|
2015-07-31, 20:33
|
#56 |
Sep 2009
40368 Posts |
You can get the backup from http://kendon.info/chris/m42463_37.backup.zip (it's 835Mb so will take a while to download).
And a core dump from re-running with ulimit -c unlimited in http://kendon.info/chris/m42463_37.dump.zip (also the log from the re-run). This is "only" 225Mb. Chris |
|
|
|
2015-08-04, 19:52
|
#57 |
Sep 2009
977 Posts |
Just a data point: I fed your dataset (unsorted) to a couple binaries of msieve, respectively built with AddressSanitizer (-fsanitize=address) and UndefinedBehaviourSanitizer (-fsanitize=undefined), with GCC 4.9.2, on Debian jessie amd64.
Neither binary crashes or otherwise complains when running -nc, both binaries build a matrix of the same dimensions and weight: Code:
matrix is 1326081 x 1326307 (379.9 MB) with weight 93617197 (70.58/col) Unsurprisingly, 1-thread LA on the asan-enabled binary is a bit less than 50% slower than 1-thread LA on the ubsan-enabled binary. Some parts of filtering are less memory-intensive than LA, so the gap between the two is narrower, less than 30%. |
|
|
|
|
2015-08-05, 05:57
|
#58 |
|
Sep 2009
977 Posts |
When I interrupt LA, whose ETA on Valgrind is > 30x longer than even the asan-enabled binary, I see
Code:
ERROR SUMMARY: 145545 errors from 1 contexts (suppressed: 0 from 0) I restarted the process, filtering "only" took 19455s last time. This run's matrix had the same dimensions and weight as the two other runs. Last fiddled with by debrouxl on 2015-08-05 at 05:57 |
|
|
|
|
2015-08-05, 16:06
|
#59 |
|
Sep 2009
207810 Posts |
That suggests the bug may be system specific. I ran it on:
Code:
chris@4core ~/ggnfs/trunk/tests/m42463_37 $ uname -a Linux 4core 3.8.0-19-generic #29-Ubuntu SMP Wed Apr 17 18:16:28 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux chris@4core ~/ggnfs/trunk/tests/m42463_37 $ ll /etc/*release -rw-r--r-- 1 root root 107 Nov 16 2014 /etc/lsb-release -rw-r--r-- 1 root root 241 Apr 19 2013 /etc/os-release chris@4core ~/ggnfs/trunk/tests/m42463_37 $ cat /etc/os-release NAME="Ubuntu" VERSION="13.04, Raring Ringtail" ID=ubuntu ID_LIKE=debian PRETTY_NAME="Ubuntu 13.04" VERSION_ID="13.04" HOME_URL="http://www.ubuntu.com/" SUPPORT_URL="http://help.ubuntu.com/" BUG_REPORT_URL="http://bugs.launchpad.net/ubuntu/" chris@4core ~/ggnfs/trunk/tests/m42463_37 $ cat /etc/lsb-release DISTRIB_ID=LinuxMint DISTRIB_RELEASE=15 DISTRIB_CODENAME=olivia DISTRIB_DESCRIPTION="Linux Mint 15 Olivia" chris@4core ~/ggnfs/trunk/tests/m42463_37 $ gcc --version gcc (Ubuntu/Linaro 4.7.3-1ubuntu1) 4.7.3 Copyright © 2012 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. Code:
cd /home/chris/ggnfs/trunk/tests/m42463_37 ulimit -c unlimited ulimit -a nice -n 19 "/home/chris/msieve-svn967/trunk/msieve" -s m42463_37.dat -l m42463_37.log -i m42463_37.ini -v -nf m42463_37.fb -nc ulimit -S -c 0 ulimit -a And the msieve binary is: Code:
chris@4core ~/ggnfs/trunk/tests/m42463_37 $ ll /home/chris/msieve-svn967/trunk/msieve
-rwxrwxr-x 1 chris chris 503198 Jun 26 2014 /home/chris/msieve-svn967/trunk/msieve
chris@4core ~/ggnfs/trunk/tests/m42463_37 $ file /home/chris/msieve-svn967/trunk/msieve
/home/chris/msieve-svn967/trunk/msieve: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.24, BuildID[sha1]=0x58c7530915df559878c2adefecd10baaa6dc5265, not stripped
chris@4core ~/ggnfs/trunk/tests/m42463_37 $ ldd /home/chris/msieve-svn967/trunk/msieve
linux-vdso.so.1 => (0x00007fff4effe000)
libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007fe6796cd000)
libz.so.1 => /lib/x86_64-linux-gnu/libz.so.1 (0x00007fe6794b6000)
libgmp.so.10 => /usr/local/lib/libgmp.so.10 (0x00007fe67923f000)
libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007fe678f3a000)
libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 (0x00007fe678d1d000)
libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007fe678954000)
/lib64/ld-linux-x86-64.so.2 (0x00007fe6798f2000)
Chris PS. Have you tried running a msieve binary built without any debugging support to see if it crashes? Reproducing the fault is always a good start when debugging. Last fiddled with by chris2be8 on 2015-08-05 at 16:10 Reason: Added PS. |
|
|
|
|
2015-08-05, 17:24
|
#60 |
|
Sep 2009
977 Posts |
The valgrind-specific output complains about manipulation of uninitialized memory:
Code:
memory use: 98.6 MB reading ideals above 720000 commencing singleton removal, initial pass ==19390== Syscall param write(buf) points to uninitialised byte(s) ==19390== at 0x6795620: __write_nocancel (syscall-template.S:81) ==19390== by 0x6730472: _IO_file_write@@GLIBC_2.2.5 (fileops.c:1253) ==19390== by 0x672FB32: new_do_write (fileops.c:530) ==19390== by 0x67312A4: _IO_do_write@@GLIBC_2.2.5 (fileops.c:503) ==19390== by 0x67309A0: _IO_file_xsputn@@GLIBC_2.2.5 (fileops.c:1324) ==19390== by 0x6726B0C: fwrite (iofwrite.c:43) ==19390== by 0x44A161: nfs_write_lp_file (in .../msieve_undefinedsanitizer/msieve) ==19390== by 0x4495D1: nfs_filter_relations (in .../msieve_undefinedsanitizer/msieve) ==19390== by 0x42A56E: factor_gnfs (in .../msieve_undefinedsanitizer/msieve) ==19390== by 0x40708F: msieve_run (in .../msieve_undefinedsanitizer/msieve) ==19390== by 0x40500D: factor_integer (in .../msieve_undefinedsanitizer/msieve) ==19390== by 0x404344: main (in .../msieve_undefinedsanitizer/msieve) ==19390== Address 0x4025006 is not stack'd, malloc'd or (recently) free'd ==19390== Uninitialised value was created by a stack allocation ==19390== at 0x449EA7: nfs_write_lp_file (in .../msieve_undefinedsanitizer/msieve) ==19390== ==19390== Warning: set address range perms: large range [0x4d8cc040, 0x6114c040) (undefined) ==19390== Warning: set address range perms: large range [0x1b8fb028, 0x2f17b058) (noaccess) memory use: 689.0 MB ==19390== Warning: set address range perms: large range [0x3a04c028, 0x6114c058) (noaccess) reading all ideals from disk ==19390== Warning: set address range perms: large range [0xa788040, 0x2e0102c4) (undefined) ==19390== Warning: set address range perms: large range [0xa788040, 0x2e010040) (defined) memory use: 568.5 MB keeping 22365672 ideals with weight <= 200, target excess is 116358 ==19390== Warning: set address range perms: large range [0xa788028, 0x2e0102dc) (noaccess) commencing in-memory singleton removal begin with 18750003 relations and 22365672 unique ideals reduce to 5953192 relations and 5741810 ideals in 23 passes max relations containing the same ideal: 93 ==19390== Warning: set address range perms: large range [0x3a04c028, 0x5d8d3960) (noaccess) removing 556781 relations and 518578 ideals in 38203 cliques [...] ==19390== ERROR SUMMARY: 145545 errors from 1 contexts (suppressed: 0 from 0) Code:
==15597== WARNING: MemorySanitizer: use-of-uninitialized-value
#0 0x7fbfd7781ed9 (.../msieve_memorysanitizer/msieve+0xc5ed9)
#0 0x7fbfd77571ba in main ??:?
#1 0x7fbfd6133b44 in __libc_start_main /build/glibc-Ir_s5K/glibc-2.19/csu/libc-start.c:287
#2 0x7fbfd7755c2c in _start ??:?
Uninitialized value was created by an allocation of 'b' in the stack frame of function 'get_cache_sizes'
#0 0x7fbfd7781190 in get_cache_sizes ??:?
SUMMARY: MemorySanitizer: use-of-uninitialized-value ??:0 ??
Exiting
EDIT: it's easy to get past that error and several other similar ones, but I hit a wall of bricks with gmp_sprintf (and if I comment that call, gmp_sscanf soon thereafter). strlen somehow barfs on the null-terminating character. Last fiddled with by debrouxl on 2015-08-05 at 18:19 |
|
|
|
|
2015-08-05, 20:07
|
#61 |
|
Sep 2009
977 Posts |
I can get the msan-enabled version to proceed further, by setting a custom memory alloc function which calls calloc() instead of malloc(), and then play whack-a-mole on the various reported uninitialized values. One of the surprising trouble-makers is mpz_cmp_ui().
Some of the hunks, in non-performance-critical paths (e.g. aprcl or cpuid), may make sense for upstream msieve ? When the code base is patched with the diff at the end of the message, the msan-enabled version goes to the LA phase: Code:
matrix is 1326129 x 1326307 (400.7 MB) with weight 117854396 (88.86/col)
sparse part has weight 90462298 (68.21/col)
saving the first 48 matrix rows for later
matrix includes 64 packed rows
matrix is 1326081 x 1326307 (379.9 MB) with weight 93617197 (70.58/col)
sparse part has weight 86322130 (65.08/col)
using block size 8192 and superblock size 786432 for processor cache size 8192 kB
==4989== WARNING: MemorySanitizer: use-of-uninitialized-value
#0 0x7f434f43f165 (.../msieve_memorysanitizer/msieve+0x16d165)
#0 0x7f434f42ed1b in block_lanczos ??:?
#1 0x7f434f406512 in nfs_solve_linear_system ??:?
#2 0x7f434f3a6f20 in factor_gnfs ??:?
#3 0x7f434f372512 in msieve_run_core common/driver.c:?
#4 0x7f434f370ad2 in msieve_run ??:?
#5 0x7f434f36cdc2 in factor_integer ??:?
#6 0x7f434f36efc8 in main ??:?
#7 0x7f434dd49b44 in __libc_start_main /build/glibc-Ir_s5K/glibc-2.19/csu/libc-start.c:287
#8 0x7f434f36c0dc in _start ??:?
Uninitialized value was created by an allocation of 'size' in the stack frame of function 'get_cache_sizes'
#0 0x7f434f387f60 in get_cache_sizes ??:?
SUMMARY: MemorySanitizer: use-of-uninitialized-value ??:0 ??
Here's a slightly cleaned up version of my local diff, I hope it applies. Compiling with -O1 and adding noinline attributes reduces inlining, and helps pinpointing the appropriate variables. Code:
Index: Makefile
===================================================================
--- Makefile (révision 988)
+++ Makefile (copie de travail)
@@ -17,9 +17,9 @@
# gcc with basic optimization (-march flag could
# get overridden by architecture-specific builds)
-CC = gcc
+CC = clang
WARN_FLAGS = -Wall -W
-OPT_FLAGS = -O3 -fomit-frame-pointer -march=native \
+OPT_FLAGS = -O1 -fno-omit-frame-pointer -march=native -fsanitize=memory -fsanitize-memory-track-origins -fPIE -pie \
-D_FILE_OFFSET_BITS=64 -DNDEBUG -D_LARGEFILE64_SOURCE
# use := instead of = so we only run the following once
Index: aprcl/mpz_aprcl32.c
===================================================================
--- aprcl/mpz_aprcl32.c (révision 988)
+++ aprcl/mpz_aprcl32.c (copie de travail)
@@ -1348,14 +1348,14 @@
void allocate_vars()
{
int i = 0;
- aiJS = malloc(PWmax * sizeof(mpz_t));
- aiJW = malloc(PWmax * sizeof(mpz_t));
- aiJX = malloc(PWmax * sizeof(mpz_t));
- aiJ0 = malloc(PWmax * sizeof(mpz_t));
- aiJ1 = malloc(PWmax * sizeof(mpz_t));
- aiJ2 = malloc(PWmax * sizeof(mpz_t));
- aiJ00 = malloc(PWmax * sizeof(mpz_t));
- aiJ01 = malloc(PWmax * sizeof(mpz_t));
+ aiJS = calloc(PWmax * sizeof(mpz_t), 1);
+ aiJW = calloc(PWmax * sizeof(mpz_t), 1);
+ aiJX = calloc(PWmax * sizeof(mpz_t), 1);
+ aiJ0 = calloc(PWmax * sizeof(mpz_t), 1);
+ aiJ1 = calloc(PWmax * sizeof(mpz_t), 1);
+ aiJ2 = calloc(PWmax * sizeof(mpz_t), 1);
+ aiJ00 = calloc(PWmax * sizeof(mpz_t), 1);
+ aiJ01 = calloc(PWmax * sizeof(mpz_t), 1);
for (i = 0 ; i < PWmax; i++)
{
mpz_init(aiJS[i]);
Index: common/driver.c
===================================================================
--- common/driver.c (révision 988)
+++ common/driver.c (copie de travail)
@@ -43,7 +43,7 @@
obj->nfs_fbfile_name = MSIEVE_DEFAULT_NFS_FBFILE;
if (nfs_fbfile_name)
obj->nfs_fbfile_name = nfs_fbfile_name;
- obj->mp_sprintf_buf = (char *)xmalloc(32 * MAX_MP_WORDS + 1);
+ obj->mp_sprintf_buf = (char *)xcalloc(32 * MAX_MP_WORDS + 1, 2);
savefile_init(&obj->savefile, savefile_name);
return obj;
Index: common/smallfact/smallfact.c
===================================================================
--- common/smallfact/smallfact.c (révision 988)
+++ common/smallfact/smallfact.c (copie de travail)
@@ -134,6 +134,7 @@
uint32 limit = 2;
mp_clear(&x);
+ mp_clear(&t);
x.nwords = 1;
x.val[0] = get_rand(&obj->seed1, &obj->seed2);
mp_copy(&x, &y);
Index: common/util.c
===================================================================
--- common/util.c (révision 988)
+++ common/util.c (copie de travail)
@@ -233,7 +233,7 @@
the L1/L2/L3 cache size of any Intel or AMD
processor made after ~1995 */
- uint32 a, b, c, d;
+ uint32 a = 0, b = 0, c = 0, d = 0;
uint8 is_intel, is_amd;
CPUID(0, a, b, c, d);
@@ -413,7 +413,7 @@
enum cpu_type cpu = cpu_generic;
#if defined(HAS_CPUID)
- uint32 a, b, c, d;
+ uint32 a = 0, b = 0, c = 0, d = 0;
CPUID(0, a, b, c, d);
if ((b & 0xff) == 'G') { /* "GenuineIntel" */
Index: demo.c
===================================================================
--- demo.c (révision 988)
+++ demo.c (copie de travail)
@@ -15,6 +15,8 @@
#include <msieve.h>
#include <signal.h>
+#include <gmp.h>
+
#ifdef HAVE_MPI
#include <mpi.h>
#endif
@@ -305,6 +307,11 @@
}
#endif
+void * new_malloc_mp(size_t size)
+{
+ return calloc(size, 1);
+}
+
/*--------------------------------------------------------------------*/
int main(int argc, char **argv) {
@@ -348,6 +355,7 @@
}
#endif
+mp_set_memory_functions(new_malloc_mp, NULL, NULL);
flags = MSIEVE_FLAG_USE_LOGFILE;
i = 1;
Index: gnfs/filter/duplicate.c
===================================================================
--- gnfs/filter/duplicate.c (révision 988)
+++ gnfs/filter/duplicate.c (copie de travail)
@@ -260,8 +260,8 @@
uint8 *hashtable;
uint32 blob[2];
uint32 log2_hashtable1_size;
- double rel_size = estimate_rel_size(savefile);
mpz_t scratch;
+ double rel_size;
uint8 *free_relation_bits;
uint32 *free_relations;
@@ -275,7 +275,9 @@
uint32 array_size;
relation_t tmp_rel;
+ mpz_init(scratch);
tmp_rel.factors = tmp_factors;
+ rel_size = estimate_rel_size(savefile);
logprintf(obj, "commencing duplicate removal, pass 1\n");
@@ -337,7 +339,6 @@
num_collisions = 0;
num_skipped_b = 0;
num_composite = 0;
- mpz_init(scratch);
savefile_read_line(buf, sizeof(buf), savefile);
while (!savefile_eof(savefile)) {
Index: gnfs/gnfs.c
===================================================================
--- gnfs/gnfs.c (révision 988)
+++ gnfs/gnfs.c (copie de travail)
@@ -268,6 +268,7 @@
savefile_t *savefile = &obj->savefile;
uint32 update = 1;
+ memset(buf, 0, sizeof(buf));
/* open the savefile; if the file already
exists and the first line contains n,
then we are restarting from a previous factorization */
@@ -294,6 +295,7 @@
up savefiles you wanted! */
savefile_open(savefile, SAVEFILE_WRITE);
+ memset(buf, 0, sizeof(buf));
gmp_sprintf(buf, "N %Zd\n", n);
savefile_write_line(savefile, buf);
savefile_flush(savefile);
Index: gnfs/poly/poly.c
===================================================================
--- gnfs/poly/poly.c (révision 988)
+++ gnfs/poly/poly.c (copie de travail)
@@ -24,14 +24,18 @@
FILE *fp;
char buf[BIGNUM_BUF_SIZE];
mpz_t read_n;
- mpz_t val, rpow;
+ mpz_t val, rpow, val2;
int32 status = 0;
+ mpz_init(read_n);
+ mpz_init(val);
+ mpz_init(val2);
+ mpz_init(rpow);
+
fp = fopen(obj->nfs_fbfile_name, "r");
if (fp == NULL)
return -1;
- buf[0] = 0;
fgets(buf, (int)sizeof(buf), fp);
if (buf[0] != 'N') {
fclose(fp);
@@ -42,10 +46,6 @@
/* check that the polynomial is for the
right number */
- mpz_init(read_n);
- mpz_init(val);
- mpz_init(rpow);
-
gmp_sscanf(buf + 2, "%Zd", read_n);
if (mpz_cmp(read_n, n) != 0) {
fclose(fp);
@@ -146,7 +146,8 @@
factors that are not reflected in the polynomials */
mpz_mod(val, val, n);
- if (mpz_cmp_ui(val, 0) != 0) {
+mpz_set_ui(val2, 0);
+ if (mpz_cmp(val, val2) != 0) {
logprintf(obj, "error: NFS input does not match polynomials\n");
logprintf(obj, "check that input doesn't have small factors\n");
exit(-1);
@@ -154,6 +155,7 @@
finished:
mpz_clear(read_n);
+ mpz_clear(val2);
mpz_clear(val);
mpz_clear(rpow);
return 0;
Index: gnfs/poly/polyutil.c
===================================================================
--- gnfs/poly/polyutil.c (révision 988)
+++ gnfs/poly/polyutil.c (copie de travail)
@@ -35,10 +35,10 @@
}
/*------------------------------------------------------------------*/
-static void poly_select_copy(poly_select_t *dest,
+/*static*/ __attribute__((noinline)) void poly_select_copy(poly_select_t *dest,
poly_select_t *src) {
- uint32 i;
+ uint32 i = 0;
dest->size_score = src->size_score;
dest->root_score = src->root_score;
@@ -87,7 +87,7 @@
}
/*------------------------------------------------------------------*/
-void analyze_poly(poly_config_t *config, poly_select_t *poly) {
+/*static*/ __attribute__((noinline)) void analyze_poly(poly_config_t *config, poly_select_t *poly) {
/* analyze a polynomial for sieving goodness
@@ -97,7 +97,7 @@
directly comparable to that of any other polynomials
given to this routine */
- uint32 i;
+ uint32 i = 0;
double root_score_r, root_score_a;
ddpoly_t ddr, dda;
mpz_poly_t *rpoly = &poly->rpoly;
Index: gnfs/poly/size_score.c
===================================================================
--- gnfs/poly/size_score.c (révision 988)
+++ gnfs/poly/size_score.c (copie de travail)
@@ -387,6 +387,13 @@
double *xrow, *yrow;
double res;
+for (i = 0; i <= MAX_POLY_DEGREE; i++) {
+for (j = 0; j <= MAX_POLY_DEGREE; j++) {
+xbinom[i][j] = 0.0;
+ybinom[i][j] = 0.0;
+}
+}
+
xbinom[1][0] = xh;
xbinom[1][1] = x0;
for (i = 2; i <= deg; i++) {
Index: gnfs/relation.c
===================================================================
--- gnfs/relation.c (révision 988)
+++ gnfs/relation.c (copie de travail)
@@ -96,7 +96,12 @@
uint32 num_factors_a;
uint32 array_size = 0;
uint8 *factors = r->factors;
+mpz_t zero, one;
+mpz_init(zero);
+mpz_init(one);
+mpz_set_ui(zero, 0);
+mpz_set_ui(one, 1);
/* read the relation coordinates */
a = strtoll(buf, &next_field, 10);
@@ -163,9 +168,9 @@
/* handle a rational factor of -1 */
eval_poly(polyval, a, b, rpoly);
- if (mpz_cmp_ui(polyval, 0) == 0)
+ if (mpz_cmp(polyval, zero) == 0)
return -6;
- if (mpz_cmp_ui(polyval, 0) < 0) {
+ if (mpz_cmp(polyval, zero) < 0) {
array_size = compress_p(factors, 0, array_size);
num_factors_r++;
mpz_abs(polyval, polyval);
@@ -203,7 +208,7 @@
/* if there are rational factors still to be accounted
for, assume they are small and find them by trial division */
- for (i = p = 0; mpz_cmp_ui(polyval, 1) != 0 &&
+ for (i = p = 0; mpz_cmp(polyval, one) != 0 &&
p < RELATION_TF_BOUND; i++) {
p += prime_delta[i];
@@ -215,13 +220,13 @@
}
}
- if (mpz_cmp_ui(polyval, 1) != 0)
+ if (mpz_cmp(polyval, one) != 0)
return -11;
/* read the algebraic factors */
eval_poly(polyval, a, b, apoly);
- if (mpz_cmp_ui(polyval, 0) == 0)
+ if (mpz_cmp(polyval, zero) == 0)
return -12;
mpz_abs(polyval, polyval);
@@ -249,7 +254,7 @@
/* if there are algebraic factors still to be accounted
for, assume they are small and find them by trial division */
- for (i = p = 0; mpz_cmp_ui(polyval, 1) != 0 &&
+ for (i = p = 0; mpz_cmp(polyval, one) != 0 &&
p < RELATION_TF_BOUND; i++) {
p += prime_delta[i];
@@ -261,7 +266,7 @@
}
}
- if (mpz_cmp_ui(polyval, 1) != 0)
+ if (mpz_cmp(polyval, one) != 0)
return -15;
r->num_factors_r = num_factors_r;
Index: include/gmp_xface.h
===================================================================
--- include/gmp_xface.h (révision 988)
+++ include/gmp_xface.h (copie de travail)
@@ -38,7 +38,7 @@
/*--------------------------------------------------------------------*/
static INLINE void gmp2mp(mpz_t src, mp_t *dest) {
- size_t count;
+ size_t count = 0;
mp_clear(dest);
mpz_export(dest->val, &count, -1, sizeof(uint32),
|
|
|
|
|
2015-08-06, 06:10
|
#62 |
|
Sep 2009
977 Posts |
Meh. 11 GB of free RAM is not enough for Valgrind to cope with the aforementioned execution :'(
This probably concludes my investigations. At least, they have turned up a reproducible warning with 145545 occurrences in nfs_write_lp_file. |
|
|
|
|
2015-08-19, 16:14
|
#63 |
Aug 2005
Seattle, WA
110110110102 Posts |
Incomplete factoring with SVN 984?
Apologies if this has been covered elsewhere; I couldn't find a prior report.
Lately when msieve finishes the square root phase of a factorization (and by "lately" I believe I mean since I started using v984), it is only printing one of the factors. Furthermore, if the cofactor is composite, msieve is not attempting to use any more dependencies, it's just quitting without finishing the factorization. Has this already been addressed with more recent versions? Is it a known issue? Thanks. |
|
|
|
|
2015-08-19, 18:10
|
#64 |
|
Tribal Bullet
Oct 2004
DD516 Posts |
Could you post a square root log that illustrates what's going on?
|
|
|
|
|
2015-08-19, 20:52
|
#65 | |
|
Aug 2005
Seattle, WA
2×877 Posts |
Quote:
Last fiddled with by jyb on 2015-08-19 at 20:53 |
|
|
|
|
|
2015-08-20, 00:37
|
#66 |
|
Aug 2015
22·5 Posts |
From where ?I can get msieve 1.53 source code and binaries ?
|
|
|
|
 |
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Msieve 1.50 feedback | firejuggler | Msieve | 99 | 2013-02-17 11:53 |
| Msieve v1.48 feedback | Jeff Gilchrist | Msieve | 48 | 2011-06-10 18:18 |
| Msieve 1.43 feedback | Jeff Gilchrist | Msieve | 47 | 2009-11-24 15:53 |
| Msieve 1.42 feedback | Andi47 | Msieve | 167 | 2009-10-18 19:37 |
| Msieve 1.41 Feedback | Batalov | Msieve | 130 | 2009-06-09 16:01 |
