View Single Post
Old 2016-05-20, 06:35   #620
Dubslow
Basketry That Evening!
 
Dubslow's Avatar
 
"Bunslow the Bold"
Jun 2011
40<A<43 -89<O<-88

3·29·83 Posts
Default

Okay, I've both fixed the bugs in yafu -nt and tested it by trying the top three from each of the three batches for the C173 under recent consideration. The results, with some caveats below:

Code:
(gdb) run "nfs($(cat num))" -nt hybrid1.job,hybrid2.job,hybrid3.job,cpucado1.job,cpucado2.job,cpucado3.job,gpumsieve1.job,gpumsieve2.job,gpumsieve3.job -v -threads 8
Starting program: /home/bill/yafu/yafu "nfs($(cat num))" -nt hybrid1.job,hybrid2.job,hybrid3.job,cpucado1.job,cpucado2.job,cpucado3.job,gpumsieve1.job,gpumsieve2.job,gpumsieve3.job -v -threads 8
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".


05/19/16 23:29:55 v1.34.5 @ Gravemind, System/Build Info: 
Using GMP-ECM 7.0-dev, Powered by GMP 6.0.0
detected        Intel(R) Core(TM) i7-2600K CPU @ 3.40GHz
detected L1 = 32768 bytes, L2 = 8388608 bytes, CL = 64 bytes
measured cpu frequency ~= 3392.300400
using 1 random witnesses for Rabin-Miller PRP checks

===============================================================
======= Welcome to YAFU (Yet Another Factoring Utility) =======
=======             bbuhrow@gmail.com                   =======
=======     Type help at any time, or quit to quit      =======
===============================================================
cached 78498 primes. pmax = 999983


>> test: starting trial sieving
nfs: parsed lpbr = 0, lpba = 0
test: warning: "hybrid1.job" is missing some paramters (0XFF). filling them.
nfs: parsed lpbr = 0, lpba = 0
test: warning: "hybrid2.job" is missing some paramters (0XFF). filling them.
nfs: parsed lpbr = 0, lpba = 0
test: warning: "hybrid3.job" is missing some paramters (0XFF). filling them.
nfs: parsed lpbr = 0, lpba = 0
test: warning: "cpucado1.job" is missing some paramters (0XFF). filling them.
nfs: parsed lpbr = 0, lpba = 0
test: warning: "cpucado2.job" is missing some paramters (0XFF). filling them.
nfs: parsed lpbr = 0, lpba = 0
test: warning: "cpucado3.job" is missing some paramters (0XFF). filling them.
nfs: parsed lpbr = 0, lpba = 0
test: warning: "gpumsieve1.job" is missing some paramters (0XFF). filling them.
nfs: parsed lpbr = 0, lpba = 0
test: warning: "gpumsieve2.job" is missing some paramters (0XFF). filling them.
nfs: parsed lpbr = 0, lpba = 0
test: warning: "gpumsieve3.job" is missing some paramters (0XFF). filling them.
test: trial sieving hybrid1.job

test: generating factor bases
test: fb generation took 38.3197 seconds
test: commencing test sieving of polynomial 0 on the algebraic side over range 65600000-65602000
gnfs-lasieve4I15e (with asm64): L1_BITS=15, SVN $Revision: 430 $
...snip...

test: found 5146 relations in a range of 1967 special-q
test: new best estimated total sieving time = 44 days 20h 19m 50s (with 8 threads)


test: trial sieving hybrid2.job

test: generating factor bases
test: fb generation took 37.5610 seconds
test: commencing test sieving of polynomial 1 on the algebraic side over range 65600000-65602000
gnfs-lasieve4I15e (with asm64): L1_BITS=15, SVN $Revision: 430 $
...snip...
test: found 4452 relations in a range of 1967 special-q
test: estimated total sieving time = 45 days 14h 35m 45s (with 8 threads)


test: trial sieving hybrid3.job

test: generating factor bases
test: fb generation took 37.7180 seconds
test: commencing test sieving of polynomial 2 on the algebraic side over range 65600000-65602000
gnfs-lasieve4I15e (with asm64): L1_BITS=15, SVN $Revision: 430 $
...snip...

test: found 4320 relations in a range of 1967 special-q
test: estimated total sieving time = 47 days 18h 52m 4s (with 8 threads)


test: trial sieving cpucado1.job

test: generating factor bases
test: fb generation took 36.9282 seconds
test: commencing test sieving of polynomial 3 on the algebraic side over range 65600000-65602000
gnfs-lasieve4I15e (with asm64): L1_BITS=15, SVN $Revision: 430 $
...snip...

test: found 4251 relations in a range of 1967 special-q
test: estimated total sieving time = 45 days 1h 5m 25s (with 8 threads)


test: trial sieving cpucado2.job

test: generating factor bases
test: fb generation took 37.1016 seconds
test: commencing test sieving of polynomial 4 on the algebraic side over range 65600000-65602000
gnfs-lasieve4I15e (with asm64): L1_BITS=15, SVN $Revision: 430 $
...snip...

test: found 4973 relations in a range of 1953 special-q
test: estimated total sieving time = 51 days 9h 45m 19s (with 8 threads)


test: trial sieving cpucado3.job

test: generating factor bases
test: fb generation took 38.3977 seconds
test: commencing test sieving of polynomial 5 on the algebraic side over range 65600000-65602000
gnfs-lasieve4I15e (with asm64): L1_BITS=15, SVN $Revision: 430 $
...snip...

test: found 4647 relations in a range of 1967 special-q
test: estimated total sieving time = 54 days 9h 10m 12s (with 8 threads)


test: trial sieving gpumsieve1.job

test: generating factor bases
test: fb generation took 38.0256 seconds
test: commencing test sieving of polynomial 6 on the algebraic side over range 65600000-65602000
gnfs-lasieve4I15e (with asm64): L1_BITS=15, SVN $Revision: 430 $
...snip...

test: found 4331 relations in a range of 1967 special-q
test: estimated total sieving time = 48 days 11h 56m 49s (with 8 threads)


test: trial sieving gpumsieve2.job

test: generating factor bases
test: fb generation took 38.1018 seconds
test: commencing test sieving of polynomial 7 on the algebraic side over range 65600000-65602000
gnfs-lasieve4I15e (with asm64): L1_BITS=15, SVN $Revision: 430 $
...snip...

test: found 5251 relations in a range of 1953 special-q
test: estimated total sieving time = 47 days 54m 14s (with 8 threads)


test: trial sieving gpumsieve3.job

test: generating factor bases
test: fb generation took 38.1440 seconds
test: commencing test sieving of polynomial 8 on the algebraic side over range 65600000-65602000
gnfs-lasieve4I15e (with asm64): L1_BITS=15, SVN $Revision: 430 $
...snip...

test: found 5579 relations in a range of 1953 special-q
test: new best estimated total sieving time = 43 days 15h 35m 58s (with 8 threads)
test: test sieving took 6544.09 seconds
test: "gpumsieve3.job" is the fastest poly
The caveats are that this took more than an hour on my primary desktop, so I can't really guarantee homogeneous computing environment, though the last 5 or 6 I think were more stable than the first 3-4. Also note that yield doesn't really correlate with speed (despite what I'm reasonably certain is an identical minrels estimate for all polys, since IIRC yafu calculates that based on lpb bounds, nothing about the actual coefficients). The top yield is the fastest, but below that there's considerable fluctuation. I'm not sure how much of this is due to the background compute environment, though I suspect with 8 threads available and a decent scheduler, the background is a relatively small factor.

All in all it does seem to be relatively a crapshoot, and I'd bet there's a decent chance some of the other top ~15-20 polys might sieve better, but I don't really want to invest time into that. (Yafu runs -nt single threaded by design, though I definitely wish I could have made use of all 8 threads available on this processor to speed it up.)

Yafu chose the following identical sieve parameters for all polys:

Code:
rlim: 65600000
alim: 65600000
lpbr: 31
lpba: 31
mfbr: 62
mfba: 62
rlambda: 2.6
alambda: 2.6

Last fiddled with by Dubslow on 2016-05-20 at 06:38
Dubslow is offline   Reply With Quote