.D. Silverman]

Originally Posted by Batalov

The update is just posted and there are two easy snfs jobs for home enthusiasts: 5^459-1 and 5^477-1. Write to Sam to see if they are still available. Good luck!

Posted where?

I looked on the extensions sub-page and did not see any base 5 extensions.
All I see is the base 3 extensions.

[R.D. Silverman]

Originally Posted by R.D. Silverman

Posted where?

I looked on the extensions sub-page and did not see any base 5 extensions.
All I see is the base 3 extensions.

Never mind. I see that Sam added the base 5 extensions to the actual
tables.

The mystery is why he did so when the base 3 extensions which have been
(presumably) around for a lot longer have NOT been added.

And what about base 6? Table 6- has only 3 composites left.

[R.D. Silverman]

Originally Posted by Batalov

The update is just posted and there are two easy snfs jobs for home enthusiasts: 5^459-1 and 5^477-1. Write to Sam to see if they are still available. Good luck!

5,457- is a GNFS C161; well within reach.

BTW, I see that he extended 5- but not 5+ to exponent 480.

He also added 2 more 6- composites.