1/*
2    SDL - Simple DirectMedia Layer
3    Copyright (C) 1997-2012 Sam Lantinga
4
5    This library is free software; you can redistribute it and/or
6    modify it under the terms of the GNU Lesser General Public
7    License as published by the Free Software Foundation; either
8    version 2.1 of the License, or (at your option) any later version.
9
10    This library is distributed in the hope that it will be useful,
11    but WITHOUT ANY WARRANTY; without even the implied warranty of
12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
13    Lesser General Public License for more details.
14
15    You should have received a copy of the GNU Lesser General Public
16    License along with this library; if not, write to the Free Software
17    Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
18
19    Sam Lantinga
20    slouken@libsdl.org
21*/
22#include "SDL_config.h"
23
24#include "SDL_mixer_MMX_VC.h"
25
26#if defined(SDL_BUGGY_MMX_MIXERS) /* buggy, so we're disabling them. --ryan. */
27#if ((defined(_MSC_VER) && defined(_M_IX86)) || defined(__WATCOMC__)) && defined(SDL_ASSEMBLY_ROUTINES)
28// MMX assembler version of SDL_MixAudio for signed little endian 16 bit samples and signed 8 bit samples
29// Copyright 2002 Stephane Marchesin (stephane.marchesin@wanadoo.fr)
30// Converted to Intel ASM notation by Cth
31// This code is licensed under the LGPL (see COPYING for details)
32//
33// Assumes buffer size in bytes is a multiple of 16
34// Assumes SDL_MIX_MAXVOLUME = 128
35
36
37////////////////////////////////////////////////
38// Mixing for 16 bit signed buffers
39////////////////////////////////////////////////
40
41void SDL_MixAudio_MMX_S16_VC(char* dst,char* src,unsigned int nSize,int volume)
42{
43	__asm
44	{
45
46		push	edi
47		push	esi
48		push	ebx
49
50		mov		edi, dst		// edi = dst
51		mov		esi, src		// esi = src
52		mov		eax, volume		// eax = volume
53		mov		ebx, nSize		// ebx = size
54		shr		ebx, 4			// process 16 bytes per iteration = 8 samples
55		jz		endS16
56
57		pxor	mm0, mm0
58		movd	mm0, eax		//%%eax,%%mm0
59		movq	mm1, mm0		//%%mm0,%%mm1
60		psllq	mm0, 16			//$16,%%mm0
61		por		mm0, mm1		//%%mm1,%%mm0
62		psllq	mm0, 16			//$16,%%mm0
63		por		mm0, mm1		//%%mm1,%%mm0
64		psllq	mm0, 16			//$16,%%mm0
65		por		mm0, mm1		//%%mm1,%%mm0			// mm0 = vol|vol|vol|vol
66
67		#ifndef __WATCOMC__
68		align	16
69		#endif
70mixloopS16:
71		movq	mm1, [esi]		//(%%esi),%%mm1\n" // mm1 = a|b|c|d
72		movq	mm2, mm1		//%%mm1,%%mm2\n" // mm2 = a|b|c|d
73		movq	mm4, [esi + 8]	//8(%%esi),%%mm4\n" // mm4 = e|f|g|h
74		// pre charger le buffer dst dans mm7
75		movq	mm7, [edi]		//(%%edi),%%mm7\n" // mm7 = dst[0]"
76		// multiplier par le volume
77		pmullw	mm1, mm0		//%%mm0,%%mm1\n" // mm1 = l(a*v)|l(b*v)|l(c*v)|l(d*v)
78		pmulhw	mm2, mm0		//%%mm0,%%mm2\n" // mm2 = h(a*v)|h(b*v)|h(c*v)|h(d*v)
79		movq	mm5, mm4		//%%mm4,%%mm5\n" // mm5 = e|f|g|h
80		pmullw	mm4, mm0		//%%mm0,%%mm4\n" // mm4 = l(e*v)|l(f*v)|l(g*v)|l(h*v)
81		pmulhw	mm5, mm0		//%%mm0,%%mm5\n" // mm5 = h(e*v)|h(f*v)|h(g*v)|h(h*v)
82		movq	mm3, mm1		//%%mm1,%%mm3\n" // mm3 = l(a*v)|l(b*v)|l(c*v)|l(d*v)
83		punpckhwd	mm1, mm2	//%%mm2,%%mm1\n" // mm1 = a*v|b*v
84		movq		mm6, mm4	//%%mm4,%%mm6\n" // mm6 = l(e*v)|l(f*v)|l(g*v)|l(h*v)
85		punpcklwd	mm3, mm2	//%%mm2,%%mm3\n" // mm3 = c*v|d*v
86		punpckhwd	mm4, mm5	//%%mm5,%%mm4\n" // mm4 = e*f|f*v
87		punpcklwd	mm6, mm5	//%%mm5,%%mm6\n" // mm6 = g*v|h*v
88		// pre charger le buffer dst dans mm5
89		movq	mm5, [edi + 8]	//8(%%edi),%%mm5\n" // mm5 = dst[1]
90		// diviser par 128
91		psrad	mm1, 7			//$7,%%mm1\n" // mm1 = a*v/128|b*v/128 , 128 = SDL_MIX_MAXVOLUME
92		add		esi, 16			//$16,%%esi\n"
93		psrad	mm3, 7			//$7,%%mm3\n" // mm3 = c*v/128|d*v/128
94		psrad	mm4, 7			//$7,%%mm4\n" // mm4 = e*v/128|f*v/128
95		// mm1 = le sample avec le volume modifie
96		packssdw	mm3, mm1	//%%mm1,%%mm3\n" // mm3 = s(a*v|b*v|c*v|d*v)
97		psrad	mm6, 7			//$7,%%mm6\n" // mm6= g*v/128|h*v/128
98		paddsw	mm3, mm7		//%%mm7,%%mm3\n" // mm3 = adjust_volume(src)+dst
99		// mm4 = le sample avec le volume modifie
100		packssdw	mm6, mm4	//%%mm4,%%mm6\n" // mm6 = s(e*v|f*v|g*v|h*v)
101		movq	[edi], mm3		//%%mm3,(%%edi)\n"
102		paddsw	mm6, mm5		//%%mm5,%%mm6\n" // mm6 = adjust_volume(src)+dst
103		movq	[edi + 8], mm6	//%%mm6,8(%%edi)\n"
104		add		edi, 16			//$16,%%edi\n"
105		dec		ebx				//%%ebx\n"
106		jnz mixloopS16
107
108endS16:
109		emms
110
111		pop		ebx
112		pop		esi
113		pop		edi
114	}
115
116}
117
118////////////////////////////////////////////////
119// Mixing for 8 bit signed buffers
120////////////////////////////////////////////////
121
122void SDL_MixAudio_MMX_S8_VC(char* dst,char* src,unsigned int nSize,int volume)
123{
124	_asm
125	{
126
127		push	edi
128		push	esi
129		push	ebx
130
131		mov		edi, dst	//movl	%0,%%edi	// edi = dst
132		mov		esi, src	//%1,%%esi	// esi = src
133		mov		eax, volume	//%3,%%eax	// eax = volume
134
135		movd	mm0, eax	//%%eax,%%mm0
136		movq	mm1, mm0	//%%mm0,%%mm1
137		psllq	mm0, 16		//$16,%%mm0
138		por		mm0, mm1	//%%mm1,%%mm0
139		psllq	mm0, 16		//$16,%%mm0
140		por		mm0, mm1	//%%mm1,%%mm0
141		psllq	mm0, 16		//$16,%%mm0
142		por		mm0, mm1	//%%mm1,%%mm0
143
144		mov		ebx, nSize	//%2,%%ebx	// ebx = size
145		shr		ebx, 3		//$3,%%ebx	// process 8 bytes per iteration = 8 samples
146		cmp		ebx, 0		//$0,%%ebx
147		je		endS8
148
149		#ifndef __WATCOMC__
150		align 16
151		#endif
152mixloopS8:
153		pxor	mm2, mm2	//%%mm2,%%mm2		// mm2 = 0
154		movq	mm1, [esi]	//(%%esi),%%mm1	// mm1 = a|b|c|d|e|f|g|h
155		movq	mm3, mm1	//%%mm1,%%mm3 	// mm3 = a|b|c|d|e|f|g|h
156		// on va faire le "sign extension" en faisant un cmp avec 0 qui retourne 1 si <0, 0 si >0
157		pcmpgtb		mm2, mm1	//%%mm1,%%mm2	// mm2 = 11111111|00000000|00000000....
158		punpckhbw	mm1, mm2	//%%mm2,%%mm1	// mm1 = 0|a|0|b|0|c|0|d
159		punpcklbw	mm3, mm2	//%%mm2,%%mm3	// mm3 = 0|e|0|f|0|g|0|h
160		movq	mm2, [edi]	//(%%edi),%%mm2	// mm2 = destination
161		pmullw	mm1, mm0	//%%mm0,%%mm1	// mm1 = v*a|v*b|v*c|v*d
162		add		esi, 8		//$8,%%esi
163		pmullw	mm3, mm0	//%%mm0,%%mm3	// mm3 = v*e|v*f|v*g|v*h
164		psraw	mm1, 7		//$7,%%mm1		// mm1 = v*a/128|v*b/128|v*c/128|v*d/128
165		psraw	mm3, 7		//$7,%%mm3		// mm3 = v*e/128|v*f/128|v*g/128|v*h/128
166		packsswb mm3, mm1	//%%mm1,%%mm3	// mm1 = v*a/128|v*b/128|v*c/128|v*d/128|v*e/128|v*f/128|v*g/128|v*h/128
167		paddsb	mm3, mm2	//%%mm2,%%mm3	// add to destination buffer
168		movq	[edi], mm3	//%%mm3,(%%edi)	// store back to ram
169		add		edi, 8		//$8,%%edi
170		dec		ebx			//%%ebx
171		jnz		mixloopS8
172
173endS8:
174		emms
175
176		pop		ebx
177		pop		esi
178		pop		edi
179	}
180}
181
182#endif /* SDL_ASSEMBLY_ROUTINES */
183#endif /* SDL_BUGGY_MMX_MIXERS */
184