Re: [help-GIFT] Optimizing

[risc]

On Fri, Sep 15, 2006 at 04:54:33PM -0500, address@hidden wrote:
> On Thu, Aug 17, 2006 at 10:45:50PM +0100, David Squire wrote:
> > address@hidden wrote:
> > 
> > >I don't find the branch idea to be a great one, because what appears to be 
> > >already happening is everyone is using their own branches, and the one in 
> > >the tree, which is the one that people new to the project will use, is 
> > >being left to rot. David's criticism of that optimization is valid, I'll 
> > >be adding a MAX_WIDTH and MAX_HEIGHT, so that both of us are satisfied. 
> > >All of my allocation changes are already in tree, and this one is the only 
> > >one that we both agree was a "badly implemented idea". 
> > 
> > Not so fast. I still haven't read everything that happened while I was
> > walking :) I had about 500 non-spam emails to deal with when I returned...
> > 
> > Cheers,
> > 
> > David
> > 
> > 
> > -- 
> > Dr David McG. Squire, Senior Lecturer, on sabbatical in 2006
> > Caulfield School of Information Technology, Monash University, Australia
> > CRICOS Provider No. 00008C        http://www.csse.monash.edu.au/~davids/
> > 
> 
> David,
> 
> Have you had time to read the other two patches (70 and 80)?
> 
> Heres a new version of patch 60, with your and johan's input.
> 
> I've been dead busy for a while, so i'm just getting around to this.
> 
> If you have no objections, i'd like to apply this, and patch 70 and 80
> come next tuesday. should make a big difference.
> 
> (yes, memset is faster. thanks johan)
> 
> Wolfgang:
> 
> I'm about 5 or 6 patches from getting to re-structuring as you 
> suggested. I just want to get this work in before tackling something
> else. :)
> 
> Julia Longtin <address@hidden>

> --- ChangeLog 2006-09-15 20:12:43.000000000 +0000
> +++ ../../dev/gift/ChangeLog  2006-09-15 21:21:52.000000000 +0000
> @@ -1,3 +1,9 @@
> +2006-09-15    <address@hidden>
> +
> +     * FeatureExtraction/gabor.c
> +     dont use calloc for conv, instead declare, and clear with memset.
> +     change to C99 datatypes in gabor_filter().
> +
>  2006-09-14    <address@hidden>
>      * scripts/perl/gift-add-collection.pl
>       Fixed a buglet in the last part of the help message, and added a

> --- ../../dev2/gift/FeatureExtraction/gabor.c 2006-08-28 17:18:09.000000000 
> +0000
> +++ FeatureExtraction/gabor.c 2006-09-15 21:44:01.000000000 +0000
> @@ -2,12 +2,16 @@
>  #include <stdlib.h>
>  #include <malloc.h>
>  #include <math.h>
> +/* for memset(), others */
>  #include <string.h>
>  #include <unistd.h>
>  #include <ppm.h>
>  
>  #include "gabor.h"
>  
> +/* for MAX_WIDTH and MAX_HEIGHT */
> +#include "gift_features.h"
> +
>  /* for uint32_t */
>  #include <stdint.h>
>  
> @@ -79,15 +83,15 @@
>       }
>  }
>  
> -void gabor_filter(double *image, int width, int height, int filter_scale, 
> int orientation, double ** kernelsxy, double *output) {
> +void gabor_filter(double *image, int width, int height, int filter_scale, 
> int orientation, double **kernelsxy, double *output) {
>  
> -     double *conv;
> -     int x, y, t_x, t_y;
> -     int i;
> +     uint32_t x, y;
> +     int32_t t_x, t_y;
> +     uint32_t i;
>       double * target_kernal;
> +     double conv[MAX_WIDTH*MAX_HEIGHT]; /* take advantage of our fixed image 
> size. */
>  
> -
> -     conv = (double *)calloc(width*height, sizeof(double));
> +     memset(&conv, 0, MAX_WIDTH*MAX_HEIGHT*sizeof(double)); 
>  
>       target_kernal=kernelsxy[filter_scale*num_gabors_per_scale+orientation];
>  
> @@ -144,5 +148,4 @@
>       }
>       }
>  
> -     free(conv);
>  }

> --- FeatureExtraction/gift_features.h 2001-02-01 11:22:30.000000000 +0000
> +++ ../../dev/gift/FeatureExtraction/gift_features.h  2006-09-15 
> 20:31:37.000000000 +0000
> @@ -24,4 +24,8 @@
>     adapting to Wolfgang's code */
>  #define WOLFGANG_HACK 
>  
> +/* fixed size buffer values, so that we dont code these into the .c files */
> +#define MAX_WIDTH  256
> +#define MAX_HEIGHT 256
> +
>  #endif

Replying to myself, just to make sure...

David,

Do you have any objections to the patch above, or the 70* and 80* patches i 
submitted last time arround?

Speak now, cause as soon as my laptop is set up, i'm commiting. :)

Julia Longtin <address@hidden>