@MinGW lost path to gmp-mparam.h and mul_fft.c:459: GNU MP assertion failed: nprime2

Torbjorn Granlund tg at gmplib.org
Sun Feb 20 12:31:43 CET 2011

Łukasz Ręcławowicz <lukasz.reclawowicz at gmail.com> writes:

  1) For undefined reference to `__mingw_vfprintf' I've added libmingwex.a to

Somebody with Windoze knowledge need to comment about this.

  2) For time.c: In function 'getrusage_backwards_p' I've corrected _ manualy
  $ diff -p -2 -r1.3 time.c
  diff: invalid option -- .
  diff: Try `diff --help' for more information.

Ok, so you get an error message from diff?  You need to give us more
context about where this error appears.

  3) But making tune(up) doesn't work for me.
  $ make tune
  make tuneup.exe
  make[1]: Entering directory `/d/gmp-5.0.1/tune'
  make[1]: `tuneup.exe' is up to date.
  make[1]: Leaving directory `/d/gmp-5.0.1/tune'
  Parameters for ./mpn/x86/p6/sse2/gmp-mparam.h
  System nie może odnaleźć określonej ścieżki. // The system can not find the
  path specified.
  System nie może odnaleźć określonej ścieżki.
  System nie może odnaleźć określonej ścieżki.
  Using: CPU cycle counter
  speed_precision 10000, speed_unittime 1 cycle, CPU freq unknown
  DEFAULT_MAX_SIZE 1000, fft_max_size 50000
  /* Generated by tuneup.c, 2011-02-20, gcc 4.5 */
  #define MOD_1_NORM_THRESHOLD 3
  #define MOD_1N_TO_MOD_1_1_THRESHOLD 8

This is not how the generated file looks like.  Did you edit it manually
before posting?

  #define MUL_FFT_TABLE3 \
  { { 565, 5}../mpn/generic/mul_fft.c:459: GNU MP assertion failed: nprime2 <
I cannot tell what causes this on your system.  The various preior
threshold values look reasonable.  Perhaps this is a compiler error,
perhaps it is some unusual threshold combination that causes tuneup to
pass bad parameters.

The shipped parameter files should be OK for most users.


More information about the gmp-discuss mailing list