[QUOTE=fivemack;297970]
Will try a C160 with the larger large primes and see if I can find the crossover point.[/QUOTE]

A C159 with Murphy 1.733 sieved with 31LP in almost exactly the same time as a similar-sized C159 with Murphy 2.183 and 30LP. So I think I prefer 31LP at that level too.

A few gratuitous and patently silly factors (non-algebraic):

82*2^17681-81 is a prp and divides MM17681
112*2^10457-111 is a prp and divides MM10457
10*2^863-9 divides MM863
1968*2^229-1967 divides MM229
1944*2^271-1943 divides MM271

Quite nice split
 

(2232.1045)

[code]
prp75 factor: 147862046859949267414841968169633849223604114880934142142092204556044870187
prp75 factor: 473607548107446907344419401894968758748844789613352095008694042735830826261
[/code]

Lucas1249 c262 = p115 * p147
[CODE]p115 = 1230907246701748850213178915950086177557919307463961418238191238338563780421891858424331033928188919064775254538919
p147 = 861662799748056902967441789531902541845512917066647559276041990818216830987090087949122773698681932527206937859817129749788517846169149201190846679[/CODE]
B+D

[COLOR=green](EDIT: the reason for it to be mildly interesting is that the previously known largest penultimate factors were p98 in Fibonacci and a recent p95 in Lucas factorizations.)[/COLOR]

[QUOTE=Batalov;301555]Lucas1249 c262 = p115 * p147
...
B+D

[COLOR=green](EDIT: the reason for it to be mildly interesting is that the previously known largest penultimate factors were p98 in Fibonacci and a recent p95 in Lucas factorizations.)[/COLOR][/QUOTE]

A second point of interest, the matrix was 14.59M^2 with a Serge
binary using zlib and aiming for density 100. This took c. 19 node-days
on our cluster, using six threads on a single node with 2 8-core cpus.

This was our first test-case for msieve-mpi; for which a "stock"
compile (no zlib or density) gave a 16.5M^2 matrix. The timing for
4 nodes, 16 cores/node is just in
[code]
Thu Jun 7 18:36:35 2012 initialized process (0,0) of 8 x 8 grid
Thu Jun 7 18:39:08 2012 matrix starts at (0, 0)
Thu Jun 7 18:39:09 2012 matrix is 2064182 x 1749228 (154.4 MB) with weight 55236916 (31.58/col)
Thu Jun 7 18:39:09 2012 sparse part has weight 22970121 (13.13/col)
Thu Jun 7 18:39:09 2012 saving the first 48 matrix rows for later
Thu Jun 7 18:39:09 2012 matrix includes 64 packed rows
Thu Jun 7 18:39:17 2012 matrix is 2064134 x 1749228 (132.4 MB) with weight 25403865 (14.52/col)
Thu Jun 7 18:39:17 2012 sparse part has weight 17224508 ( 9.85/col)
Thu Jun 7 18:39:17 2012 using block size 262144 for processor cache size 10240 kB
Thu Jun 7 18:39:18 2012 commencing Lanczos iteration
Thu Jun 7 18:39:18 2012 memory use: 207.8 MB
Thu Jun 7 18:39:38 2012 linear algebra at 0.0%, ETA 57h19m [/code]
which looks like a 50% savings in node-days.

Bruce (for Batalov+Dodson, with thanks to Jason, Greg and Lehigh's HPC group)

Slightly nicer split than fivemack's above:
[url]http://factordb.com/index.php?id=1100000000520244816[/url]
[code]PRP46 = 1620979858139715654164015438978958404260540673
PRP46 = 4376012889530283193868474423962504495931269487[/code]
854628.2079

Amazing split
 

[url]http://factordb.com/index.php?id=1100000000520518828[/url]

[code]starting SIQS on c92: 34164189955789536053590089105629252537838430612716698097997774549446989508175466797330208963
...
PRP46 = 5567633869391458324478744885324975214045868961
PRP46 = 6136213471868199135213045365634727120644813283[/code]
Ratio is 1.102.


[URL="http://factordb.com/sequences.php?se=1&aq=544608&action=range&fr=1138&to=1138"]A544608.1138[/URL]

There exists a couple of p76 . p76 splits in aliquot sequences.
One is in [URL="http://factordb.com/sequences.php?se=1&aq=9120&action=range&fr=869&to=870"]9120[/URL]:i869
Another is in 3906:i1848 ...

If you are interested in the ratio (and at least 45 digits), then there are some ratios like 1.00054 (11820:i1061), 1.00093(277344:i1508), 1.00086 (405420:i1101)...

There's also a 32090656388032554397472480978461 * 32091162549556209617436569529601 split (ratio 1.00002), if you look at smaller numbers.

There's a p108 . p108 nice split outside of aliquot sequences.

I had two p67 split in one of my sequences, I forgot which one, but I remember I was boasting on the forum about it. That is the max brilliant I ever split in... real life (i.e. not counting deliberately-manufactured brilliant numbers some people are posting here from time to time and ask for help to factor them).

I don't know if this is the right place to tell...

[code]
Trial-factoring M1951951 in [2^1, 2^41-1]
M1951951 has a factor: 54562292909897 - Program: L5.0x
M1951951 has a factor: 81981943 - Program: L5.0x
M1951951 has a factor: 1511251214927 - Program: L5.0x
M1951951 has 3 factors in [2^1, 2^41-1].
[/code]

54562292909897 was not found. I added it to mersenne-aries site; is there a way to show it to the GIMPS archive?

I'm mailing George as well.

Luigi

[QUOTE=ET_;303951]I don't know if this is the right place to tell...

[code]
Trial-factoring M1951951 in [2^1, 2^41-1]
M1951951 has a factor: 54562292909897 - Program: L5.0x
M1951951 has a factor: 81981943 - Program: L5.0x
M1951951 has a factor: 1511251214927 - Program: L5.0x
M1951951 has 3 factors in [2^1, 2^41-1].
[/code]

54562292909897 was not found. I added it to mersenne-aries site; is there a way to show it to the GIMPS archive?

I'm mailing George as well.

Luigi[/QUOTE]

Submitting it via the manual pages should work I believe.