[Top][All Lists]

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

Re: [Bug-gnubg] Bug in sigmoid?

From: Joseph Heled
Subject: Re: [Bug-gnubg] Bug in sigmoid?
Date: Mon, 05 May 2003 11:15:17 +1200
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2) Gecko/20021202

Even if it is quicker I would advise against using this code until all net can be re-trained with the new sigmoid.


Joern Thyssen wrote:
On Sat, Apr 19, 2003 at 04:23:25PM +0200, Olivier Baur wrote

Le vendredi, 18 avr 2003, à 23:01 Europe/Paris, Joseph Heled a écrit :

Third, I need to see numbers on how faster this is on non scalar, regular x86 machine.

I don't know if this is a typo, but please note a "scalar" processor is what you call a "regular" processor; on the other hand, "non-scalar" and "super-scalar" refer to vector-computing.

Please note the speed increase I have measured in sigmoid2 (+60%) was for a regular scalar (ie *non* vector) implementation (on a PPC G4 processor); with a vector implementation of sigmoid2 (on the same processor), I actually got a whopping +250% speed increase...

So let me know what figures you get on a scalar x86 :-)

I tried your code:

I pasted it into neuralnet.c, made a call to ComputeSigTable and
replaced all calls to sigmoid with sigmoid2.

I analysed a 172 move match on 2-ply. It took 475 seconds with the old
code and 463 seconds with the new code.

In your posted code you use 201 points, but I tried with both 201 and
1001 points.

I also got slightly different results with typical differences in the
third or fourth digit.


Bug-gnubg mailing list

reply via email to

[Prev in Thread] Current Thread [Next in Thread]