Re: patch-fu

Hi!

I feel an urge to ask some questions... (Perhaps silly, but anyway...  
I know I could probably find the answers by digging in the code a bit  
deeper, but I admit I'm lazy...)

* The function gabor_filter, in gabor.c, now uses a fixed array of  
65536 doubles, instead of callocing the size indicated by the width  
and height that are passed as parameters to gabor_filter(...). Very  
well...
Are the width and height always 256, or can they be 128*512 or  
2*32768 or whatever?

* I guess that most modern CPUs have some kind of SSE2-ish features  
that gcc could use, but what would the effect of the patch be for an  
architecture that lacks it? (Something seriously old, pre MMX, or  
something else that perhaps one would not use for this application  
anyway...)

* Wouldn't memset be faster than looping and setting to zero?:
	for (i = 0; i < width*height; i++)
	  {
	    conv[i]= 0;   /* needs to be zeroed */
	}
and isn't width*height always 65536?




/Jonas





> Yet another patch..
>
> This one is a case of "add code to gain speed".
>
> By adding a special case handling of "kernel being tested
> is within the borders of the image",you gain speed by not
> performing a conditional in the loop, which allows for
> gcc 4.1's autovec engine to change your loop to a series
> of SSE2 instructions.
>
> Get all that? :)
>
> Anyways, this brings the speed down to ~1.3 on my machine.
>
> the 70-* and 80-* patches have equivilents for convolutions
> two and four, i'm just not getting it quite right yet...
>
> I'm also attaching my testing program, so that others can
> check my work.
>
> Julia Longtin <[email protected]>
> <80-ChangeLog>
> <80-FeatureExtraction_gabor.c_move_conditional_and_parallelize.patch>
> _______________________________________________
> help-GIFT mailing list
> [email protected]
> http://lists.gnu.org/mailman/listinfo/help-gift

_______________________________________________
help-GIFT mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/help-gift