0
I Use This!
High Activity

Commits : Listings

Analyzed 21 days ago. based on code collected 21 days ago.
Jan 29, 2019 — Jan 29, 2020
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
README.params: new timings merge_linalg_sqrt.sh: added parameter 'prune' Anon32 More... about 12 years ago
got rid of 'dif' field (thus less memory used) put back 'mid' in prune() to refresh from time to time less verbose Anon32 More... about 12 years ago
The connected components were not sorted, since qsort was given length 0!!! Changed strategy for recomputing the connected components, new strategy should be better. Added some comments for possible optimizations (see FIXME's). Now prune works better, new timings being computed, see README.params. Anon32 More... about 12 years ago
kprune should be 1.0 by default (prune is not called) iprune should be at least 128 removed unused parameters to avoid compiler warnings Anon32 More... about 12 years ago
added -O2 to default OPTCFLAGS Anon32 More... about 12 years ago
added information about regression after r462 Anon32 More... about 12 years ago
better this way. Anon32 More... about 12 years ago
with strategies clearly isolated; introducing -prune for pruning the 2-primes. Anon32 More... about 12 years ago
updated with new version of the code Anon32 More... about 12 years ago
Since I am at it, clearing in allsqrt.c can be benefitial too. Anon32 More... about 12 years ago
stupid debugging message. Anon32 More... about 12 years ago
even less memory consuming. It can now handle a starting file of 21M relations in less than 3Go. Anon32 More... about 12 years ago
to take into account the modifs in purge.c . Anon32 More... about 12 years ago
Preparing for a more drastic version, when memory becomes prohibitive. Cleaned the options, etc. and modified merge_linalg_sqrt.sh accordingly. Anon32 More... about 12 years ago
freeing space in relations can be useful here and there... Anon32 More... about 12 years ago
print warning instead of error when given large prime does not divide norm Anon32 More... about 12 years ago
added new data for c85a and rsa110 Anon32 More... about 12 years ago
changed verbose/timing options to 'no' Note: ALLOW_VERBOSE = 'no' is not enough, the messages "squfof: ..." still appear in the output file Anon32 More... about 12 years ago
Makefile: added dependency wrt TIFALIB checknorms.c: print only first warnings for perfect squares print "... nth relation read" only in verbose mode Anon32 More... about 12 years ago
fixed bugs introduced in ggnfs2cado -> convert_rels change Anon32 More... about 12 years ago
Even better. For large examples, please use the new versions... Anon32 More... about 12 years ago
hashpair has more data; purge.c uses poly to compute less stupid hashing parameters. Anon32 More... about 12 years ago
now uses verbose mode; purge uses poly as first argument. Anon32 More... about 12 years ago
Misc, including some timings to watch how things are going on; prune works but is not connected, since the best strategy has to be thought of before. Anon32 More... about 12 years ago
Preparing the return of the pruning phase. Parameters tuning seem appropriate. Anon32 More... about 12 years ago
use new checknorms everywhere Anon32 More... about 12 years ago
now use Jerome's checknorms Anon32 More... about 12 years ago
added optimization by default changed ECM path to something less specific Anon32 More... about 12 years ago
USE_TAB set to 1, which means a faster version is on, reusing the length of row's that were already computed during the set up of the minimal spanning tree algo. Anon32 More... about 12 years ago
polyselect can be given a floating-point argument Anon32 More... about 12 years ago