Patch by Dzonatas and Callum Lerwick. Fp/vectorization patch which basically converts most of the irreversible decode codepath to floating point, eliminating a few rounds of int/fp conversion, resulting in a vast performance improvement, and an increase in accuracy.

ChangeLog

@@ -5,6 +5,12 @@ What's New for OpenJPEG
 ! : changed
 + : added
 
+November 13, 2007
+! [FOD] Patch by Dzonatas and Callum Lerwick.
+        Fp/vectorization patch which basically converts most of the irreversible decode codepath to floating point,
+        eliminating a few rounds of int/fp conversion, resulting in a vast performance improvement,
+	and an increase in accuracy.
+
 November 8, 2007
 ! [FOD] In t1.c, small change to avoid calling twice t1_getwmsedec()
         Patches from Callum Lewick:
@@ -12,7 +18,8 @@ November 8, 2007
 		- Fixed some spelling errors in dwt.c.
 
 November 5, 2007
-*+ [GB] Fixed a bug which prevented JPWL from working on multi-tiled images; added some more fields in the interface info structures (keep a list of markers, save start packet number for each tile)
+*+ [GB] Fixed a bug which prevented JPWL from working on multi-tiled images; added some more fields in the interface info structures 
+(keep a list of markers, save start packet number for each tile)
 
 October 23, 2007
 * [GB] Improved success for the linux build; OPJViewer shows all the COM contents
@@ -39,7 +46,8 @@ October 12, 2007
 
 October 10, 2007
 * [FOD] Patch from Callum Lewick. Clean up of j2klib.h for the aligned malloc stuff. 
-        It makes it work right with mingw, as _mm_malloc() isn't a macro, attempts to pave the way to using cmake to check for this stuff and combines a patch from Dana Fagerstrom at Sun that makes it use memalign() on Solaris
+        It makes it work right with mingw, as _mm_malloc() isn't a macro, attempts to pave the way to using cmake 
+        to check for this stuff and combines a patch from Dana Fagerstrom at Sun that makes it use memalign() on Solaris
         convert.c: Changed some error comments for TIFF images
 
 September 27, 2007
@@ -91,10 +99,12 @@ September 6, 2007
 * [Mathieu Malaterre] Fix unitialized read in img_fol (we may need a smarter initialize than memset)
 
 September 4, 2007
-+ [GB] Added some fields in the codestream_info structure: they are used to record the position of single tile parts. Changed also the write_index function in the codec, to reflect the presence of this new information.
++ [GB] Added some fields in the codestream_info structure: they are used to record the position of single tile parts. 
+		Changed also the write_index function in the codec, to reflect the presence of this new information.
 
 September 3, 2007
-+ [GB] Added the knowledge of JPSEC SEC and INSEC markers (you have to compile the JPWL project). Management of these markers is limited to skipping them without crashing: no real security function at this stage. Deprecated USE_JPSEC will be removed next
++ [GB] Added the knowledge of JPSEC SEC and INSEC markers (you have to compile the JPWL project). Management of these markers is limited to skipping them without crashing: 
+		no real security function at this stage. Deprecated USE_JPSEC will be removed next
 
 August 31, 2007
 * [GB] Fixed save capabilities in OPJViewer due to recent code upgrade
@@ -146,12 +156,15 @@ July 17, 2007
 + [FOD] Added support for RAW images. This module has been developped by the University of Perugia team. Thanks to them ! [image_to_j2k.c j2k_to_image.c convert.c convert.h]
 
 July 13, 2007
-! [FOD] Modified the memory allocation for codestreams containing multiple tiles. The memory is now allocated for each tile indenpendently, leading to an important decrease of the virtual memory needed. [j2k.c tcd.h tcd.c]
+! [FOD] Modified the memory allocation for codestreams containing multiple tiles. The memory is now allocated for each tile indenpendently, 
+		leading to an important decrease of the virtual memory needed. [j2k.c tcd.h tcd.c]
 ! [FOD] Modified old comments about the ability to decode mega-images and comments about the disk size necessary to do this. [image_to_j2k.c and frames_to_mj2.c]
 * [FOD] Added 2000 bytes for the memory allocation in cio.c for the minimum size of headers (useful in case of very small images) [cio.c]
 
 July 12, 2007
-* [GB] fixed a bug in JPWL module, which prevented to exploit the full error correction capability of RS codes (e.g. it gave up at 5 errors, even if 6 were correctable); defined a JPWL_MAXIMUM_EPB_ROOM for better customization of the maximum dimension of EPBs (the dimension is pre-calculated on an hypothesis, if it goes beyond 65535 there will be problems, thus we give a little less than the max, let's say 65450)
+* [GB] fixed a bug in JPWL module, which prevented to exploit the full error correction capability of RS codes (e.g. it gave up at 5 errors, 
+		even if 6 were correctable); defined a JPWL_MAXIMUM_EPB_ROOM for better customization of the maximum dimension of EPBs (the dimension 
+		is pre-calculated on an hypothesis, if it goes beyond 65535 there will be problems, thus we give a little less than the max, let's say 65450)
 
 July 8, 2007
 * [ANTONIN] fixed the size of the memory allocation in cio.c (confusion between bits and bytes)
@@ -186,13 +199,16 @@ May 31, 2007
 * [FOD] Fixed the parameters used for cinema compression (9-7 transform used instead of 5-3). Modified "image_to_j2k.c"
 
 May 24, 2007
-* [FOD] Bug fixed by Sylvain Munaut. Change in the reading of the POC marker. Since COD/COC can be anywhere in the header, the decoder cannot always know while decoding the POC marker the value of numlayers and numresolution.
+* [FOD] Bug fixed by Sylvain Munaut. Change in the reading of the POC marker. Since COD/COC can be anywhere in the header, the decoder cannot always know while decoding the POC marker 
+		the value of numlayers and numresolution.
 
 May 23, 2007
 ! [FOD] Patch suggested by Callum Lerwick <seg@haxxed.com>: "This makes the t1 data arrays dynamic, which greatly reduces cache thrashing. Also, some minor cleanup to prevent unnecessary casts"
 
 May 22, 2007
-! [FOD] Patch suggested by Callum Lerwick <seg@haxxed.com>: "Some formatting cleanups, so that the long function definitions and calls fit on screen. Use of prefix increment which is theoretically faster, in practice any sane compiler can optimize a postfix increment but its best not to count on such things. Consolidation of some redundant calculations in the inner loops, which becomes very useful in the future autovectorize patch."
+! [FOD] Patch suggested by Callum Lerwick <seg@haxxed.com>: "Some formatting cleanups, 
+		so that the long function definitions and calls fit on screen. Use of prefix increment which is theoretically faster, in practice any sane compiler can optimize a postfix 
+		increment but its best not to count on such things. Consolidation of some redundant calculations in the inner loops, which becomes very useful in the future autovectorize patch."
 ! [FOD] Patch suggested by Callum Lerwick <seg@haxxed.com>: "This changes the flag code in t1 to use a flag_t type, which can then be changed to reduce RAM usage. It is now typedef to a short."
 ! [FOD] Patch suggested by Callum Lerwick <seg@haxxed.com>: "This patch makes the t1 LUTs static. I actually intend this as a prelude to possibly eliminating some or all of the LUTs entirely."
 
@@ -241,14 +257,17 @@ March 29, 2007
 * [Parvatha] Equation to check multiple tile precincts. Modification pi.c
 ! [Parvatha] array size generation of pi->include in pi_initialise_encode().Modification in pi.c
 * [Parvatha] modification in pi_create_encode for tile part generation.Modification in pi.c
-+ [Parvatha] In tcd_rateallocate a variable stable_threshold which holds the valid threshold value. This is used to avoid error in case of a wrong threshold value in the last iteration. Modification in tcd.c.
++ [Parvatha] In tcd_rateallocate a variable stable_threshold which holds the valid threshold value. 
+			 This is used to avoid error in case of a wrong threshold value in the last iteration. Modification in tcd.c.
 
 March 28, 2007
 * [FOD] Fixed an historical bug in t1.c that leaded to the inclusion of useless 0xFF in the codestream. Thanks to Sylvain, Pascal and Parvatha !
 
 March 27, 2007
-+ [GB] Improved parsing in OPJViewer, as well some minor aesthetic modifications; support for image rendering with bit depths lower than 8 bits; can display an arbitrary frame of an MJ2 file (only in B/W, though); can reload a file; better resizing capabilities
++ [GB] Improved parsing in OPJViewer, as well some minor aesthetic modifications; support for image rendering with bit depths lower than 8 bits; 
-* [GB] Following to Hervé's suggestions, all the exit() calls, added by JPWL strict checking in t2.c and j2k.c, have been substituted with (object free'ing + opj_evt_message(EVT_ERROR) + return)
+		can display an arbitrary frame of an MJ2 file (only in B/W, though); can reload a file; better resizing capabilities
+* [GB] Following to Hervé's suggestions, all the exit() calls, added by JPWL strict checking in t2.c and j2k.c, 
+		have been substituted with (object free'ing + opj_evt_message(EVT_ERROR) + return)
 + [GB] Added linking to TIFF library in the JPWL VC6 workspaces
 
 March 23, 2007
@@ -294,7 +313,8 @@ VERSION 1.1.1 RELEASED
 ----------------------
 
 February 23, 2007
-* [GB] Fixed a copy-and-paste type assignment error (bool instead of int) in the JPWL section of decoder parameters structure in openjpeg.h; minor type-casting in jpwl_lib.c. As a result, now OPJViewer should run correctly when built against the most current SVN trunk of LibOpenJPEG.lib
+* [GB] Fixed a copy-and-paste type assignment error (bool instead of int) in the JPWL section of decoder parameters structure in openjpeg.h; minor type-casting in jpwl_lib.c. 
+		As a result, now OPJViewer should run correctly when built against the most current SVN trunk of LibOpenJPEG.lib
 + [GB] Linux makefile for the JPWL module; newlines at end of JPWL files
 
 February 22, 2007

libopenjpeg/dwt.c

@@ -5,6 +5,8 @@
  * Copyright (c) 2002-2003, Yannick Verschueren
  * Copyright (c) 2003-2007, Francois-Olivier Devaux and Antonin Descampe
  * Copyright (c) 2005, Herve Drolon, FreeImage Team
+ * Copyright (c) 2007, Jonathan Ballard <dzonatas@dzonux.net>
+ * Copyright (c) 2007, Callum Lerwick <seg@haxxed.com>
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -29,6 +31,9 @@
  * POSSIBILITY OF SUCH DAMAGE.
  */
 
+#ifdef __SSE__
+#include <xmmintrin.h>
+#endif
 
 #include "opj_includes.h"
 
@@ -41,12 +46,32 @@
 /** @name Local data structures */
 /*@{*/
 
-typedef struct dwt_local{
+typedef struct dwt_local {
-	int *	mem ;
+	int* mem;
+	int dn;
+	int sn;
+	int cas;
+} dwt_t;
+
+typedef union {
+	float	f[4];
+} v4;
+
+typedef struct v4dwt_local {
+	v4*	wavelet ;
 	int		dn ;
 	int		sn ;
 	int		cas ;
-	} dwt_t ; 
+} v4dwt_t ;
+
+static const float alpha =  1.586134342f; //  12994
+static const float beta  =  0.052980118f; //    434
+static const float gamma = -0.882911075f; //  -7233
+static const float delta = -0.443506852f; //  -3633
+
+static const float K      = 1.230174105f; //  10078
+/* FIXME: What is this constant? */
+static const float c13318 = 1.625732422f;
 
 /*@}*/
 
@@ -87,17 +112,13 @@ Forward 9-7 wavelet transform in 1-D
 */
 static void dwt_encode_1_real(int *a, int dn, int sn, int cas);
 /**
-Inverse 9-7 wavelet transform in 1-D
+Explicit calculation of the Quantization Stepsizes 
-*/
-static void dwt_decode_1_real(dwt_t *v);
-/**
-FIXME : comment ???
 */
 static void dwt_encode_stepsize(int stepsize, int numbps, opj_stepsize_t *bandno_stepsize);
 /**
 Inverse wavelet transform in 2-D.
 */
-static void dwt_decode_tile(opj_tcd_tilecomp_t * tilec, int stop , DWT1DFN fn);
+static void dwt_decode_tile(opj_tcd_tilecomp_t* tilec, int i, DWT1DFN fn);
 
 /*@}*/
 
@@ -285,102 +306,6 @@ static void dwt_encode_1_real(int *a, int dn, int sn, int cas) {
 	}
 }
 
-static void dwt_decode_sm(dwt_t* v, int k, int n, int x) {
-	int m = k > n ? n : k;
-	int l = v->mem[1];			//D(0);
-	int j;
-	int i;
-	for (i = 0; i < m; i++) {
-		j = l;
-		WS(i) -= fix_mul( ( l = WD(i) ) + j , x);
-	}
-	if( i < k ) {
-		l = fix_mul( l + l , x );
-		for (; i < k; i++)
-			WS(i) -= l;
-	}
-}
-
-static void dwt_decode_sp(dwt_t* v, int k, int n, int x) {
-	int m = k > n ? n : k;
-	int l = v->mem[1];			//D(0);
-	int j;
-	int i;
-	for (i = 0; i < m; i++) {
-		j = l;
-		WS(i) += fix_mul( ( l = WD(i) ) + j , x);
-	}
-	if( i < k ) {
-		l = fix_mul( l + l , x );
-		for (; i < k; i++)
-			WS(i) += l;
-	}
-}
-
-static void dwt_decode_dm(dwt_t* v, int k, int n, int x) {
-	int m = k >= n ? n-1 : k;
-	int l = v->mem[0];				//S(0);
-	int i;
-	int j;
-	for (i = 0; i < m; i++) {
-		j = l;
-		WD(i) -=  fix_mul( ( l = WS(i+1) ) + j , x);
-	}
-	if( i < k ) {
-		l = fix_mul( l + l , x );
-		for (; i < k; i++)
-			WD(i) -= l;
-	}
-}
-
-static void dwt_decode_dp(dwt_t* v, int k, int n, int x) {
-	int m = k >= n ? n-1 : k;
-	int l = v->mem[0];				//S(0);
-	int i;
-	int j;
-	for (i = 0; i < m; i++) {
-		j = l;
-		WD(i) +=  fix_mul( ( l = WS(i+1) ) + j , x);
-	}
-
-	if( i < k ) {
-		l = fix_mul( l + l , x );
-		for (; i < k; i++)
-			WD(i) += l;
-	}
-}
-
-
-/* <summary>                             */
-/* Inverse 9-7 wavelet transform in 1-D. */
-/* </summary>                            */
-static void dwt_decode_1_real(dwt_t* v) {
-	int i;
-	if (!v->cas) {
-		if ((v->dn > 0) || (v->sn > 1)) {	/* NEW :  CASE ONE ELEMENT */
-			for (i = 0; i < v->sn; i++)
-				WS(i) = fix_mul(WS(i), 10078);	/* 10076 */
-			for (i = 0; i < v->dn; i++)
-				WD(i) = fix_mul(WD(i), 13318);	/* 13320 */
-			dwt_decode_sm(v, v->sn, v->dn, 3633);
-			dwt_decode_dm(v, v->dn, v->sn, 7233);
-			dwt_decode_sp(v, v->sn, v->dn, 434);
-			dwt_decode_dp(v, v->dn, v->sn, 12994);
-		}
-	} else {
-		if ((v->sn > 0) || (v->dn > 1)) {	/* NEW :  CASE ONE ELEMENT */
-			for (i = 0; i < v->sn; i++)
-				WD(i) = fix_mul(WD(i), 10078);	/* 10076 */
-			for (i = 0; i < v->dn; i++)
-				WS(i) = fix_mul(WS(i), 13318);	/* 13320 */
-			dwt_decode_dm(v, v->sn, v->dn, 3633);
-			dwt_decode_sm(v, v->dn, v->sn, 7233);
-			dwt_decode_dp(v, v->sn, v->dn, 434);
-			dwt_decode_sp(v, v->dn, v->sn, 12994);
-		}
-	}
-}
-
 static void dwt_encode_stepsize(int stepsize, int numbps, opj_stepsize_t *bandno_stepsize) {
 	int p, n;
 	p = int_floorlog2(stepsize) - 13;
@@ -454,8 +379,8 @@ void dwt_encode(opj_tcd_tilecomp_t * tilec) {
 /* <summary>                            */
 /* Inverse 5-3 wavelet transform in 2-D. */
 /* </summary>                           */
-void dwt_decode(opj_tcd_tilecomp_t * tilec, int stop) {
+void dwt_decode(opj_tcd_tilecomp_t* tilec, int numres) {
-	dwt_decode_tile(tilec, stop, &dwt_decode_1);
+	dwt_decode_tile(tilec, numres, &dwt_decode_1);
 }
 
 
@@ -534,14 +459,6 @@ void dwt_encode_real(opj_tcd_tilecomp_t * tilec) {
 }
 
 
-/* <summary>                             */
-/* Inverse 9-7 wavelet transform in 2-D. */
-/* </summary>                            */
-void dwt_decode_real(opj_tcd_tilecomp_t * tilec, int stop) {
-	dwt_decode_tile(tilec, stop, dwt_decode_1_real);
-}
-
-
 /* <summary>                          */
 /* Get gain of 9-7 wavelet transform. */
 /* </summary>                         */
@@ -582,7 +499,7 @@ void dwt_calc_explicit_stepsizes(opj_tccp_t * tccp, int prec) {
 /* <summary>                             */
 /* Determine maximum computed resolution level for inverse wavelet transform */
 /* </summary>                            */
-static int dwt_decode_max_resolution(opj_tcd_resolution_t* r, int i) {
+static int dwt_decode_max_resolution(opj_tcd_resolution_t* restrict r, int i) {
 	int mr	= 1;
 	int w;
 	while( --i ) {
@@ -599,62 +516,310 @@ static int dwt_decode_max_resolution(opj_tcd_resolution_t* r, int i) {
 /* <summary>                            */
 /* Inverse wavelet transform in 2-D.     */
 /* </summary>                           */
-static void dwt_decode_tile(opj_tcd_tilecomp_t * tilec, int stop, DWT1DFN dwt_1D) {
+static void dwt_decode_tile(opj_tcd_tilecomp_t* tilec, int numres, DWT1DFN dwt_1D) {
-	opj_tcd_resolution_t* tr;
-	int i, j, k;
-	int *a = NULL;
-	int *aj = NULL;
-	int w; //, l;
-	int rw;			/* width of the resolution level computed  */
-	int rh;			/* height of the resolution level computed  */
 	dwt_t h;
 	dwt_t v;
 
-	if( 1 > ( i = tilec->numresolutions - stop ) )
+	opj_tcd_resolution_t* tr = tilec->resolutions;
-		return ;
 
-	tr = tilec->resolutions;
+	int rw = tr->x1 - tr->x0;	/* width of the resolution level computed */
+	int rh = tr->y1 - tr->y0;	/* height of the resolution level computed */
 
-	w = tilec->x1-tilec->x0;
+	int w = tilec->x1 - tilec->x0;
-	a = tilec->data;
 
-	h.mem = (int*) opj_aligned_malloc(dwt_decode_max_resolution(tr, i) * sizeof(int));
+	h.mem = opj_aligned_malloc(dwt_decode_max_resolution(tr, numres) * sizeof(int));
 	v.mem = h.mem;
 
+	while( --numres) {
+		int * restrict tiledp = tilec->data;
+		int j;
+
+		++tr;
+		h.sn = rw;
+		v.sn = rh;
+
 		rw = tr->x1 - tr->x0;
 		rh = tr->y1 - tr->y0;
 
-	while( --i ) {
+		h.dn = rw - h.sn;
-		tr++;
-		h.sn = rw;
-		v.sn = rh;
-		h.dn = ( rw = tr->x1 - tr->x0 ) - h.sn;
-		v.dn = ( rh = tr->y1 - tr->y0 ) - v.sn;
-				
 		h.cas = tr->x0 % 2;
-		v.cas = tr->y0 % 2;
 
-		aj = a;
+		for(j = 0; j < rh; ++j) {
-		j = rh;
+			dwt_interleave_h(&h, &tiledp[j*w]);
-		while( j-- ) {		
-			dwt_interleave_h(&h, aj);
 			(dwt_1D)(&h);
-			k = rw;
+			memcpy(&tiledp[j*w], h.mem, rw * sizeof(int));
-			while( k-- )
-				aj[k] = h.mem[k];
-			aj += w;
 		}
 
-		aj = a;
+		v.dn = rh - v.sn;
-		j = rw;
+		v.cas = tr->y0 % 2;
-		while( j-- ) {
+
-			dwt_interleave_v(&v, aj, w);
+		for(j = 0; j < rw; ++j){
+			int k;
+			dwt_interleave_v(&v, &tiledp[j], w);
 			(dwt_1D)(&v);
-			k = rh;
+			for(k = 0; k < rh; ++k) {
-			while( k-- )
+				tiledp[k * w + j] = v.mem[k];
-				aj[k * w] = v.mem[k];
+			}
-			aj++;
 		}
 	}
 	opj_aligned_free(h.mem);
 }
+
+static void v4dwt_interleave_h(v4dwt_t* restrict w, float* restrict a, int x, int size){
+	float* restrict bi = (float*) (w->wavelet + w->cas);
+	int count = w->sn;
+	int i, k;
+	for(k = 0; k < 2; ++k){
+		for(i = 0; i < count; ++i){
+			int j = i;
+			bi[i*8    ] = a[j];
+			j += x;
+			if(j > size) continue;
+			bi[i*8 + 1] = a[j];
+			j += x;
+			if(j > size) continue;
+			bi[i*8 + 2] = a[j];
+			j += x;
+			if(j > size) continue;
+			bi[i*8 + 3] = a[j];
+		}
+		bi = (float*) (w->wavelet + 1 - w->cas);
+		a += w->sn;
+		size -= w->sn;
+		count = w->dn;
+	}
+}
+
+static void v4dwt_interleave_v(v4dwt_t* restrict v , float* restrict a , int x){
+	v4* restrict bi = v->wavelet + v->cas;
+	int i;
+	for(i = 0; i < v->sn; ++i){
+		memcpy(&bi[i*2], &a[i*x], 4 * sizeof(float));
+	}
+	a += v->sn * x;
+	bi = v->wavelet + 1 - v->cas;
+	for(i = 0; i < v->dn; ++i){
+		memcpy(&bi[i*2], &a[i*x], 4 * sizeof(float));
+	}
+}
+
+#ifdef __SSE__
+
+static void v4dwt_decode_step1_sse(v4* w, int count, const __m128 c){
+	__m128* restrict vw = (__m128*) w;
+	int i;
+	for(i = 0; i < count; ++i){
+		__m128 tmp = vw[i*2];
+		vw[i*2] = tmp * c;
+	}
+}
+
+static void v4dwt_decode_step2_sse(v4* l, v4* w, int k, int m, __m128 c){
+	__m128* restrict vl = (__m128*) l;
+	__m128* restrict vw = (__m128*) w;
+	int i;
+	for(i = 0; i < m; ++i){
+		__m128 tmp1 = vl[ 0];
+		__m128 tmp2 = vw[-1];
+		__m128 tmp3 = vw[ 0];
+		vw[-1] = tmp2 + ((tmp1 + tmp3) * c);
+		vl = vw;
+		vw += 2;
+	}
+	if(m >= k){
+		return;
+	}
+	c += c;
+	c *= vl[0];
+	for(; m < k; ++m){
+		__m128 tmp = vw[-1];
+		vw[-1] = tmp + c;
+		vw += 2;
+	}
+}
+
+#else
+
+static void v4dwt_decode_step1(v4* w, int count, const float c){
+	float* restrict fw = (float*) w;
+	int i;
+	for(i = 0; i < count; ++i){
+		float tmp1 = fw[i*8    ];
+		float tmp2 = fw[i*8 + 1];
+		float tmp3 = fw[i*8 + 2];
+		float tmp4 = fw[i*8 + 3];
+		fw[i*8    ] = tmp1 * c;
+		fw[i*8 + 1] = tmp2 * c;
+		fw[i*8 + 2] = tmp3 * c;
+		fw[i*8 + 3] = tmp4 * c;
+	}
+}
+
+static void v4dwt_decode_step2(v4* l, v4* w, int k, int m, float c){
+	float* restrict fl = (float*) l;
+	float* restrict fw = (float*) w;
+	int i;
+	for(i = 0; i < m; ++i){
+		float tmp1_1 = fl[0];
+		float tmp1_2 = fl[1];
+		float tmp1_3 = fl[2];
+		float tmp1_4 = fl[3];
+		float tmp2_1 = fw[-4];
+		float tmp2_2 = fw[-3];
+		float tmp2_3 = fw[-2];
+		float tmp2_4 = fw[-1];
+		float tmp3_1 = fw[0];
+		float tmp3_2 = fw[1];
+		float tmp3_3 = fw[2];
+		float tmp3_4 = fw[3];
+		fw[-4] = tmp2_1 + ((tmp1_1 + tmp3_1) * c);
+		fw[-3] = tmp2_2 + ((tmp1_2 + tmp3_2) * c);
+		fw[-2] = tmp2_3 + ((tmp1_3 + tmp3_3) * c);
+		fw[-1] = tmp2_4 + ((tmp1_4 + tmp3_4) * c);
+		fl = fw;
+		fw += 8;
+	}
+	if(m < k){
+		float c1;
+		float c2;
+		float c3;
+		float c4;
+		c += c;
+		c1 = fl[0] * c;
+		c2 = fl[1] * c;
+		c3 = fl[2] * c;
+		c4 = fl[3] * c;
+		for(; m < k; ++m){
+			float tmp1 = fw[-4];
+			float tmp2 = fw[-3];
+			float tmp3 = fw[-2];
+			float tmp4 = fw[-1];
+			fw[-4] = tmp1 + c1;
+			fw[-3] = tmp2 + c2;
+			fw[-2] = tmp3 + c3;
+			fw[-1] = tmp4 + c4;
+			fw += 8;
+		}
+	}
+}
+
+#endif
+
+/* <summary>                             */
+/* Inverse 9-7 wavelet transform in 1-D. */
+/* </summary>                            */
+static void v4dwt_decode(v4dwt_t* restrict dwt){
+	int a, b;
+	if(dwt->cas == 0) {
+		if(!((dwt->dn > 0) || (dwt->sn > 1))){
+			return;
+		}
+		a = 0;
+		b = 1;
+	}else{
+		if(!((dwt->sn > 0) || (dwt->dn > 1))) {
+			return;
+		}
+		a = 1;
+		b = 0;
+	}
+#ifdef __SSE__
+	v4dwt_decode_step1_sse(dwt->wavelet+a, dwt->sn, _mm_set1_ps(K));
+	v4dwt_decode_step1_sse(dwt->wavelet+b, dwt->dn, _mm_set1_ps(c13318));
+	v4dwt_decode_step2_sse(dwt->wavelet+b, dwt->wavelet+a+1, dwt->sn, int_min(dwt->sn, dwt->dn-a), _mm_set1_ps(delta));
+	v4dwt_decode_step2_sse(dwt->wavelet+a, dwt->wavelet+b+1, dwt->dn, int_min(dwt->dn, dwt->sn-b), _mm_set1_ps(gamma));
+	v4dwt_decode_step2_sse(dwt->wavelet+b, dwt->wavelet+a+1, dwt->sn, int_min(dwt->sn, dwt->dn-a), _mm_set1_ps(beta));
+	v4dwt_decode_step2_sse(dwt->wavelet+a, dwt->wavelet+b+1, dwt->dn, int_min(dwt->dn, dwt->sn-b), _mm_set1_ps(alpha));
+#else
+	v4dwt_decode_step1(dwt->wavelet+a, dwt->sn, K);
+	v4dwt_decode_step1(dwt->wavelet+b, dwt->dn, c13318);
+	v4dwt_decode_step2(dwt->wavelet+b, dwt->wavelet+a+1, dwt->sn, int_min(dwt->sn, dwt->dn-a), delta);
+	v4dwt_decode_step2(dwt->wavelet+a, dwt->wavelet+b+1, dwt->dn, int_min(dwt->dn, dwt->sn-b), gamma);
+	v4dwt_decode_step2(dwt->wavelet+b, dwt->wavelet+a+1, dwt->sn, int_min(dwt->sn, dwt->dn-a), beta);
+	v4dwt_decode_step2(dwt->wavelet+a, dwt->wavelet+b+1, dwt->dn, int_min(dwt->dn, dwt->sn-b), alpha);
+#endif
+}
+
+/* <summary>                             */
+/* Inverse 9-7 wavelet transform in 2-D. */
+/* </summary>                            */
+void dwt_decode_real(opj_tcd_tilecomp_t* restrict tilec, int numres){
+	v4dwt_t h;
+	v4dwt_t v;
+
+	opj_tcd_resolution_t* res = tilec->resolutions;
+
+	int rw = res->x1 - res->x0;	/* width of the resolution level computed */
+	int rh = res->y1 - res->y0;	/* height of the resolution level computed */
+
+	int w = tilec->x1 - tilec->x0;
+
+	h.wavelet = (v4*) opj_aligned_malloc((dwt_decode_max_resolution(res, numres)+5) * sizeof(v4));
+	v.wavelet = h.wavelet;
+
+	while( --numres) {
+		float * restrict aj = (float*) tilec->data;
+		int bufsize = (tilec->x1 - tilec->x0) * (tilec->y1 - tilec->y0);
+		int j;
+
+		h.sn = rw;
+		v.sn = rh;
+
+		++res;
+
+		rw = res->x1 - res->x0;	/* width of the resolution level computed */
+		rh = res->y1 - res->y0;	/* height of the resolution level computed */
+
+		h.dn = rw - h.sn;
+		h.cas = res->x0 % 2;
+
+		for(j = rh; j > 0; j -= 4){
+			v4dwt_interleave_h(&h, aj, w, bufsize);
+			v4dwt_decode(&h);
+			if(j >= 4){
+				int k;
+				for(k = rw; --k >= 0;){
+					aj[k    ] = h.wavelet[k].f[0];
+					aj[k+w  ] = h.wavelet[k].f[1];
+					aj[k+w*2] = h.wavelet[k].f[2];
+					aj[k+w*3] = h.wavelet[k].f[3];
+				}
+			}else{
+				int k;
+				for(k = rw; --k >= 0;){
+					switch(j) {
+						case 3: aj[k+w*2] = h.wavelet[k].f[2];
+						case 2: aj[k+w  ] = h.wavelet[k].f[1];
+						case 1: aj[k    ] = h.wavelet[k].f[0];
+					}
+				}
+			}
+			aj += w*4;
+			bufsize -= w*4;
+		}
+
+		v.dn = rh - v.sn;
+		v.cas = res->y0 % 2;
+
+		aj = (float*) tilec->data;
+		for(j = rw; j > 0; j -= 4){
+			v4dwt_interleave_v(&v, aj, w);
+			v4dwt_decode(&v);
+			if(j >= 4){
+				int k;
+				for(k = 0; k < rh; ++k){
+					memcpy(&aj[k*w], &v.wavelet[k], 4 * sizeof(float));
+				}
+			}else{
+				int k;
+				for(k = 0; k < rh; ++k){
+					memcpy(&aj[k*w], &v.wavelet[k], j * sizeof(float));
+				}
+			}
+			aj += 4;
+		}
+	}
+
+	opj_aligned_free(h.wavelet);
+}
+

libopenjpeg/dwt.h

@@ -57,9 +57,9 @@ void dwt_encode(opj_tcd_tilecomp_t * tilec);
 Inverse 5-3 wavelet tranform in 2-D.
 Apply a reversible inverse DWT transform to a component of an image.
 @param tilec Tile component information (current tile)
-@param stop FIXME Number of decoded resolution levels ?
+@param numres Number of resolution levels to decode
 */
-void dwt_decode(opj_tcd_tilecomp_t * tilec, int stop);
+void dwt_decode(opj_tcd_tilecomp_t* tilec, int numres);
 /**
 Get the gain of a subband for the reversible 5-3 DWT.
 @param orient Number that identifies the subband (0->LL, 1->HL, 2->LH, 3->HH)
@@ -83,9 +83,9 @@ void dwt_encode_real(opj_tcd_tilecomp_t * tilec);
 Inverse 9-7 wavelet transform in 2-D. 
 Apply an irreversible inverse DWT transform to a component of an image.
 @param tilec Tile component information (current tile)
-@param stop FIXME Number of decoded resolution levels ?
+@param numres Number of resolution levels to decode
 */
-void dwt_decode_real(opj_tcd_tilecomp_t * tilec, int stop);
+void dwt_decode_real(opj_tcd_tilecomp_t* tilec, int numres);
 /**
 Get the gain of a subband for the irreversible 9-7 DWT.
 @param orient Number that identifies the subband (0->LL, 1->HL, 2->LH, 3->HH)
@@ -100,9 +100,9 @@ Get the norm of a wavelet function of a subband at a specified level for the irr
 */
 double dwt_getnorm_real(int level, int orient);
 /**
-FIXME : comment ???
+Explicit calculation of the Quantization Stepsizes 
-@param tccp
+@param tccp Tile-component coding parameters
-@param prec
+@param prec Precint analyzed
 */
 void dwt_calc_explicit_stepsizes(opj_tccp_t * tccp, int prec);
 /* ----------------------------------------------------------------------- */

libopenjpeg/mct.c

@@ -44,16 +44,20 @@ static const double mct_norms_real[3] = { 1.732, 1.805, 1.573 };
 /* <summary> */
 /* Foward reversible MCT. */
 /* </summary> */
-void mct_encode(int *c0, int *c1, int *c2, int n) {
+void mct_encode(
+		int* restrict c0,
+		int* restrict c1,
+		int* restrict c2,
+		int n)
+{
 	int i;
-	for (i = 0; i < n; i++) {
+	for(i = 0; i < n; ++i) {
-		int r, g, b, y, u, v;
+		int r = c0[i];
-		r = c0[i];
+		int g = c1[i];
-		g = c1[i];
+		int b = c2[i];
-		b = c2[i];
+		int y = (r + (g * 2) + b) >> 2;
-		y = (r + (g << 1) + b) >> 2;
+		int u = b - g;
-		u = b - g;
+		int v = r - g;
-		v = r - g;
 		c0[i] = y;
 		c1[i] = u;
 		c2[i] = v;
@@ -63,16 +67,20 @@ void mct_encode(int *c0, int *c1, int *c2, int n) {
 /* <summary> */
 /* Inverse reversible MCT. */
 /* </summary> */
-void mct_decode(int *c0, int *c1, int *c2, int n) {
+void mct_decode(
+		int* restrict c0,
+		int* restrict c1, 
+		int* restrict c2, 
+		int n)
+{
 	int i;
-	for (i = 0; i < n; i++) {
+	for (i = 0; i < n; ++i) {
-		int y, u, v, r, g, b;
+		int y = c0[i];
-		y = c0[i];
+		int u = c1[i];
-		u = c1[i];
+		int v = c2[i];
-		v = c2[i];
+		int g = y - ((u + v) >> 2);
-		g = y - ((u + v) >> 2);
+		int r = v + g;
-		r = v + g;
+		int b = u + g;
-		b = u + g;
 		c0[i] = r;
 		c1[i] = g;
 		c2[i] = b;
@@ -89,16 +97,20 @@ double mct_getnorm(int compno) {
 /* <summary> */
 /* Foward irreversible MCT. */
 /* </summary> */
-void mct_encode_real(int *c0, int *c1, int *c2, int n) {
+void mct_encode_real(
+		int* restrict c0,
+		int* restrict c1,
+		int* restrict c2,
+		int n)
+{
 	int i;
-	for (i = 0; i < n; i++) {
+	for(i = 0; i < n; ++i) {
-		int r, g, b, y, u, v;
+		int r = c0[i];
-		r = c0[i];
+		int g = c1[i];
-		g = c1[i];
+		int b = c2[i];
-		b = c2[i];
+		int y =  fix_mul(r, 2449) + fix_mul(g, 4809) + fix_mul(b, 934);
-		y = fix_mul(r, 2449) + fix_mul(g, 4809) + fix_mul(b, 934);
+		int u = -fix_mul(r, 1382) - fix_mul(g, 2714) + fix_mul(b, 4096);
-		u = -fix_mul(r, 1382) - fix_mul(g, 2714) + fix_mul(b, 4096);
+		int v =  fix_mul(r, 4096) - fix_mul(g, 3430) - fix_mul(b, 666);
-		v = fix_mul(r, 4096) - fix_mul(g, 3430) - fix_mul(b, 666);
 		c0[i] = y;
 		c1[i] = u;
 		c2[i] = v;
@@ -108,16 +120,20 @@ void mct_encode_real(int *c0, int *c1, int *c2, int n) {
 /* <summary> */
 /* Inverse irreversible MCT. */
 /* </summary> */
-void mct_decode_real(int *c0, int *c1, int *c2, int n) {
+void mct_decode_real(
+		float* restrict c0,
+		float* restrict c1,
+		float* restrict c2,
+		int n)
+{
 	int i;
-	for (i = 0; i < n; i++) {
+	for(i = 0; i < n; ++i) {
-		int y, u, v, r, g, b;
+		float y = c0[i];
-		y = c0[i];
+		float u = c1[i];
-		u = c1[i];
+		float v = c2[i];
-		v = c2[i];
+		float r = y + (v * 1.402f);
-		r = y + fix_mul(v, 11485);
+		float g = y - (u * 0.34413f) - (v * (0.71414f));
-		g = y - fix_mul(u, 2819) - fix_mul(v, 5850);
+		float b = y + (u * 1.772f);
-		b = y + fix_mul(u, 14516);
 		c0[i] = r;
 		c1[i] = g;
 		c2[i] = b;

libopenjpeg/mct.h

@@ -83,7 +83,7 @@ Apply an irreversible multi-component inverse transform to an image
 @param c2 Samples for blue chrominance component
 @param n Number of samples for each component
 */
-void mct_decode_real(int *c0, int *c1, int *c2, int n);
+void mct_decode_real(float* c0, float* c1, float* c2, int n);
 /**
 Get norm of the basis function used for the irreversible multi-component transform
 @param compno Number of the component (0->Y, 1->U, 2->V)

libopenjpeg/opj_includes.h

@@ -76,6 +76,30 @@ Most compilers implement their own version of this keyword ...
 	#endif /* defined(<Compiler>) */
 #endif /* INLINE */
 
+/* Are restricted pointers available? (C99) */
+#if (__STDC_VERSION__ != 199901L)
+	/* Not a C99 compiler */
+	#ifdef __GNUC__
+		#define restrict __restrict__
+	#else
+		#define restrict /* restrict */
+	#endif
+#endif
+
+/* MSVC does not have lrintf */
+#ifdef _MSC_VER
+static INLINE long lrintf(float f){
+	int i;
+
+	_asm{
+		fld f
+		fistp i
+	};
+
+	return i;
+}
+#endif
+
 #include "j2k_lib.h"
 #include "opj_malloc.h"
 #include "event.h"

libopenjpeg/t1.c

@@ -802,22 +802,20 @@ static void t1_encode_cblk(
 		int numcomps,
 		opj_tcd_tile_t * tile)
 {
-	int i, j;
-	int passno;
-	int bpno, passtype;
-	int max;
-	int nmsedec = 0;
 	double cumwmsedec = 0.0;
-	char type = T1_TYPE_MQ;
-	double tempwmsedec;
 
 	opj_mqc_t *mqc = t1->mqc;	/* MQC component */
 
+	int passno, bpno, passtype;
+	int nmsedec = 0;
+	int i, max;
+	char type = T1_TYPE_MQ;
+	double tempwmsedec;
+
 	max = 0;
-	for (j = 0; j < t1->h; ++j) {
+	for (i = 0; i < t1->w * t1->h; ++i) {
-		for (i = 0; i < t1->w; ++i) {
+		int tmp = abs(t1->data[i]);
-			max = int_max(max, int_abs(t1->data[(j * t1->w) + i]));
+		max = int_max(max, tmp);
-		}
 	}
 
 	cblk->numbps = max ? (int_floorlog2(max) + 1) - T1_NMSEDEC_FRACBITS : 0;
@@ -932,13 +930,13 @@ static void t1_decode_cblk(
 		int roishift,
 		int cblksty)
 {
+	opj_raw_t *raw = t1->raw;	/* RAW component */
+	opj_mqc_t *mqc = t1->mqc;	/* MQC component */
+
 	int bpno, passtype;
 	int segno, passno;
 	char type = T1_TYPE_MQ; /* BYPASS mode */
 
-	opj_raw_t *raw = t1->raw;	/* RAW component */
-	opj_mqc_t *mqc = t1->mqc;	/* MQC component */
-
 	if(!allocate_buffers(
 				t1,
 				cblk->x1 - cblk->x0,
@@ -1034,23 +1032,29 @@ void t1_encode_cblks(
 	tile->distotile = 0;		/* fixed_quality */
 
 	for (compno = 0; compno < tile->numcomps; ++compno) {
-		opj_tcd_tilecomp_t *tilec = &tile->comps[compno];
+		opj_tcd_tilecomp_t* tilec = &tile->comps[compno];
+		opj_tccp_t* tccp = &tcp->tccps[compno];
+		int tile_w = tilec->x1 - tilec->x0;
 
 		for (resno = 0; resno < tilec->numresolutions; ++resno) {
 			opj_tcd_resolution_t *res = &tilec->resolutions[resno];
 
 			for (bandno = 0; bandno < res->numbands; ++bandno) {
-				opj_tcd_band_t *band = &res->bands[bandno];
+				opj_tcd_band_t* restrict band = &res->bands[bandno];
 
 				for (precno = 0; precno < res->pw * res->ph; ++precno) {
 					opj_tcd_precinct_t *prc = &band->precincts[precno];
 
 					for (cblkno = 0; cblkno < prc->cw * prc->ch; ++cblkno) {
-						int x, y, w, i, j;
 						opj_tcd_cblk_t *cblk = &prc->cblks[cblkno];
+						int* restrict datap;
+						int* restrict tiledp;
+						int cblk_w;
+						int cblk_h;
+						int i, j;
 
-						x = cblk->x0 - band->x0;
+						int x = cblk->x0 - band->x0;
-						y = cblk->y0 - band->y0;
+						int y = cblk->y0 - band->y0;
 						if (band->bandno & 1) {
 							opj_tcd_resolution_t *pres = &tilec->resolutions[resno - 1];
 							x += pres->x1 - pres->x0;
@@ -1068,20 +1072,25 @@ void t1_encode_cblks(
 							return;
 						}
 
-						w = tilec->x1 - tilec->x0;
+						datap=t1->data;
-						if (tcp->tccps[compno].qmfbid == 1) {
+						cblk_w = t1->w;
-							for (j = 0; j < t1->h; ++j) {
+						cblk_h = t1->h;
-								for (i = 0; i < t1->w; ++i) {
+
-									t1->data[(j * t1->w) + i] =
+						tiledp=&tilec->data[(y * tile_w) + x];
-										tilec->data[(x + i) + (y + j) * w] << T1_NMSEDEC_FRACBITS;
+						if (tccp->qmfbid == 1) {
+							for (j = 0; j < cblk_h; ++j) {
+								for (i = 0; i < cblk_w; ++i) {
+									int tmp = tiledp[(j * tile_w) + i];
+									datap[(j * cblk_w) + i] = tmp << T1_NMSEDEC_FRACBITS;
 								}
 							}
-						} else {		/* if (tcp->tccps[compno].qmfbid == 0) */
+						} else {		/* if (tccp->qmfbid == 0) */
-							for (j = 0; j < t1->h; ++j) {
+							for (j = 0; j < cblk_h; ++j) {
-								for (i = 0; i < t1->w; ++i) {
+								for (i = 0; i < cblk_w; ++i) {
-									t1->data[(j * t1->w) + i] = 
+									int tmp = tiledp[(j * tile_w) + i];
+									datap[(j * cblk_w) + i] =
 										fix_mul(
-										tilec->data[x + i + (y + j) * w],
+										tmp,
 										8192 * 8192 / ((int) floor(band->stepsize * 8192))) >> (11 - T1_NMSEDEC_FRACBITS);
 								}
 							}
@@ -1093,9 +1102,9 @@ void t1_encode_cblks(
 								band->bandno,
 								compno,
 								tilec->numresolutions - 1 - resno,
-								tcp->tccps[compno].qmfbid,
+								tccp->qmfbid,
 								band->stepsize,
-								tcp->tccps[compno].cblksty,
+								tccp->cblksty,
 								tile->numcomps,
 								tile);
 
@@ -1107,80 +1116,80 @@ void t1_encode_cblks(
 }
 
 void t1_decode_cblks(
-		opj_t1_t *t1,
+		opj_t1_t* t1,
-		opj_tcd_tile_t *tile,
+		opj_tcd_tilecomp_t* tilec,
-		opj_tcp_t *tcp)
+		opj_tccp_t* tccp)
 {
-	int compno, resno, bandno, precno, cblkno;
+	int resno, bandno, precno, cblkno;
 
-	for (compno = 0; compno < tile->numcomps; ++compno) {
+	int tile_w = tilec->x1 - tilec->x0;
-		opj_tcd_tilecomp_t *tilec = &tile->comps[compno];
 
 	for (resno = 0; resno < tilec->numresolutions; ++resno) {
-			opj_tcd_resolution_t *res = &tilec->resolutions[resno];
+		opj_tcd_resolution_t* res = &tilec->resolutions[resno];
 
 		for (bandno = 0; bandno < res->numbands; ++bandno) {
-				opj_tcd_band_t *band = &res->bands[bandno];
+			opj_tcd_band_t* restrict band = &res->bands[bandno];
 
 			for (precno = 0; precno < res->pw * res->ph; ++precno) {
-					opj_tcd_precinct_t *prc = &band->precincts[precno];
+				opj_tcd_precinct_t* prc = &band->precincts[precno];
 
 				for (cblkno = 0; cblkno < prc->cw * prc->ch; ++cblkno) {
-						int x, y, w, i, j, cblk_w, cblk_h;
+					opj_tcd_cblk_t* cblk = &prc->cblks[cblkno];
-						opj_tcd_cblk_t *cblk = &prc->cblks[cblkno];
+					int* restrict datap;
+					void* restrict tiledp;
+					int cblk_w, cblk_h;
+					int x, y;
+					int i, j;
 
 					t1_decode_cblk(
 							t1,
 							cblk,
 							band->bandno,
-								tcp->tccps[compno].roishift,
+							tccp->roishift,
-								tcp->tccps[compno].cblksty);
+							tccp->cblksty);
 
 					x = cblk->x0 - band->x0;
 					y = cblk->y0 - band->y0;
 					if (band->bandno & 1) {
-							opj_tcd_resolution_t *pres = &tilec->resolutions[resno - 1];
+						opj_tcd_resolution_t* pres = &tilec->resolutions[resno - 1];
 						x += pres->x1 - pres->x0;
 					}
 					if (band->bandno & 2) {
-							opj_tcd_resolution_t *pres = &tilec->resolutions[resno - 1];
+						opj_tcd_resolution_t* pres = &tilec->resolutions[resno - 1];
 						y += pres->y1 - pres->y0;
 					}
 
-						cblk_w = cblk->x1 - cblk->x0;
+					datap=t1->data;
-						cblk_h = cblk->y1 - cblk->y0;
+					cblk_w = t1->w;
+					cblk_h = t1->h;
 
-						if (tcp->tccps[compno].roishift) {
+					if (tccp->roishift) {
-							int thresh = 1 << tcp->tccps[compno].roishift;
+						int thresh = 1 << tccp->roishift;
 						for (j = 0; j < cblk_h; ++j) {
 							for (i = 0; i < cblk_w; ++i) {
-									int val = t1->data[(j * t1->w) + i];
+								int val = datap[(j * cblk_w) + i];
-									int mag = int_abs(val);
+								int mag = abs(val);
 								if (mag >= thresh) {
-										mag >>= tcp->tccps[compno].roishift;
+									mag >>= tccp->roishift;
-										t1->data[(j * t1->w) + i] = val < 0 ? -mag : mag;
+									datap[(j * cblk_w) + i] = val < 0 ? -mag : mag;
 								}
 							}
 						}
 					}
 
-						w = tilec->x1 - tilec->x0;
+					tiledp=(void*)&tilec->data[(y * tile_w) + x];
-						if (tcp->tccps[compno].qmfbid == 1) {
+					if (tccp->qmfbid == 1) {
 						for (j = 0; j < cblk_h; ++j) {
 							for (i = 0; i < cblk_w; ++i) {
-									tilec->data[x + i + (y + j) * w] = t1->data[(j * t1->w) + i]/2;
+								int tmp = datap[(j * cblk_w) + i];
+								((int*)tiledp)[(j * tile_w) + i] = tmp / 2;
 							}
 						}
-						} else {		/* if (tcp->tccps[compno].qmfbid == 0) */
+					} else {		/* if (tccp->qmfbid == 0) */
 						for (j = 0; j < cblk_h; ++j) {
 							for (i = 0; i < cblk_w; ++i) {
-									if (t1->data[(j * t1->w) + i] >> 1 == 0) {
+								float tmp = datap[(j * cblk_w) + i] * band->stepsize;
-										tilec->data[x + i + (y + j) * w] = 0;
+								((float*)tiledp)[(j * tile_w) + i] = tmp;
-									} else {
-										double tmp = (double)(t1->data[(j * t1->w) + i] * band->stepsize * 4096.0);
-										int tmp2 = ((int) (floor(fabs(tmp)))) + ((int) floor(fabs(tmp*2))%2);									
-										tilec->data[x + i + (y + j) * w] = ((tmp<0)?-tmp2:tmp2);
-									}
 							}
 						}
 					}
@@ -1188,6 +1197,5 @@ void t1_decode_cblks(
 			} /* precno */
 		} /* bandno */
 	} /* resno */
-	} /* compno */
 }
 

libopenjpeg/t1.h

@@ -138,7 +138,7 @@ Decode the code-blocks of a tile
 @param tile The tile to decode
 @param tcp Tile coding parameters
 */
-void t1_decode_cblks(opj_t1_t *t1, opj_tcd_tile_t *tile, opj_tcp_t *tcp);
+void t1_decode_cblks(opj_t1_t* t1, opj_tcd_tilecomp_t* tilec, opj_tccp_t* tccp);
 /* ----------------------------------------------------------------------- */
 /*@}*/
 

libopenjpeg/tcd.c

@@ -671,7 +671,8 @@ void tcd_malloc_decode_tile(opj_tcd_t *tcd, opj_image_t * image, opj_cp_t * cp,
 		tilec->x1 = int_ceildiv(tile->x1, image->comps[compno].dx);
 		tilec->y1 = int_ceildiv(tile->y1, image->comps[compno].dy);
 
-		tilec->data = (int*) opj_aligned_malloc((tilec->x1 - tilec->x0) * (tilec->y1 - tilec->y0) * sizeof(int));
+		/* The +3 is headroom required by the vectorized DWT */
+		tilec->data = (int*) opj_aligned_malloc((((tilec->x1 - tilec->x0) * (tilec->y1 - tilec->y0))+3) * sizeof(int));
 		tilec->numresolutions = tccp->numresolutions;
 		tilec->resolutions = (opj_tcd_resolution_t *) opj_malloc(tilec->numresolutions * sizeof(opj_tcd_resolution_t));
 		
@@ -756,7 +757,7 @@ void tcd_malloc_decode_tile(opj_tcd_t *tcd, opj_image_t * image, opj_cp_t * cp,
 				ss = &tccp->stepsizes[resno == 0 ? 0 : 3 * (resno - 1) + bandno + 1];
 				gain = tccp->qmfbid == 0 ? dwt_getgain_real(band->bandno) : dwt_getgain(band->bandno);
 				numbps = image->comps[compno].prec + gain;
-				band->stepsize = (float)((1.0 + ss->mant / 2048.0) * pow(2.0, numbps - ss->expn));
+				band->stepsize = (float)(((1.0 + ss->mant / 2048.0) * pow(2.0, numbps - ss->expn)) * 0.5);
 				band->numbps = ss->expn + tccp->numgbits - 1;	/* WHY -1 ? */
 				
 				band->precincts = (opj_tcd_precinct_t *) opj_malloc(res->pw * res->ph * sizeof(opj_tcd_precinct_t));
@@ -1350,7 +1351,9 @@ bool tcd_decode_tile(opj_tcd_t *tcd, unsigned char *src, int len, int tileno, op
 	
 	t1_time = opj_clock();	/* time needed to decode a tile */
 	t1 = t1_create(tcd->cinfo);
-	t1_decode_cblks(t1, tile, tcd->tcp);
+	for (compno = 0; compno < tile->numcomps; ++compno) {
+		t1_decode_cblks(t1, &tile->comps[compno], &tcd->tcp->tccps[compno]);
+	}
 	t1_destroy(t1);
 	t1_time = opj_clock() - t1_time;
 	opj_event_msg(tcd->cinfo, EVT_INFO, "- tiers-1 took %f s\n", t1_time);
@@ -1360,6 +1363,8 @@ bool tcd_decode_tile(opj_tcd_t *tcd, unsigned char *src, int len, int tileno, op
 	dwt_time = opj_clock();	/* time needed to decode a tile */
 	for (compno = 0; compno < tile->numcomps; compno++) {
 		opj_tcd_tilecomp_t *tilec = &tile->comps[compno];
+		int numres2decode;
+
 		if (tcd->cp->reduce != 0) {
 			tcd->image->comps[compno].resno_decoded =
 				tile->comps[compno].numresolutions - tcd->cp->reduce - 1;
@@ -1370,12 +1375,14 @@ bool tcd_decode_tile(opj_tcd_t *tcd, unsigned char *src, int len, int tileno, op
 			}
 		}
 
+		numres2decode = tcd->image->comps[compno].resno_decoded + 1;
+		if(numres2decode > 0){
 			if (tcd->tcp->tccps[compno].qmfbid == 1) {
-			dwt_decode(tilec, tilec->numresolutions - 1 - tcd->image->comps[compno].resno_decoded);
+				dwt_decode(tilec, numres2decode);
 			} else {
-			dwt_decode_real(tilec, tilec->numresolutions - 1 - tcd->image->comps[compno].resno_decoded);
+				dwt_decode_real(tilec, numres2decode);
+			}
 		}
-
 	}
 	dwt_time = opj_clock() - dwt_time;
 	opj_event_msg(tcd->cinfo, EVT_INFO, "- dwt took %f s\n", dwt_time);
@@ -1383,48 +1390,55 @@ bool tcd_decode_tile(opj_tcd_t *tcd, unsigned char *src, int len, int tileno, op
 	/*----------------MCT-------------------*/
 
 	if (tcd->tcp->mct) {
+		int n = (tile->comps[0].x1 - tile->comps[0].x0) * (tile->comps[0].y1 - tile->comps[0].y0);
 		if (tcd->tcp->tccps[0].qmfbid == 1) {
-			mct_decode(tile->comps[0].data, tile->comps[1].data, tile->comps[2].data, 
+			mct_decode(
-				(tile->comps[0].x1 - tile->comps[0].x0) * (tile->comps[0].y1 - tile->comps[0].y0));
+					tile->comps[0].data,
+					tile->comps[1].data,
+					tile->comps[2].data, 
+					n);
 		} else {
-			mct_decode_real(tile->comps[0].data, tile->comps[1].data, tile->comps[2].data, 
+			mct_decode_real(
-				(tile->comps[0].x1 - tile->comps[0].x0) * (tile->comps[0].y1 - tile->comps[0].y0));
+					(float*)tile->comps[0].data,
+					(float*)tile->comps[1].data,
+					(float*)tile->comps[2].data, 
+					n);
 		}
 	}
 
-	
 	/*---------------TILE-------------------*/
 
-	for (compno = 0; compno < tile->numcomps; compno++) {
+	for (compno = 0; compno < tile->numcomps; ++compno) {
-		opj_tcd_tilecomp_t *tilec = &tile->comps[compno];
+		opj_tcd_tilecomp_t* tilec = &tile->comps[compno];
-		opj_tcd_resolution_t *res =	&tilec->resolutions[tcd->image->comps[compno].resno_decoded];
+		opj_image_comp_t* imagec = &tcd->image->comps[compno];
-		int adjust = tcd->image->comps[compno].sgnd ? 0 : 1 << (tcd->image->comps[compno].prec - 1);
+		opj_tcd_resolution_t* res = &tilec->resolutions[imagec->resno_decoded];
-		int min = tcd->image->comps[compno].sgnd ? 
+		int adjust = imagec->sgnd ? 0 : 1 << (imagec->prec - 1);
-			-(1 << (tcd->image->comps[compno].prec - 1)) : 0;
+		int min = imagec->sgnd ? -(1 << (imagec->prec - 1)) : 0;
-		int max = tcd->image->comps[compno].sgnd ? 
+		int max = imagec->sgnd ?  (1 << (imagec->prec - 1)) - 1 : (1 << imagec->prec) - 1;
-			(1 << (tcd->image->comps[compno].prec - 1)) - 1 : (1 << tcd->image->comps[compno].prec) - 1;
 
 		int tw = tilec->x1 - tilec->x0;
-		int w = tcd->image->comps[compno].w;
+		int w = imagec->w;
+
+		int offset_x = int_ceildivpow2(imagec->x0, imagec->factor);
+		int offset_y = int_ceildivpow2(imagec->y0, imagec->factor);
 
 		int i, j;
-		int offset_x = int_ceildivpow2(tcd->image->comps[compno].x0, tcd->image->comps[compno].factor);
+		if(tcd->tcp->tccps[compno].qmfbid == 1) {
-		int offset_y = int_ceildivpow2(tcd->image->comps[compno].y0, tcd->image->comps[compno].factor);
+			for(j = res->y0; j < res->y1; ++j) {
-		
+				for(i = res->x0; i < res->x1; ++i) {
-		for (j = res->y0; j < res->y1; j++) {
+					int v = tilec->data[i - res->x0 + (j - res->y0) * tw];
-			for (i = res->x0; i < res->x1; i++) {
-				int v;
-				float tmp = (float)((tilec->data[i - res->x0 + (j - res->y0) * tw]) / 8192.0);
-
-				if (tcd->tcp->tccps[compno].qmfbid == 1) {
-					v = tilec->data[i - res->x0 + (j - res->y0) * tw];
-				} else {
-					int tmp2 = ((int) (floor(fabs(tmp)))) + ((int) floor(fabs(tmp*2))%2);
-					v = ((tmp < 0) ? -tmp2:tmp2);
-				}
 					v += adjust;
-				
+					imagec->data[(i - offset_x) + (j - offset_y) * w] = int_clamp(v, min, max);
-				tcd->image->comps[compno].data[(i - offset_x) + (j - offset_y) * w] = int_clamp(v, min, max);
+				}
+			}
+		}else{
+			for(j = res->y0; j < res->y1; ++j) {
+				for(i = res->x0; i < res->x1; ++i) {
+					float tmp = ((float*)tilec->data)[i - res->x0 + (j - res->y0) * tw];
+					int v = lrintf(tmp);
+					v += adjust;
+					imagec->data[(i - offset_x) + (j - offset_y) * w] = int_clamp(v, min, max);
+				}
 			}
 		}
 	}
@@ -1475,8 +1489,3 @@ void tcd_free_decode_tile(opj_tcd_t *tcd, int tileno) {
 	if (tile->comps != NULL) opj_free(tile->comps);
 }
 
-
-
-
-
-