Karim BELABAS on Sun, 22 Jun 2003 21:03:39 +0200 (MEST)

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: pari-2.2.6.alpha released

On Sun, 22 Jun 2003, Alexander Shumakovitch wrote:
> Dear PARI developers,
> On Thu, Jun 19, 2003 at 03:57:48PM +0200, Karim BELABAS wrote:
>> Dear PARI enthousiast,
>>   I would like to announce the release of pari-2.2.6.ALPHA. As usual, you can
>> fetch the sources from the link
> I'm now performing basic benchmarking of the new version of Pari and it
> seams to be much slower than the old one (i.e. 2.2.5). On the same
> hardware (Pentium III 800Mhz with Linux 2.4.20) and with the same
> compiler (gcc 2.95.4) Pari 2.2.6.alpha performance is somewhere around
>   +++ Total bench for gp-sta is 1376
>   +++ Total bench for gp-dyn is 1586
> 2.2.5, on the other hand, gives
>   +++ Total bench for gp-sta is 1262
>   +++ Total bench for gp-dyn is 1298
> The difference is as much as %%10--20!
> Is it something to be expected?

Actual time has (presumably) remained more or less the same.

This is a bug in the benchmark files: they compute total time using the
'gettime' routine without initializing the timers first, i.e without calling
gettime() at the beginning of the script.

It happened to be correct up to 2.2.5 because a few timers (including the
relevant one) were initialized on GP's startup. I removed this while cleaning
up the startup code.

I have updated all benchmark files in CVS to do the right thing.
Is it better now ?


Karim Belabas                     Tel: (+33) (0)1 69 15 57 48
Dép. de Mathématiques, Bât. 425   Fax: (+33) (0)1 69 15 60 19
Université Paris-Sud              http://www.math.u-psud.fr/~belabas/
F-91405 Orsay (France)            http://www.parigp-home.de/  [PARI/GP]