sergiotarxz/libvips
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

faster vips_linear()

Browse Source 
put the 1ary path back, faster with gcc 4.8
This commit is contained in:
John Cupitt 2013-11-30 17:26:13 +00:00
parent a44377f03e
commit 1cffe216f5
1 changed files with 35 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

45
libvips/arithmetic/linear.c
View File

@ -38,6 +38,8 @@
* 31/10/11
* - rework as a class
* - removed the 1-ary constant path, no faster
* 30/11/13
* - 1ary is back, faster with gcc 4.8
*/
/*
@ -160,6 +162,19 @@ vips_linear_build( VipsObject *object )
return( 0 );
}
/* Non-complex input, any output, all bands of the constant equal.
*/
#define LOOP1( IN, OUT ) { \
IN * __restrict__ p = (IN *) in[0]; \
OUT * __restrict__ q = (OUT *) out; \
OUT a1 = a[0]; \
OUT b1 = b[0]; \
int sz = width * nb; \
\
for( x = 0; x < sz; x++ ) \
q[x] = a1 * (OUT) p[x] + b1; \
}
/* Non-complex input, any output.
*/
#define LOOPN( IN, OUT ) { \
@ -171,6 +186,16 @@ vips_linear_build( VipsObject *object )
q[i] = a[k] * (OUT) p[i] + b[k]; \
}
#define LOOP( IN, OUT ) { \
if( linear->a->n == 1 && linear->b->n == 1 ) { \
LOOP1( IN, OUT ); \
} \
else { \
LOOPN( IN, OUT ); \
} \
}
/* Complex input, complex output.
*/
#define LOOPCMPLXN( IN, OUT ) { \
@ -194,21 +219,21 @@ vips_linear_buffer( VipsArithmetic *arithmetic,
{
VipsImage *im = arithmetic->ready[0];
VipsLinear *linear = (VipsLinear *) arithmetic;
double *a = linear->a_ready;
double *b = linear->b_ready;
double * __restrict__ a = linear->a_ready;
double * __restrict__ b = linear->b_ready;
int nb = im->Bands;
int i, x, k;
switch( vips_image_get_format( im ) ) {
case VIPS_FORMAT_UCHAR: LOOPN( unsigned char, float ); break;
case VIPS_FORMAT_CHAR: LOOPN( signed char, float ); break;
case VIPS_FORMAT_USHORT: LOOPN( unsigned short, float ); break;
case VIPS_FORMAT_SHORT: LOOPN( signed short, float ); break;
case VIPS_FORMAT_UINT: LOOPN( unsigned int, float ); break;
case VIPS_FORMAT_INT: LOOPN( signed int, float ); break;
case VIPS_FORMAT_FLOAT: LOOPN( float, float ); break;
case VIPS_FORMAT_DOUBLE: LOOPN( double, double ); break;
case VIPS_FORMAT_UCHAR: LOOP( unsigned char, float ); break;
case VIPS_FORMAT_CHAR: LOOP( signed char, float ); break;
case VIPS_FORMAT_USHORT: LOOP( unsigned short, float ); break;
case VIPS_FORMAT_SHORT: LOOP( signed short, float ); break;
case VIPS_FORMAT_UINT: LOOP( unsigned int, float ); break;
case VIPS_FORMAT_INT: LOOP( signed int, float ); break;
case VIPS_FORMAT_FLOAT: LOOP( float, float ); break;
case VIPS_FORMAT_DOUBLE: LOOP( double, double ); break;
case VIPS_FORMAT_COMPLEX: LOOPCMPLXN( float, float ); break;
case VIPS_FORMAT_DPCOMPLEX: LOOPCMPLXN( double, double ); break;