|Charles Greathouse on Thu, 10 May 2012 02:49:03 +0200|
[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]
|Re: Errors in poliscycloprod|
Ah, excellent. It works in gp-gitd1d257d.exe and gp-gmp-gitd1d257d.exe. I think, just in case, poliscycloprod(x^3-1) or similar should be added to the test suite. Charles Greathouse Analyst/Programmer Case Western Reserve University On Wed, May 9, 2012 at 4:56 PM, Bill Allombert <Bill.Allombert@math.u-bordeaux1.fr> wrote: > On Wed, May 09, 2012 at 08:47:39AM -0400, Charles Greathouse wrote: >> Further, on the Windows git version, >> >> ? poliscycloprod(x^3-1) >> *** at top-level: poliscycloprod(x^3-1 >> *** ^-------------------- >> *** poliscycloprod: the PARI stack overflows ! >> current stack size: 157286400 (150.000 Mbytes) >> [hint] you can increase GP stack with allocatemem() >> >> which seems to happen for x^k - 1 for all k not a power of 2. > > You did not specify which GIT version, but I will hazard a guess: > gp-git48332bc.exe but 48332bc is not the latest git (48332bc is from Wed May 2) > > Unfortunately, gp-gitd1d257d.exe did not pass the test suite, so > was not put on the server. I just added it. Please check it. > > Under wine it does not exhibit this problem. > > Now, the test-cyclo bug: > On linux 32bit without GMP, the stack is set to 4MB. and the 'portable' FFT code use more > PARI stack than the GMP FFT code (which use malloc instead), so the stack overflow > only on 32bit/without GMP. > > Cheers, > Bill.