From c05011c393626ea0a692adc050f4bc8df24d3383 Mon Sep 17 00:00:00 2001 From: twinaphex Date: Sat, 20 Oct 2012 05:53:03 +0200 Subject: [PATCH] No resampler --- Makefile | 4 - mednafen/mednafen.cpp | 20 +- mednafen/psx/spu.cpp | 38 - mednafen/psx/spu.h | 3 - mednafen/resampler/Makefile.am.inc | 1 - mednafen/resampler/arch.h | 239 ------ mednafen/resampler/fixed_generic.h | 106 --- mednafen/resampler/resample.c | 1158 ---------------------------- mednafen/resampler/resample_sse.h | 128 --- mednafen/resampler/resampler.h | 349 --------- mednafen/resampler/stack_alloc.h | 115 --- mednafen/sound/Fir_Resampler.cpp | 199 ----- mednafen/sound/WAVRecord.cpp | 99 --- mednafen/sound/WAVRecord.h | 31 - mednafen/sound/sound.cpp | 41 - 15 files changed, 2 insertions(+), 2529 deletions(-) delete mode 100644 mednafen/resampler/Makefile.am.inc delete mode 100644 mednafen/resampler/arch.h delete mode 100644 mednafen/resampler/fixed_generic.h delete mode 100644 mednafen/resampler/resample.c delete mode 100644 mednafen/resampler/resample_sse.h delete mode 100644 mednafen/resampler/resampler.h delete mode 100644 mednafen/resampler/stack_alloc.h delete mode 100644 mednafen/sound/Fir_Resampler.cpp delete mode 100644 mednafen/sound/WAVRecord.cpp delete mode 100644 mednafen/sound/WAVRecord.h delete mode 100644 mednafen/sound/sound.cpp diff --git a/Makefile b/Makefile index 0d517ff..1f9f267 100644 --- a/Makefile +++ b/Makefile @@ -96,10 +96,7 @@ MEDNAFEN_SOURCES := $(MEDNAFEN_DIR)/cdrom/cdromif.cpp \ $(MEDNAFEN_DIR)/string/escape.cpp \ $(MEDNAFEN_DIR)/string/ConvertUTF.cpp \ $(MEDNAFEN_DIR)/sound/Blip_Buffer.cpp \ - $(MEDNAFEN_DIR)/sound/Fir_Resampler.cpp \ $(MEDNAFEN_DIR)/sound/Stereo_Buffer.cpp \ - $(MEDNAFEN_DIR)/sound/WAVRecord.cpp \ - $(MEDNAFEN_DIR)/sound/sound.cpp \ $(MEDNAFEN_DIR)/file.cpp \ $(MEDNAFEN_DIR)/okiadpcm.cpp \ $(MEDNAFEN_DIR)/md5.cpp @@ -116,7 +113,6 @@ SOURCES_C := $(MEDNAFEN_DIR)/trio/trio.c \ $(MEDNAFEN_DIR)/trio/trionan.c \ $(MEDNAFEN_DIR)/trio/triostr.c \ $(MEDNAFEN_DIR)/string/world_strtod.c \ - $(MEDNAFEN_DIR)/resampler/resample.c \ $(MEDNAFEN_DIR)/compress/blz.c \ $(MEDNAFEN_DIR)/compress/unzip.c \ $(MEDNAFEN_DIR)/compress/minilzo.c \ diff --git a/mednafen/mednafen.cpp b/mednafen/mednafen.cpp index 0dd8a54..8ccf9c0 100644 --- a/mednafen/mednafen.cpp +++ b/mednafen/mednafen.cpp @@ -33,13 +33,12 @@ #include "video.h" #include "video/Deinterlacer.h" #include "file.h" -#include "sound/WAVRecord.h" +#include "FileWrapper.h" #include "cdrom/cdromif.h" #include "mempatcher.h" #include "compress/minilzo.h" #include "md5.h" #include "clamp.h" -#include "Fir_Resampler.h" #include "string/escape.h" @@ -106,7 +105,6 @@ static MDFNSetting RenamedSettings[] = MDFNGI *MDFNGameInfo = NULL; -static Fir_Resampler<16> ff_resampler; static double LastSoundMultiplier; static MDFN_PixelFormat last_pixel_format; @@ -967,23 +965,9 @@ void MDFNI_Emulate(EmulateSpecStruct *espec) { espec->SoundFormatChanged = true; last_sound_rate = espec->SoundRate; - - ff_resampler.buffer_size((espec->SoundRate / 2) * 2); } - if(espec->NeedRewind) - { - if(MDFNGameInfo->GameType == GMT_PLAYER) - { - espec->NeedRewind = 0; - MDFN_DispMessage(_("Music player rewinding is unsupported.")); - } - } - - // Don't even save states with state rewinding if netplay is enabled, it will degrade netplay performance, and can cause - // desynchs with some emulation(IE SNES based on bsnes). - - espec->NeedSoundReverse = false; + espec->NeedSoundReverse = false; MDFNGameInfo->Emulate(espec); diff --git a/mednafen/psx/spu.cpp b/mednafen/psx/spu.cpp index 946f5e1..4eaa54f 100644 --- a/mednafen/psx/spu.cpp +++ b/mednafen/psx/spu.cpp @@ -96,8 +96,6 @@ PS_SPU::PS_SPU() IntermediateBufferPos = 0; memset(IntermediateBuffer, 0, sizeof(IntermediateBuffer)); - - resampler = NULL; } PS_SPU::~PS_SPU() @@ -880,10 +878,6 @@ int32 PS_SPU::UpdateFromCDC(int32 clocks) IntermediateBuffer[IntermediateBufferPos][1] = output_r; IntermediateBufferPos++; - //resampler.buffer()[0] = (int16)(rand() & 0xFFFF) >> 1; - //resampler.buffer()[1] = (int16)(rand() & 0xFFFF) >> 1; - //resampler.write(2); - sample_clocks--; // Clock global sweep @@ -1179,21 +1173,8 @@ void PS_SPU::StartFrame(double rate, uint32 quality) { if((int)rate != last_rate || quality != last_quality) { - //double ratio = (double)44100 / (rate ? rate : 44100); - //resampler.time_ratio(ratio, 0.9965); int err = 0; - if(resampler) - { - speex_resampler_destroy(resampler); - resampler = NULL; - } - - if((int)rate && (int)rate != 44100) - { - resampler = speex_resampler_init(2, 44100, (int)rate, quality, &err); - } - last_rate = (int)rate; last_quality = quality; } @@ -1213,25 +1194,6 @@ int32 PS_SPU::EndFrame(int16 *SoundBuf) return(ret); } - else if(resampler) - { - spx_uint32_t in_len; - spx_uint32_t out_len; - - in_len = IntermediateBufferPos; - out_len = 524288; //8192; // FIXME, real size. - - speex_resampler_process_interleaved_int(resampler, (const spx_int16_t *)IntermediateBuffer, &in_len, (spx_int16_t *)SoundBuf, &out_len); - - assert(in_len <= IntermediateBufferPos); - - if((IntermediateBufferPos - in_len) > 0) - memmove(IntermediateBuffer, IntermediateBuffer + in_len, (IntermediateBufferPos - in_len) * sizeof(int16) * 2); - - IntermediateBufferPos -= in_len; - - return(out_len); - } else { IntermediateBufferPos = 0; diff --git a/mednafen/psx/spu.h b/mednafen/psx/spu.h index c994d17..fedf44e 100644 --- a/mednafen/psx/spu.h +++ b/mednafen/psx/spu.h @@ -1,8 +1,6 @@ #ifndef __MDFN_PSX_SPU_H #define __MDFN_PSX_SPU_H -#include - namespace MDFN_IEN_PSX { @@ -242,7 +240,6 @@ class PS_SPU int last_rate; uint32 last_quality; - SpeexResamplerState *resampler; // Buffers 44.1KHz samples, should have enough for one video frame(~735 frames NTSC, ~882 PAL) plus jitter plus enough for the resampler leftovers. // We'll just go with 4096 because powers of 2 are AWESOME and such. diff --git a/mednafen/resampler/Makefile.am.inc b/mednafen/resampler/Makefile.am.inc deleted file mode 100644 index 3d01319..0000000 --- a/mednafen/resampler/Makefile.am.inc +++ /dev/null @@ -1 +0,0 @@ -mednafen_SOURCES += resampler/resample.c diff --git a/mednafen/resampler/arch.h b/mednafen/resampler/arch.h deleted file mode 100644 index d38c36c..0000000 --- a/mednafen/resampler/arch.h +++ /dev/null @@ -1,239 +0,0 @@ -/* Copyright (C) 2003 Jean-Marc Valin */ -/** - @file arch.h - @brief Various architecture definitions Speex -*/ -/* - Redistribution and use in source and binary forms, with or without - modification, are permitted provided that the following conditions - are met: - - - Redistributions of source code must retain the above copyright - notice, this list of conditions and the following disclaimer. - - - Redistributions in binary form must reproduce the above copyright - notice, this list of conditions and the following disclaimer in the - documentation and/or other materials provided with the distribution. - - - Neither the name of the Xiph.org Foundation nor the names of its - contributors may be used to endorse or promote products derived from - this software without specific prior written permission. - - THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS - ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT - LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR - A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR - CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, - EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, - PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR - PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF - LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING - NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS - SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -*/ - -#ifndef ARCH_H -#define ARCH_H - -#ifndef SPEEX_VERSION -#define SPEEX_MAJOR_VERSION 1 /**< Major Speex version. */ -#define SPEEX_MINOR_VERSION 1 /**< Minor Speex version. */ -#define SPEEX_MICRO_VERSION 15 /**< Micro Speex version. */ -#define SPEEX_EXTRA_VERSION "" /**< Extra Speex version. */ -#define SPEEX_VERSION "speex-1.2beta3" /**< Speex version string. */ -#endif - -/* A couple test to catch stupid option combinations */ -#ifdef FIXED_POINT - -#ifdef FLOATING_POINT -#error You cannot compile as floating point and fixed point at the same time -#endif -#ifdef _USE_SSE -#error SSE is only for floating-point -#endif -#if ((defined (ARM4_ASM)||defined (ARM4_ASM)) && defined(BFIN_ASM)) || (defined (ARM4_ASM)&&defined(ARM5E_ASM)) -#error Make up your mind. What CPU do you have? -#endif -#ifdef VORBIS_PSYCHO -#error Vorbis-psy model currently not implemented in fixed-point -#endif - -#else - -#ifndef FLOATING_POINT -#error You now need to define either FIXED_POINT or FLOATING_POINT -#endif -#if defined (ARM4_ASM) || defined(ARM5E_ASM) || defined(BFIN_ASM) -#error I suppose you can have a [ARM4/ARM5E/Blackfin] that has float instructions? -#endif -#ifdef FIXED_POINT_DEBUG -#error "Don't you think enabling fixed-point is a good thing to do if you want to debug that?" -#endif - - -#endif - -#ifndef OUTSIDE_SPEEX -#include "speex/speex_types.h" -#endif - -#define ABS(x) ((x) < 0 ? (-(x)) : (x)) /**< Absolute integer value. */ -#define ABS16(x) ((x) < 0 ? (-(x)) : (x)) /**< Absolute 16-bit value. */ -#define MIN16(a,b) ((a) < (b) ? (a) : (b)) /**< Maximum 16-bit value. */ -#define MAX16(a,b) ((a) > (b) ? (a) : (b)) /**< Maximum 16-bit value. */ -#define ABS32(x) ((x) < 0 ? (-(x)) : (x)) /**< Absolute 32-bit value. */ -#define MIN32(a,b) ((a) < (b) ? (a) : (b)) /**< Maximum 32-bit value. */ -#define MAX32(a,b) ((a) > (b) ? (a) : (b)) /**< Maximum 32-bit value. */ - -#ifdef FIXED_POINT - -typedef spx_int16_t spx_word16_t; -typedef spx_int32_t spx_word32_t; -typedef spx_word32_t spx_mem_t; -typedef spx_word16_t spx_coef_t; -typedef spx_word16_t spx_lsp_t; -typedef spx_word32_t spx_sig_t; - -#define Q15ONE 32767 - -#define LPC_SCALING 8192 -#define SIG_SCALING 16384 -#define LSP_SCALING 8192. -#define GAMMA_SCALING 32768. -#define GAIN_SCALING 64 -#define GAIN_SCALING_1 0.015625 - -#define LPC_SHIFT 13 -#define LSP_SHIFT 13 -#define SIG_SHIFT 14 -#define GAIN_SHIFT 6 - -#define VERY_SMALL 0 -#define VERY_LARGE32 ((spx_word32_t)2147483647) -#define VERY_LARGE16 ((spx_word16_t)32767) -#define Q15_ONE ((spx_word16_t)32767) - - -#ifdef FIXED_DEBUG -#include "fixed_debug.h" -#else - -#include "fixed_generic.h" - -#ifdef ARM5E_ASM -#include "fixed_arm5e.h" -#elif defined (ARM4_ASM) -#include "fixed_arm4.h" -#elif defined (BFIN_ASM) -#include "fixed_bfin.h" -#endif - -#endif - - -#else - -typedef float spx_mem_t; -typedef float spx_coef_t; -typedef float spx_lsp_t; -typedef float spx_sig_t; -typedef float spx_word16_t; -typedef float spx_word32_t; - -#define Q15ONE 1.0f -#define LPC_SCALING 1.f -#define SIG_SCALING 1.f -#define LSP_SCALING 1.f -#define GAMMA_SCALING 1.f -#define GAIN_SCALING 1.f -#define GAIN_SCALING_1 1.f - - -#define VERY_SMALL 1e-15f -#define VERY_LARGE32 1e15f -#define VERY_LARGE16 1e15f -#define Q15_ONE ((spx_word16_t)1.f) - -#define QCONST16(x,bits) (x) -#define QCONST32(x,bits) (x) - -#define NEG16(x) (-(x)) -#define NEG32(x) (-(x)) -#define EXTRACT16(x) (x) -#define EXTEND32(x) (x) -#define SHR16(a,shift) (a) -#define SHL16(a,shift) (a) -#define SHR32(a,shift) (a) -#define SHL32(a,shift) (a) -#define PSHR16(a,shift) (a) -#define PSHR32(a,shift) (a) -#define VSHR32(a,shift) (a) -#define SATURATE16(x,a) (x) -#define SATURATE32(x,a) (x) - -#define PSHR(a,shift) (a) -#define SHR(a,shift) (a) -#define SHL(a,shift) (a) -#define SATURATE(x,a) (x) - -#define ADD16(a,b) ((a)+(b)) -#define SUB16(a,b) ((a)-(b)) -#define ADD32(a,b) ((a)+(b)) -#define SUB32(a,b) ((a)-(b)) -#define MULT16_16_16(a,b) ((a)*(b)) -#define MULT16_16(a,b) ((spx_word32_t)(a)*(spx_word32_t)(b)) -#define MAC16_16(c,a,b) ((c)+(spx_word32_t)(a)*(spx_word32_t)(b)) - -#define MULT16_32_Q11(a,b) ((a)*(b)) -#define MULT16_32_Q13(a,b) ((a)*(b)) -#define MULT16_32_Q14(a,b) ((a)*(b)) -#define MULT16_32_Q15(a,b) ((a)*(b)) -#define MULT16_32_P15(a,b) ((a)*(b)) - -#define MAC16_32_Q11(c,a,b) ((c)+(a)*(b)) -#define MAC16_32_Q15(c,a,b) ((c)+(a)*(b)) - -#define MAC16_16_Q11(c,a,b) ((c)+(a)*(b)) -#define MAC16_16_Q13(c,a,b) ((c)+(a)*(b)) -#define MAC16_16_P13(c,a,b) ((c)+(a)*(b)) -#define MULT16_16_Q11_32(a,b) ((a)*(b)) -#define MULT16_16_Q13(a,b) ((a)*(b)) -#define MULT16_16_Q14(a,b) ((a)*(b)) -#define MULT16_16_Q15(a,b) ((a)*(b)) -#define MULT16_16_P15(a,b) ((a)*(b)) -#define MULT16_16_P13(a,b) ((a)*(b)) -#define MULT16_16_P14(a,b) ((a)*(b)) - -#define DIV32_16(a,b) (((spx_word32_t)(a))/(spx_word16_t)(b)) -#define PDIV32_16(a,b) (((spx_word32_t)(a))/(spx_word16_t)(b)) -#define DIV32(a,b) (((spx_word32_t)(a))/(spx_word32_t)(b)) -#define PDIV32(a,b) (((spx_word32_t)(a))/(spx_word32_t)(b)) - - -#endif - - -#if defined (CONFIG_TI_C54X) || defined (CONFIG_TI_C55X) - -/* 2 on TI C5x DSP */ -#define BYTES_PER_CHAR 2 -#define BITS_PER_CHAR 16 -#define LOG2_BITS_PER_CHAR 4 - -#else - -#define BYTES_PER_CHAR 1 -#define BITS_PER_CHAR 8 -#define LOG2_BITS_PER_CHAR 3 - -#endif - - - -#ifdef FIXED_DEBUG -extern long long spx_mips; -#endif - - -#endif diff --git a/mednafen/resampler/fixed_generic.h b/mednafen/resampler/fixed_generic.h deleted file mode 100644 index 3fb096e..0000000 --- a/mednafen/resampler/fixed_generic.h +++ /dev/null @@ -1,106 +0,0 @@ -/* Copyright (C) 2003 Jean-Marc Valin */ -/** - @file fixed_generic.h - @brief Generic fixed-point operations -*/ -/* - Redistribution and use in source and binary forms, with or without - modification, are permitted provided that the following conditions - are met: - - - Redistributions of source code must retain the above copyright - notice, this list of conditions and the following disclaimer. - - - Redistributions in binary form must reproduce the above copyright - notice, this list of conditions and the following disclaimer in the - documentation and/or other materials provided with the distribution. - - - Neither the name of the Xiph.org Foundation nor the names of its - contributors may be used to endorse or promote products derived from - this software without specific prior written permission. - - THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS - ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT - LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR - A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR - CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, - EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, - PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR - PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF - LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING - NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS - SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -*/ - -#ifndef FIXED_GENERIC_H -#define FIXED_GENERIC_H - -#define QCONST16(x,bits) ((spx_word16_t)(.5+(x)*(((spx_word32_t)1)<<(bits)))) -#define QCONST32(x,bits) ((spx_word32_t)(.5+(x)*(((spx_word32_t)1)<<(bits)))) - -#define NEG16(x) (-(x)) -#define NEG32(x) (-(x)) -#define EXTRACT16(x) ((spx_word16_t)(x)) -#define EXTEND32(x) ((spx_word32_t)(x)) -#define SHR16(a,shift) ((a) >> (shift)) -#define SHL16(a,shift) ((a) << (shift)) -#define SHR32(a,shift) ((a) >> (shift)) -#define SHL32(a,shift) ((a) << (shift)) -#define PSHR16(a,shift) (SHR16((a)+((1<<((shift))>>1)),shift)) -#define PSHR32(a,shift) (SHR32((a)+((EXTEND32(1)<<((shift))>>1)),shift)) -#define VSHR32(a, shift) (((shift)>0) ? SHR32(a, shift) : SHL32(a, -(shift))) -#define SATURATE16(x,a) (((x)>(a) ? (a) : (x)<-(a) ? -(a) : (x))) -#define SATURATE32(x,a) (((x)>(a) ? (a) : (x)<-(a) ? -(a) : (x))) - -#define SHR(a,shift) ((a) >> (shift)) -#define SHL(a,shift) ((spx_word32_t)(a) << (shift)) -#define PSHR(a,shift) (SHR((a)+((EXTEND32(1)<<((shift))>>1)),shift)) -#define SATURATE(x,a) (((x)>(a) ? (a) : (x)<-(a) ? -(a) : (x))) - - -#define ADD16(a,b) ((spx_word16_t)((spx_word16_t)(a)+(spx_word16_t)(b))) -#define SUB16(a,b) ((spx_word16_t)(a)-(spx_word16_t)(b)) -#define ADD32(a,b) ((spx_word32_t)(a)+(spx_word32_t)(b)) -#define SUB32(a,b) ((spx_word32_t)(a)-(spx_word32_t)(b)) - - -/* result fits in 16 bits */ -#define MULT16_16_16(a,b) ((((spx_word16_t)(a))*((spx_word16_t)(b)))) - -/* (spx_word32_t)(spx_word16_t) gives TI compiler a hint that it's 16x16->32 multiply */ -#define MULT16_16(a,b) (((spx_word32_t)(spx_word16_t)(a))*((spx_word32_t)(spx_word16_t)(b))) - -#define MAC16_16(c,a,b) (ADD32((c),MULT16_16((a),(b)))) -#define MULT16_32_Q12(a,b) ADD32(MULT16_16((a),SHR((b),12)), SHR(MULT16_16((a),((b)&0x00000fff)),12)) -#define MULT16_32_Q13(a,b) ADD32(MULT16_16((a),SHR((b),13)), SHR(MULT16_16((a),((b)&0x00001fff)),13)) -#define MULT16_32_Q14(a,b) ADD32(MULT16_16((a),SHR((b),14)), SHR(MULT16_16((a),((b)&0x00003fff)),14)) - -#define MULT16_32_Q11(a,b) ADD32(MULT16_16((a),SHR((b),11)), SHR(MULT16_16((a),((b)&0x000007ff)),11)) -#define MAC16_32_Q11(c,a,b) ADD32(c,ADD32(MULT16_16((a),SHR((b),11)), SHR(MULT16_16((a),((b)&0x000007ff)),11))) - -#define MULT16_32_P15(a,b) ADD32(MULT16_16((a),SHR((b),15)), PSHR(MULT16_16((a),((b)&0x00007fff)),15)) -#define MULT16_32_Q15(a,b) ADD32(MULT16_16((a),SHR((b),15)), SHR(MULT16_16((a),((b)&0x00007fff)),15)) -#define MAC16_32_Q15(c,a,b) ADD32(c,ADD32(MULT16_16((a),SHR((b),15)), SHR(MULT16_16((a),((b)&0x00007fff)),15))) - - -#define MAC16_16_Q11(c,a,b) (ADD32((c),SHR(MULT16_16((a),(b)),11))) -#define MAC16_16_Q13(c,a,b) (ADD32((c),SHR(MULT16_16((a),(b)),13))) -#define MAC16_16_P13(c,a,b) (ADD32((c),SHR(ADD32(4096,MULT16_16((a),(b))),13))) - -#define MULT16_16_Q11_32(a,b) (SHR(MULT16_16((a),(b)),11)) -#define MULT16_16_Q13(a,b) (SHR(MULT16_16((a),(b)),13)) -#define MULT16_16_Q14(a,b) (SHR(MULT16_16((a),(b)),14)) -#define MULT16_16_Q15(a,b) (SHR(MULT16_16((a),(b)),15)) - -#define MULT16_16_P13(a,b) (SHR(ADD32(4096,MULT16_16((a),(b))),13)) -#define MULT16_16_P14(a,b) (SHR(ADD32(8192,MULT16_16((a),(b))),14)) -#define MULT16_16_P15(a,b) (SHR(ADD32(16384,MULT16_16((a),(b))),15)) - -#define MUL_16_32_R15(a,bh,bl) ADD32(MULT16_16((a),(bh)), SHR(MULT16_16((a),(bl)),15)) - -#define DIV32_16(a,b) ((spx_word16_t)(((spx_word32_t)(a))/((spx_word16_t)(b)))) -#define PDIV32_16(a,b) ((spx_word16_t)(((spx_word32_t)(a)+((spx_word16_t)(b)>>1))/((spx_word16_t)(b)))) -#define DIV32(a,b) (((spx_word32_t)(a))/((spx_word32_t)(b))) -#define PDIV32(a,b) (((spx_word32_t)(a)+((spx_word16_t)(b)>>1))/((spx_word32_t)(b))) - -#endif diff --git a/mednafen/resampler/resample.c b/mednafen/resampler/resample.c deleted file mode 100644 index c47fef5..0000000 --- a/mednafen/resampler/resample.c +++ /dev/null @@ -1,1158 +0,0 @@ -/* Copyright (C) 2007-2008 Jean-Marc Valin - Copyright (C) 2008 Thorvald Natvig - - File: resample.c - Arbitrary resampling code - - Redistribution and use in source and binary forms, with or without - modification, are permitted provided that the following conditions are - met: - - 1. Redistributions of source code must retain the above copyright notice, - this list of conditions and the following disclaimer. - - 2. Redistributions in binary form must reproduce the above copyright - notice, this list of conditions and the following disclaimer in the - documentation and/or other materials provided with the distribution. - - 3. The name of the author may not be used to endorse or promote products - derived from this software without specific prior written permission. - - THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR - IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES - OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE - DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, - INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES - (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR - SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, - STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN - ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE - POSSIBILITY OF SUCH DAMAGE. -*/ - -/* - The design goals of this code are: - - Very fast algorithm - - SIMD-friendly algorithm - - Low memory requirement - - Good *perceptual* quality (and not best SNR) - - Warning: This resampler is relatively new. Although I think I got rid of - all the major bugs and I don't expect the API to change anymore, there - may be something I've missed. So use with caution. - - This algorithm is based on this original resampling algorithm: - Smith, Julius O. Digital Audio Resampling Home Page - Center for Computer Research in Music and Acoustics (CCRMA), - Stanford University, 2007. - Web published at http://www-ccrma.stanford.edu/~jos/resample/. - - There is one main difference, though. This resampler uses cubic - interpolation instead of linear interpolation in the above paper. This - makes the table much smaller and makes it possible to compute that table - on a per-stream basis. In turn, being able to tweak the table for each - stream makes it possible to both reduce complexity on simple ratios - (e.g. 2/3), and get rid of the rounding operations in the inner loop. - The latter both reduces CPU time and makes the algorithm more SIMD-friendly. -*/ - -#ifdef HAVE_CONFIG_H -#include "config.h" -#endif - -/* Begin Mednafen modifications */ - -#ifdef ARCH_X86_64 -#define _USE_SSE -#define _USE_SSE2 -#endif - -#ifdef EXPORT - #undef EXPORT -#endif - -#define EXPORT - -#ifdef FIXED_POINT - #undef FIXED_POINT -#endif - -//#define FIXED_POINT 1 -#define FLOATING_POINT 1 - -#define OUTSIDE_SPEEX - -/* End Mednafen modifications */ - -#ifdef OUTSIDE_SPEEX -#include -static void *speex_alloc (int size) {return calloc(size,1);} -static void *speex_realloc (void *ptr, int size) {return realloc(ptr, size);} -static void speex_free (void *ptr) {free(ptr);} - -/* Begin Mednafen modifications */ -#include "resampler.h" -/* End Mednafen modifications */ -#include "arch.h" -#else /* OUTSIDE_SPEEX */ - -#include "speex/speex_resampler.h" -#include "arch.h" -#include "os_support.h" -#endif /* OUTSIDE_SPEEX */ - -#include "stack_alloc.h" -#include - -#ifndef M_PI -#define M_PI 3.14159263 -#endif - -#ifdef FIXED_POINT -#define WORD2INT(x) ((x) < -32767 ? -32768 : ((x) > 32766 ? 32767 : (x))) -#else -#define WORD2INT(x) ((x) < -32767.5f ? -32768 : ((x) > 32766.5f ? 32767 : floor(.5+(x)))) -#endif - -#define IMAX(a,b) ((a) > (b) ? (a) : (b)) -#define IMIN(a,b) ((a) < (b) ? (a) : (b)) - -#ifndef NULL -#define NULL 0 -#endif - -#ifdef _USE_SSE -#include "resample_sse.h" -#endif - -/* Numer of elements to allocate on the stack */ -#ifdef VAR_ARRAYS -#define FIXED_STACK_ALLOC 8192 -#else -#define FIXED_STACK_ALLOC 1024 -#endif - -typedef int (*resampler_basic_func)(SpeexResamplerState *, spx_uint32_t , const spx_word16_t *, spx_uint32_t *, spx_word16_t *, spx_uint32_t *); - -struct SpeexResamplerState_ { - spx_uint32_t in_rate; - spx_uint32_t out_rate; - spx_uint32_t num_rate; - spx_uint32_t den_rate; - - int quality; - spx_uint32_t nb_channels; - spx_uint32_t filt_len; - spx_uint32_t mem_alloc_size; - spx_uint32_t buffer_size; - int int_advance; - int frac_advance; - float cutoff; - spx_uint32_t oversample; - int initialised; - int started; - - /* These are per-channel */ - spx_int32_t *last_sample; - spx_uint32_t *samp_frac_num; - spx_uint32_t *magic_samples; - - spx_word16_t *mem; - spx_word16_t *sinc_table; - spx_uint32_t sinc_table_length; - resampler_basic_func resampler_ptr; - - int in_stride; - int out_stride; -} ; - -static double kaiser12_table[68] = { - 0.99859849, 1.00000000, 0.99859849, 0.99440475, 0.98745105, 0.97779076, - 0.96549770, 0.95066529, 0.93340547, 0.91384741, 0.89213598, 0.86843014, - 0.84290116, 0.81573067, 0.78710866, 0.75723148, 0.72629970, 0.69451601, - 0.66208321, 0.62920216, 0.59606986, 0.56287762, 0.52980938, 0.49704014, - 0.46473455, 0.43304576, 0.40211431, 0.37206735, 0.34301800, 0.31506490, - 0.28829195, 0.26276832, 0.23854851, 0.21567274, 0.19416736, 0.17404546, - 0.15530766, 0.13794294, 0.12192957, 0.10723616, 0.09382272, 0.08164178, - 0.07063950, 0.06075685, 0.05193064, 0.04409466, 0.03718069, 0.03111947, - 0.02584161, 0.02127838, 0.01736250, 0.01402878, 0.01121463, 0.00886058, - 0.00691064, 0.00531256, 0.00401805, 0.00298291, 0.00216702, 0.00153438, - 0.00105297, 0.00069463, 0.00043489, 0.00025272, 0.00013031, 0.0000527734, - 0.00001000, 0.00000000}; -/* -static double kaiser12_table[36] = { - 0.99440475, 1.00000000, 0.99440475, 0.97779076, 0.95066529, 0.91384741, - 0.86843014, 0.81573067, 0.75723148, 0.69451601, 0.62920216, 0.56287762, - 0.49704014, 0.43304576, 0.37206735, 0.31506490, 0.26276832, 0.21567274, - 0.17404546, 0.13794294, 0.10723616, 0.08164178, 0.06075685, 0.04409466, - 0.03111947, 0.02127838, 0.01402878, 0.00886058, 0.00531256, 0.00298291, - 0.00153438, 0.00069463, 0.00025272, 0.0000527734, 0.00000500, 0.00000000}; -*/ -static double kaiser10_table[36] = { - 0.99537781, 1.00000000, 0.99537781, 0.98162644, 0.95908712, 0.92831446, - 0.89005583, 0.84522401, 0.79486424, 0.74011713, 0.68217934, 0.62226347, - 0.56155915, 0.50119680, 0.44221549, 0.38553619, 0.33194107, 0.28205962, - 0.23636152, 0.19515633, 0.15859932, 0.12670280, 0.09935205, 0.07632451, - 0.05731132, 0.04193980, 0.02979584, 0.02044510, 0.01345224, 0.00839739, - 0.00488951, 0.00257636, 0.00115101, 0.00035515, 0.00000000, 0.00000000}; - -static double kaiser8_table[36] = { - 0.99635258, 1.00000000, 0.99635258, 0.98548012, 0.96759014, 0.94302200, - 0.91223751, 0.87580811, 0.83439927, 0.78875245, 0.73966538, 0.68797126, - 0.63451750, 0.58014482, 0.52566725, 0.47185369, 0.41941150, 0.36897272, - 0.32108304, 0.27619388, 0.23465776, 0.19672670, 0.16255380, 0.13219758, - 0.10562887, 0.08273982, 0.06335451, 0.04724088, 0.03412321, 0.02369490, - 0.01563093, 0.00959968, 0.00527363, 0.00233883, 0.00050000, 0.00000000}; - -static double kaiser6_table[36] = { - 0.99733006, 1.00000000, 0.99733006, 0.98935595, 0.97618418, 0.95799003, - 0.93501423, 0.90755855, 0.87598009, 0.84068475, 0.80211977, 0.76076565, - 0.71712752, 0.67172623, 0.62508937, 0.57774224, 0.53019925, 0.48295561, - 0.43647969, 0.39120616, 0.34752997, 0.30580127, 0.26632152, 0.22934058, - 0.19505503, 0.16360756, 0.13508755, 0.10953262, 0.08693120, 0.06722600, - 0.05031820, 0.03607231, 0.02432151, 0.01487334, 0.00752000, 0.00000000}; - -struct FuncDef { - double *table; - int oversample; -}; - -static struct FuncDef _KAISER12 = {kaiser12_table, 64}; -#define KAISER12 (&_KAISER12) -/*static struct FuncDef _KAISER12 = {kaiser12_table, 32}; -#define KAISER12 (&_KAISER12)*/ -static struct FuncDef _KAISER10 = {kaiser10_table, 32}; -#define KAISER10 (&_KAISER10) -static struct FuncDef _KAISER8 = {kaiser8_table, 32}; -#define KAISER8 (&_KAISER8) -static struct FuncDef _KAISER6 = {kaiser6_table, 32}; -#define KAISER6 (&_KAISER6) - -struct QualityMapping { - int base_length; - int oversample; - float downsample_bandwidth; - float upsample_bandwidth; - struct FuncDef *window_func; -}; - - -/* This table maps conversion quality to internal parameters. There are two - reasons that explain why the up-sampling bandwidth is larger than the - down-sampling bandwidth: - 1) When up-sampling, we can assume that the spectrum is already attenuated - close to the Nyquist rate (from an A/D or a previous resampling filter) - 2) Any aliasing that occurs very close to the Nyquist rate will be masked - by the sinusoids/noise just below the Nyquist rate (guaranteed only for - up-sampling). -*/ -static const struct QualityMapping quality_map[11] = { - { 8, 4, 0.830f, 0.860f, KAISER6 }, /* Q0 */ - { 16, 4, 0.850f, 0.880f, KAISER6 }, /* Q1 */ - { 32, 4, 0.882f, 0.910f, KAISER6 }, /* Q2 */ /* 82.3% cutoff ( ~60 dB stop) 6 */ - { 48, 8, 0.895f, 0.917f, KAISER8 }, /* Q3 */ /* 84.9% cutoff ( ~80 dB stop) 8 */ - { 64, 8, 0.921f, 0.940f, KAISER8 }, /* Q4 */ /* 88.7% cutoff ( ~80 dB stop) 8 */ - { 80, 16, 0.922f, 0.940f, KAISER10}, /* Q5 */ /* 89.1% cutoff (~100 dB stop) 10 */ - { 96, 16, 0.940f, 0.945f, KAISER10}, /* Q6 */ /* 91.5% cutoff (~100 dB stop) 10 */ - {128, 16, 0.950f, 0.950f, KAISER10}, /* Q7 */ /* 93.1% cutoff (~100 dB stop) 10 */ - {160, 16, 0.960f, 0.960f, KAISER10}, /* Q8 */ /* 94.5% cutoff (~100 dB stop) 10 */ - {192, 32, 0.968f, 0.968f, KAISER12}, /* Q9 */ /* 95.5% cutoff (~100 dB stop) 10 */ - {256, 32, 0.975f, 0.975f, KAISER12}, /* Q10 */ /* 96.6% cutoff (~100 dB stop) 10 */ -}; -/*8,24,40,56,80,104,128,160,200,256,320*/ -static double compute_func(float x, struct FuncDef *func) -{ - float y, frac; - double interp[4]; - int ind; - y = x*func->oversample; - ind = (int)floor(y); - frac = (y-ind); - /* CSE with handle the repeated powers */ - interp[3] = -0.1666666667*frac + 0.1666666667*(frac*frac*frac); - interp[2] = frac + 0.5*(frac*frac) - 0.5*(frac*frac*frac); - /*interp[2] = 1.f - 0.5f*frac - frac*frac + 0.5f*frac*frac*frac;*/ - interp[0] = -0.3333333333*frac + 0.5*(frac*frac) - 0.1666666667*(frac*frac*frac); - /* Just to make sure we don't have rounding problems */ - interp[1] = 1.f-interp[3]-interp[2]-interp[0]; - - /*sum = frac*accum[1] + (1-frac)*accum[2];*/ - return interp[0]*func->table[ind] + interp[1]*func->table[ind+1] + interp[2]*func->table[ind+2] + interp[3]*func->table[ind+3]; -} - -#if 0 -#include -int main(int argc, char **argv) -{ - int i; - for (i=0;i<256;i++) - { - printf ("%f\n", compute_func(i/256., KAISER12)); - } - return 0; -} -#endif - -#ifdef FIXED_POINT -/* The slow way of computing a sinc for the table. Should improve that some day */ -static spx_word16_t sinc(float cutoff, float x, int N, struct FuncDef *window_func) -{ - /*fprintf (stderr, "%f ", x);*/ - float xx = x * cutoff; - if (fabs(x)<1e-6f) - return WORD2INT(32768.*cutoff); - else if (fabs(x) > .5f*N) - return 0; - /*FIXME: Can it really be any slower than this? */ - return WORD2INT(32768.*cutoff*sin(M_PI*xx)/(M_PI*xx) * compute_func(fabs(2.*x/N), window_func)); -} -#else -/* The slow way of computing a sinc for the table. Should improve that some day */ -static spx_word16_t sinc(float cutoff, float x, int N, struct FuncDef *window_func) -{ - /*fprintf (stderr, "%f ", x);*/ - float xx = x * cutoff; - if (fabs(x)<1e-6) - return cutoff; - else if (fabs(x) > .5*N) - return 0; - /*FIXME: Can it really be any slower than this? */ - return cutoff*sin(M_PI*xx)/(M_PI*xx) * compute_func(fabs(2.*x/N), window_func); -} -#endif - -#ifdef FIXED_POINT -static void cubic_coef(spx_word16_t x, spx_word16_t interp[4]) -{ - /* Compute interpolation coefficients. I'm not sure whether this corresponds to cubic interpolation - but I know it's MMSE-optimal on a sinc */ - spx_word16_t x2, x3; - x2 = MULT16_16_P15(x, x); - x3 = MULT16_16_P15(x, x2); - interp[0] = PSHR32(MULT16_16(QCONST16(-0.16667f, 15),x) + MULT16_16(QCONST16(0.16667f, 15),x3),15); - interp[1] = EXTRACT16(EXTEND32(x) + SHR32(SUB32(EXTEND32(x2),EXTEND32(x3)),1)); - interp[3] = PSHR32(MULT16_16(QCONST16(-0.33333f, 15),x) + MULT16_16(QCONST16(.5f,15),x2) - MULT16_16(QCONST16(0.16667f, 15),x3),15); - /* Just to make sure we don't have rounding problems */ - interp[2] = Q15_ONE-interp[0]-interp[1]-interp[3]; - if (interp[2]<32767) - interp[2]+=1; -} -#else -static void cubic_coef(spx_word16_t frac, spx_word16_t interp[4]) -{ - /* Compute interpolation coefficients. I'm not sure whether this corresponds to cubic interpolation - but I know it's MMSE-optimal on a sinc */ - interp[0] = -0.16667f*frac + 0.16667f*frac*frac*frac; - interp[1] = frac + 0.5f*frac*frac - 0.5f*frac*frac*frac; - /*interp[2] = 1.f - 0.5f*frac - frac*frac + 0.5f*frac*frac*frac;*/ - interp[3] = -0.33333f*frac + 0.5f*frac*frac - 0.16667f*frac*frac*frac; - /* Just to make sure we don't have rounding problems */ - interp[2] = 1.-interp[0]-interp[1]-interp[3]; -} -#endif - -static int resampler_basic_direct_single(SpeexResamplerState *st, spx_uint32_t channel_index, const spx_word16_t *in, spx_uint32_t *in_len, spx_word16_t *out, spx_uint32_t *out_len) -{ - const int N = st->filt_len; - int out_sample = 0; - int last_sample = st->last_sample[channel_index]; - spx_uint32_t samp_frac_num = st->samp_frac_num[channel_index]; - const spx_word16_t *sinc_table = st->sinc_table; - const int out_stride = st->out_stride; - const int int_advance = st->int_advance; - const int frac_advance = st->frac_advance; - const spx_uint32_t den_rate = st->den_rate; - spx_word32_t sum; - int j; - - while (!(last_sample >= (spx_int32_t)*in_len || out_sample >= (spx_int32_t)*out_len)) - { - const spx_word16_t *sinc = & sinc_table[samp_frac_num*N]; - const spx_word16_t *iptr = & in[last_sample]; - -#ifndef OVERRIDE_INNER_PRODUCT_SINGLE - float accum[4] = {0,0,0,0}; - - for(j=0;j= den_rate) - { - samp_frac_num -= den_rate; - last_sample++; - } - } - - st->last_sample[channel_index] = last_sample; - st->samp_frac_num[channel_index] = samp_frac_num; - return out_sample; -} - -#ifdef FIXED_POINT -#else -/* This is the same as the previous function, except with a double-precision accumulator */ -static int resampler_basic_direct_double(SpeexResamplerState *st, spx_uint32_t channel_index, const spx_word16_t *in, spx_uint32_t *in_len, spx_word16_t *out, spx_uint32_t *out_len) -{ - const int N = st->filt_len; - int out_sample = 0; - int last_sample = st->last_sample[channel_index]; - spx_uint32_t samp_frac_num = st->samp_frac_num[channel_index]; - const spx_word16_t *sinc_table = st->sinc_table; - const int out_stride = st->out_stride; - const int int_advance = st->int_advance; - const int frac_advance = st->frac_advance; - const spx_uint32_t den_rate = st->den_rate; - double sum; - int j; - - while (!(last_sample >= (spx_int32_t)*in_len || out_sample >= (spx_int32_t)*out_len)) - { - const spx_word16_t *sinc = & sinc_table[samp_frac_num*N]; - const spx_word16_t *iptr = & in[last_sample]; - -#ifndef OVERRIDE_INNER_PRODUCT_DOUBLE - double accum[4] = {0,0,0,0}; - - for(j=0;j= den_rate) - { - samp_frac_num -= den_rate; - last_sample++; - } - } - - st->last_sample[channel_index] = last_sample; - st->samp_frac_num[channel_index] = samp_frac_num; - return out_sample; -} -#endif - -static int resampler_basic_interpolate_single(SpeexResamplerState *st, spx_uint32_t channel_index, const spx_word16_t *in, spx_uint32_t *in_len, spx_word16_t *out, spx_uint32_t *out_len) -{ - const int N = st->filt_len; - int out_sample = 0; - int last_sample = st->last_sample[channel_index]; - spx_uint32_t samp_frac_num = st->samp_frac_num[channel_index]; - const int out_stride = st->out_stride; - const int int_advance = st->int_advance; - const int frac_advance = st->frac_advance; - const spx_uint32_t den_rate = st->den_rate; - int j; - spx_word32_t sum; - - while (!(last_sample >= (spx_int32_t)*in_len || out_sample >= (spx_int32_t)*out_len)) - { - const spx_word16_t *iptr = & in[last_sample]; - - const int offset = samp_frac_num*st->oversample/st->den_rate; -#ifdef FIXED_POINT - const spx_word16_t frac = PDIV32(SHL32((samp_frac_num*st->oversample) % st->den_rate,15),st->den_rate); -#else - const spx_word16_t frac = ((float)((samp_frac_num*st->oversample) % st->den_rate))/st->den_rate; -#endif - spx_word16_t interp[4]; - - -#ifndef OVERRIDE_INTERPOLATE_PRODUCT_SINGLE - spx_word32_t accum[4] = {0,0,0,0}; - - for(j=0;jsinc_table[4+(j+1)*st->oversample-offset-2]); - accum[1] += MULT16_16(curr_in,st->sinc_table[4+(j+1)*st->oversample-offset-1]); - accum[2] += MULT16_16(curr_in,st->sinc_table[4+(j+1)*st->oversample-offset]); - accum[3] += MULT16_16(curr_in,st->sinc_table[4+(j+1)*st->oversample-offset+1]); - } - - cubic_coef(frac, interp); - sum = MULT16_32_Q15(interp[0],accum[0]) + MULT16_32_Q15(interp[1],accum[1]) + MULT16_32_Q15(interp[2],accum[2]) + MULT16_32_Q15(interp[3],accum[3]); -#else - cubic_coef(frac, interp); - sum = interpolate_product_single(iptr, st->sinc_table + st->oversample + 4 - offset - 2, N, st->oversample, interp); -#endif - - out[out_stride * out_sample++] = PSHR32(sum,15); - last_sample += int_advance; - samp_frac_num += frac_advance; - if (samp_frac_num >= den_rate) - { - samp_frac_num -= den_rate; - last_sample++; - } - } - - st->last_sample[channel_index] = last_sample; - st->samp_frac_num[channel_index] = samp_frac_num; - return out_sample; -} - -#ifdef FIXED_POINT -#else -/* This is the same as the previous function, except with a double-precision accumulator */ -static int resampler_basic_interpolate_double(SpeexResamplerState *st, spx_uint32_t channel_index, const spx_word16_t *in, spx_uint32_t *in_len, spx_word16_t *out, spx_uint32_t *out_len) -{ - const int N = st->filt_len; - int out_sample = 0; - int last_sample = st->last_sample[channel_index]; - spx_uint32_t samp_frac_num = st->samp_frac_num[channel_index]; - const int out_stride = st->out_stride; - const int int_advance = st->int_advance; - const int frac_advance = st->frac_advance; - const spx_uint32_t den_rate = st->den_rate; - int j; - spx_word32_t sum; - - while (!(last_sample >= (spx_int32_t)*in_len || out_sample >= (spx_int32_t)*out_len)) - { - const spx_word16_t *iptr = & in[last_sample]; - - const int offset = samp_frac_num*st->oversample/st->den_rate; -#ifdef FIXED_POINT - const spx_word16_t frac = PDIV32(SHL32((samp_frac_num*st->oversample) % st->den_rate,15),st->den_rate); -#else - const spx_word16_t frac = ((float)((samp_frac_num*st->oversample) % st->den_rate))/st->den_rate; -#endif - spx_word16_t interp[4]; - - -#ifndef OVERRIDE_INTERPOLATE_PRODUCT_DOUBLE - double accum[4] = {0,0,0,0}; - - for(j=0;jsinc_table[4+(j+1)*st->oversample-offset-2]); - accum[1] += MULT16_16(curr_in,st->sinc_table[4+(j+1)*st->oversample-offset-1]); - accum[2] += MULT16_16(curr_in,st->sinc_table[4+(j+1)*st->oversample-offset]); - accum[3] += MULT16_16(curr_in,st->sinc_table[4+(j+1)*st->oversample-offset+1]); - } - - cubic_coef(frac, interp); - sum = MULT16_32_Q15(interp[0],accum[0]) + MULT16_32_Q15(interp[1],accum[1]) + MULT16_32_Q15(interp[2],accum[2]) + MULT16_32_Q15(interp[3],accum[3]); -#else - cubic_coef(frac, interp); - sum = interpolate_product_double(iptr, st->sinc_table + st->oversample + 4 - offset - 2, N, st->oversample, interp); -#endif - - out[out_stride * out_sample++] = PSHR32(sum,15); - last_sample += int_advance; - samp_frac_num += frac_advance; - if (samp_frac_num >= den_rate) - { - samp_frac_num -= den_rate; - last_sample++; - } - } - - st->last_sample[channel_index] = last_sample; - st->samp_frac_num[channel_index] = samp_frac_num; - return out_sample; -} -#endif - -static void update_filter(SpeexResamplerState *st) -{ - spx_uint32_t old_length; - - old_length = st->filt_len; - st->oversample = quality_map[st->quality].oversample; - st->filt_len = quality_map[st->quality].base_length; - - if (st->num_rate > st->den_rate) - { - /* down-sampling */ - st->cutoff = quality_map[st->quality].downsample_bandwidth * st->den_rate / st->num_rate; - /* FIXME: divide the numerator and denominator by a certain amount if they're too large */ - st->filt_len = st->filt_len*st->num_rate / st->den_rate; - /* Round down to make sure we have a multiple of 4 */ - st->filt_len &= (~0x3); - if (2*st->den_rate < st->num_rate) - st->oversample >>= 1; - if (4*st->den_rate < st->num_rate) - st->oversample >>= 1; - if (8*st->den_rate < st->num_rate) - st->oversample >>= 1; - if (16*st->den_rate < st->num_rate) - st->oversample >>= 1; - if (st->oversample < 1) - st->oversample = 1; - } else { - /* up-sampling */ - st->cutoff = quality_map[st->quality].upsample_bandwidth; - } - - /* Choose the resampling type that requires the least amount of memory */ - if (st->den_rate <= st->oversample) - { - spx_uint32_t i; - if (!st->sinc_table) - st->sinc_table = (spx_word16_t *)speex_alloc(st->filt_len*st->den_rate*sizeof(spx_word16_t)); - else if (st->sinc_table_length < st->filt_len*st->den_rate) - { - st->sinc_table = (spx_word16_t *)speex_realloc(st->sinc_table,st->filt_len*st->den_rate*sizeof(spx_word16_t)); - st->sinc_table_length = st->filt_len*st->den_rate; - } - for (i=0;iden_rate;i++) - { - spx_int32_t j; - for (j=0;jfilt_len;j++) - { - st->sinc_table[i*st->filt_len+j] = sinc(st->cutoff,((j-(spx_int32_t)st->filt_len/2+1)-((float)i)/st->den_rate), st->filt_len, quality_map[st->quality].window_func); - } - } -#ifdef FIXED_POINT - st->resampler_ptr = resampler_basic_direct_single; -#else - if (st->quality>8) - st->resampler_ptr = resampler_basic_direct_double; - else - st->resampler_ptr = resampler_basic_direct_single; -#endif - /*fprintf (stderr, "resampler uses direct sinc table and normalised cutoff %f\n", cutoff);*/ - } else { - spx_int32_t i; - if (!st->sinc_table) - st->sinc_table = (spx_word16_t *)speex_alloc((st->filt_len*st->oversample+8)*sizeof(spx_word16_t)); - else if (st->sinc_table_length < st->filt_len*st->oversample+8) - { - st->sinc_table = (spx_word16_t *)speex_realloc(st->sinc_table,(st->filt_len*st->oversample+8)*sizeof(spx_word16_t)); - st->sinc_table_length = st->filt_len*st->oversample+8; - } - for (i=-4;i<(spx_int32_t)(st->oversample*st->filt_len+4);i++) - st->sinc_table[i+4] = sinc(st->cutoff,(i/(float)st->oversample - st->filt_len/2), st->filt_len, quality_map[st->quality].window_func); -#ifdef FIXED_POINT - st->resampler_ptr = resampler_basic_interpolate_single; -#else - if (st->quality>8) - st->resampler_ptr = resampler_basic_interpolate_double; - else - st->resampler_ptr = resampler_basic_interpolate_single; -#endif - /*fprintf (stderr, "resampler uses interpolated sinc table and normalised cutoff %f\n", cutoff);*/ - } - st->int_advance = st->num_rate/st->den_rate; - st->frac_advance = st->num_rate%st->den_rate; - - - /* Here's the place where we update the filter memory to take into account - the change in filter length. It's probably the messiest part of the code - due to handling of lots of corner cases. */ - if (!st->mem) - { - spx_uint32_t i; - st->mem_alloc_size = st->filt_len-1 + st->buffer_size; - st->mem = (spx_word16_t*)speex_alloc(st->nb_channels*st->mem_alloc_size * sizeof(spx_word16_t)); - for (i=0;inb_channels*st->mem_alloc_size;i++) - st->mem[i] = 0; - /*speex_warning("init filter");*/ - } else if (!st->started) - { - spx_uint32_t i; - st->mem_alloc_size = st->filt_len-1 + st->buffer_size; - st->mem = (spx_word16_t*)speex_realloc(st->mem, st->nb_channels*st->mem_alloc_size * sizeof(spx_word16_t)); - for (i=0;inb_channels*st->mem_alloc_size;i++) - st->mem[i] = 0; - /*speex_warning("reinit filter");*/ - } else if (st->filt_len > old_length) - { - spx_int32_t i; - /* Increase the filter length */ - /*speex_warning("increase filter size");*/ - int old_alloc_size = st->mem_alloc_size; - if ((st->filt_len-1 + st->buffer_size) > st->mem_alloc_size) - { - st->mem_alloc_size = st->filt_len-1 + st->buffer_size; - st->mem = (spx_word16_t*)speex_realloc(st->mem, st->nb_channels*st->mem_alloc_size * sizeof(spx_word16_t)); - } - for (i=st->nb_channels-1;i>=0;i--) - { - spx_int32_t j; - spx_uint32_t olen = old_length; - /*if (st->magic_samples[i])*/ - { - /* Try and remove the magic samples as if nothing had happened */ - - /* FIXME: This is wrong but for now we need it to avoid going over the array bounds */ - olen = old_length + 2*st->magic_samples[i]; - for (j=old_length-2+st->magic_samples[i];j>=0;j--) - st->mem[i*st->mem_alloc_size+j+st->magic_samples[i]] = st->mem[i*old_alloc_size+j]; - for (j=0;jmagic_samples[i];j++) - st->mem[i*st->mem_alloc_size+j] = 0; - st->magic_samples[i] = 0; - } - if (st->filt_len > olen) - { - /* If the new filter length is still bigger than the "augmented" length */ - /* Copy data going backward */ - for (j=0;jmem[i*st->mem_alloc_size+(st->filt_len-2-j)] = st->mem[i*st->mem_alloc_size+(olen-2-j)]; - /* Then put zeros for lack of anything better */ - for (;jfilt_len-1;j++) - st->mem[i*st->mem_alloc_size+(st->filt_len-2-j)] = 0; - /* Adjust last_sample */ - st->last_sample[i] += (st->filt_len - olen)/2; - } else { - /* Put back some of the magic! */ - st->magic_samples[i] = (olen - st->filt_len)/2; - for (j=0;jfilt_len-1+st->magic_samples[i];j++) - st->mem[i*st->mem_alloc_size+j] = st->mem[i*st->mem_alloc_size+j+st->magic_samples[i]]; - } - } - } else if (st->filt_len < old_length) - { - spx_uint32_t i; - /* Reduce filter length, this a bit tricky. We need to store some of the memory as "magic" - samples so they can be used directly as input the next time(s) */ - for (i=0;inb_channels;i++) - { - spx_uint32_t j; - spx_uint32_t old_magic = st->magic_samples[i]; - st->magic_samples[i] = (old_length - st->filt_len)/2; - /* We must copy some of the memory that's no longer used */ - /* Copy data going backward */ - for (j=0;jfilt_len-1+st->magic_samples[i]+old_magic;j++) - st->mem[i*st->mem_alloc_size+j] = st->mem[i*st->mem_alloc_size+j+st->magic_samples[i]]; - st->magic_samples[i] += old_magic; - } - } - -} - -EXPORT SpeexResamplerState *speex_resampler_init(spx_uint32_t nb_channels, spx_uint32_t in_rate, spx_uint32_t out_rate, int quality, int *err) -{ - return speex_resampler_init_frac(nb_channels, in_rate, out_rate, in_rate, out_rate, quality, err); -} - -EXPORT SpeexResamplerState *speex_resampler_init_frac(spx_uint32_t nb_channels, spx_uint32_t ratio_num, spx_uint32_t ratio_den, spx_uint32_t in_rate, spx_uint32_t out_rate, int quality, int *err) -{ - spx_uint32_t i; - SpeexResamplerState *st; - if (quality > 10 || quality < 0) - { - if (err) - *err = RESAMPLER_ERR_INVALID_ARG; - return NULL; - } - st = (SpeexResamplerState *)speex_alloc(sizeof(SpeexResamplerState)); - st->initialised = 0; - st->started = 0; - st->in_rate = 0; - st->out_rate = 0; - st->num_rate = 0; - st->den_rate = 0; - st->quality = -1; - st->sinc_table_length = 0; - st->mem_alloc_size = 0; - st->filt_len = 0; - st->mem = 0; - st->resampler_ptr = 0; - - st->cutoff = 1.f; - st->nb_channels = nb_channels; - st->in_stride = 1; - st->out_stride = 1; - -#ifdef FIXED_POINT - st->buffer_size = 160; -#else - st->buffer_size = 160; -#endif - - /* Per channel data */ - st->last_sample = (spx_int32_t*)speex_alloc(nb_channels*sizeof(int)); - st->magic_samples = (spx_uint32_t*)speex_alloc(nb_channels*sizeof(int)); - st->samp_frac_num = (spx_uint32_t*)speex_alloc(nb_channels*sizeof(int)); - for (i=0;ilast_sample[i] = 0; - st->magic_samples[i] = 0; - st->samp_frac_num[i] = 0; - } - - speex_resampler_set_quality(st, quality); - speex_resampler_set_rate_frac(st, ratio_num, ratio_den, in_rate, out_rate); - - - update_filter(st); - - st->initialised = 1; - if (err) - *err = RESAMPLER_ERR_SUCCESS; - - return st; -} - -EXPORT void speex_resampler_destroy(SpeexResamplerState *st) -{ - speex_free(st->mem); - speex_free(st->sinc_table); - speex_free(st->last_sample); - speex_free(st->magic_samples); - speex_free(st->samp_frac_num); - speex_free(st); -} - -static int speex_resampler_process_native(SpeexResamplerState *st, spx_uint32_t channel_index, spx_uint32_t *in_len, spx_word16_t *out, spx_uint32_t *out_len) -{ - int j=0; - const int N = st->filt_len; - int out_sample = 0; - spx_word16_t *mem = st->mem + channel_index * st->mem_alloc_size; - spx_uint32_t ilen; - - st->started = 1; - - /* Call the right resampler through the function ptr */ - out_sample = st->resampler_ptr(st, channel_index, mem, in_len, out, out_len); - - if (st->last_sample[channel_index] < (spx_int32_t)*in_len) - *in_len = st->last_sample[channel_index]; - *out_len = out_sample; - st->last_sample[channel_index] -= *in_len; - - ilen = *in_len; - - for(j=0;jmagic_samples[channel_index]; - spx_word16_t *mem = st->mem + channel_index * st->mem_alloc_size; - const int N = st->filt_len; - - speex_resampler_process_native(st, channel_index, &tmp_in_len, *out, &out_len); - - st->magic_samples[channel_index] -= tmp_in_len; - - /* If we couldn't process all "magic" input samples, save the rest for next time */ - if (st->magic_samples[channel_index]) - { - spx_uint32_t i; - for (i=0;imagic_samples[channel_index];i++) - mem[N-1+i]=mem[N-1+i+tmp_in_len]; - } - *out += out_len*st->out_stride; - return out_len; -} - -#ifdef FIXED_POINT -EXPORT int speex_resampler_process_int(SpeexResamplerState *st, spx_uint32_t channel_index, const spx_int16_t *in, spx_uint32_t *in_len, spx_int16_t *out, spx_uint32_t *out_len) -#else -EXPORT int speex_resampler_process_float(SpeexResamplerState *st, spx_uint32_t channel_index, const float *in, spx_uint32_t *in_len, float *out, spx_uint32_t *out_len) -#endif -{ - int j; - spx_uint32_t ilen = *in_len; - spx_uint32_t olen = *out_len; - spx_word16_t *x = st->mem + channel_index * st->mem_alloc_size; - const int filt_offs = st->filt_len - 1; - const spx_uint32_t xlen = st->mem_alloc_size - filt_offs; - const int istride = st->in_stride; - - if (st->magic_samples[channel_index]) - olen -= speex_resampler_magic(st, channel_index, &out, olen); - if (! st->magic_samples[channel_index]) { - while (ilen && olen) { - spx_uint32_t ichunk = (ilen > xlen) ? xlen : ilen; - spx_uint32_t ochunk = olen; - - if (in) { - for(j=0;jout_stride; - if (in) - in += ichunk * istride; - } - } - *in_len -= ilen; - *out_len -= olen; - return RESAMPLER_ERR_SUCCESS; -} - -#ifdef FIXED_POINT -EXPORT int speex_resampler_process_float(SpeexResamplerState *st, spx_uint32_t channel_index, const float *in, spx_uint32_t *in_len, float *out, spx_uint32_t *out_len) -#else -EXPORT int speex_resampler_process_int(SpeexResamplerState *st, spx_uint32_t channel_index, const spx_int16_t *in, spx_uint32_t *in_len, spx_int16_t *out, spx_uint32_t *out_len) -#endif -{ - int j; - const int istride_save = st->in_stride; - const int ostride_save = st->out_stride; - spx_uint32_t ilen = *in_len; - spx_uint32_t olen = *out_len; - spx_word16_t *x = st->mem + channel_index * st->mem_alloc_size; - const spx_uint32_t xlen = st->mem_alloc_size - (st->filt_len - 1); -#ifdef VAR_ARRAYS - const unsigned int ylen = (olen < FIXED_STACK_ALLOC) ? olen : FIXED_STACK_ALLOC; - VARDECL(spx_word16_t *ystack); - ALLOC(ystack, ylen, spx_word16_t); -#else - const unsigned int ylen = FIXED_STACK_ALLOC; - spx_word16_t ystack[FIXED_STACK_ALLOC]; -#endif - - st->out_stride = 1; - - while (ilen && olen) { - spx_word16_t *y = ystack; - spx_uint32_t ichunk = (ilen > xlen) ? xlen : ilen; - spx_uint32_t ochunk = (olen > ylen) ? ylen : olen; - spx_uint32_t omagic = 0; - - if (st->magic_samples[channel_index]) { - omagic = speex_resampler_magic(st, channel_index, &y, ochunk); - ochunk -= omagic; - olen -= omagic; - } - if (! st->magic_samples[channel_index]) { - if (in) { - for(j=0;jfilt_len-1]=WORD2INT(in[j*istride_save]); -#else - x[j+st->filt_len-1]=in[j*istride_save]; -#endif - } else { - for(j=0;jfilt_len-1]=0; - } - - speex_resampler_process_native(st, channel_index, &ichunk, y, &ochunk); - } else { - ichunk = 0; - ochunk = 0; - } - - for (j=0;jout_stride = ostride_save; - *in_len -= ilen; - *out_len -= olen; - - return RESAMPLER_ERR_SUCCESS; -} - -EXPORT int speex_resampler_process_interleaved_float(SpeexResamplerState *st, const float *in, spx_uint32_t *in_len, float *out, spx_uint32_t *out_len) -{ - spx_uint32_t i; - int istride_save, ostride_save; - spx_uint32_t bak_len = *out_len; - istride_save = st->in_stride; - ostride_save = st->out_stride; - st->in_stride = st->out_stride = st->nb_channels; - for (i=0;inb_channels;i++) - { - *out_len = bak_len; - if (in != NULL) - speex_resampler_process_float(st, i, in+i, in_len, out+i, out_len); - else - speex_resampler_process_float(st, i, NULL, in_len, out+i, out_len); - } - st->in_stride = istride_save; - st->out_stride = ostride_save; - return RESAMPLER_ERR_SUCCESS; -} - -EXPORT int speex_resampler_process_interleaved_int(SpeexResamplerState *st, const spx_int16_t *in, spx_uint32_t *in_len, spx_int16_t *out, spx_uint32_t *out_len) -{ - spx_uint32_t i; - int istride_save, ostride_save; - spx_uint32_t bak_len = *out_len; - istride_save = st->in_stride; - ostride_save = st->out_stride; - st->in_stride = st->out_stride = st->nb_channels; - for (i=0;inb_channels;i++) - { - *out_len = bak_len; - if (in != NULL) - speex_resampler_process_int(st, i, in+i, in_len, out+i, out_len); - else - speex_resampler_process_int(st, i, NULL, in_len, out+i, out_len); - } - st->in_stride = istride_save; - st->out_stride = ostride_save; - return RESAMPLER_ERR_SUCCESS; -} - -EXPORT int speex_resampler_set_rate(SpeexResamplerState *st, spx_uint32_t in_rate, spx_uint32_t out_rate) -{ - return speex_resampler_set_rate_frac(st, in_rate, out_rate, in_rate, out_rate); -} - -EXPORT void speex_resampler_get_rate(SpeexResamplerState *st, spx_uint32_t *in_rate, spx_uint32_t *out_rate) -{ - *in_rate = st->in_rate; - *out_rate = st->out_rate; -} - -EXPORT int speex_resampler_set_rate_frac(SpeexResamplerState *st, spx_uint32_t ratio_num, spx_uint32_t ratio_den, spx_uint32_t in_rate, spx_uint32_t out_rate) -{ - spx_uint32_t fact; - spx_uint32_t old_den; - spx_uint32_t i; - if (st->in_rate == in_rate && st->out_rate == out_rate && st->num_rate == ratio_num && st->den_rate == ratio_den) - return RESAMPLER_ERR_SUCCESS; - - old_den = st->den_rate; - st->in_rate = in_rate; - st->out_rate = out_rate; - st->num_rate = ratio_num; - st->den_rate = ratio_den; - /* FIXME: This is terribly inefficient, but who cares (at least for now)? */ - for (fact=2;fact<=IMIN(st->num_rate, st->den_rate);fact++) - { - while ((st->num_rate % fact == 0) && (st->den_rate % fact == 0)) - { - st->num_rate /= fact; - st->den_rate /= fact; - } - } - - if (old_den > 0) - { - for (i=0;inb_channels;i++) - { - st->samp_frac_num[i]=st->samp_frac_num[i]*st->den_rate/old_den; - /* Safety net */ - if (st->samp_frac_num[i] >= st->den_rate) - st->samp_frac_num[i] = st->den_rate-1; - } - } - - if (st->initialised) - update_filter(st); - return RESAMPLER_ERR_SUCCESS; -} - -EXPORT void speex_resampler_get_ratio(SpeexResamplerState *st, spx_uint32_t *ratio_num, spx_uint32_t *ratio_den) -{ - *ratio_num = st->num_rate; - *ratio_den = st->den_rate; -} - -EXPORT int speex_resampler_set_quality(SpeexResamplerState *st, int quality) -{ - if (quality > 10 || quality < 0) - return RESAMPLER_ERR_INVALID_ARG; - if (st->quality == quality) - return RESAMPLER_ERR_SUCCESS; - st->quality = quality; - if (st->initialised) - update_filter(st); - return RESAMPLER_ERR_SUCCESS; -} - -EXPORT void speex_resampler_get_quality(SpeexResamplerState *st, int *quality) -{ - *quality = st->quality; -} - -EXPORT void speex_resampler_set_input_stride(SpeexResamplerState *st, spx_uint32_t stride) -{ - st->in_stride = stride; -} - -EXPORT void speex_resampler_get_input_stride(SpeexResamplerState *st, spx_uint32_t *stride) -{ - *stride = st->in_stride; -} - -EXPORT void speex_resampler_set_output_stride(SpeexResamplerState *st, spx_uint32_t stride) -{ - st->out_stride = stride; -} - -EXPORT void speex_resampler_get_output_stride(SpeexResamplerState *st, spx_uint32_t *stride) -{ - *stride = st->out_stride; -} - -EXPORT int speex_resampler_get_input_latency(SpeexResamplerState *st) -{ - return st->filt_len / 2; -} - -EXPORT int speex_resampler_get_output_latency(SpeexResamplerState *st) -{ - return ((st->filt_len / 2) * st->den_rate + (st->num_rate >> 1)) / st->num_rate; -} - -EXPORT int speex_resampler_skip_zeros(SpeexResamplerState *st) -{ - spx_uint32_t i; - for (i=0;inb_channels;i++) - st->last_sample[i] = st->filt_len/2; - return RESAMPLER_ERR_SUCCESS; -} - -EXPORT int speex_resampler_reset_mem(SpeexResamplerState *st) -{ - spx_uint32_t i; - for (i=0;inb_channels*(st->filt_len-1);i++) - st->mem[i] = 0; - return RESAMPLER_ERR_SUCCESS; -} - -EXPORT const char *speex_resampler_strerror(int err) -{ - switch (err) - { - case RESAMPLER_ERR_SUCCESS: - return "Success."; - case RESAMPLER_ERR_ALLOC_FAILED: - return "Memory allocation failed."; - case RESAMPLER_ERR_BAD_STATE: - return "Bad resampler state."; - case RESAMPLER_ERR_INVALID_ARG: - return "Invalid argument."; - case RESAMPLER_ERR_PTR_OVERLAP: - return "Input and output buffers overlap."; - default: - return "Unknown error. Bad error code or strange version mismatch."; - } -} diff --git a/mednafen/resampler/resample_sse.h b/mednafen/resampler/resample_sse.h deleted file mode 100644 index 4bd35a2..0000000 --- a/mednafen/resampler/resample_sse.h +++ /dev/null @@ -1,128 +0,0 @@ -/* Copyright (C) 2007-2008 Jean-Marc Valin - * Copyright (C) 2008 Thorvald Natvig - */ -/** - @file resample_sse.h - @brief Resampler functions (SSE version) -*/ -/* - Redistribution and use in source and binary forms, with or without - modification, are permitted provided that the following conditions - are met: - - - Redistributions of source code must retain the above copyright - notice, this list of conditions and the following disclaimer. - - - Redistributions in binary form must reproduce the above copyright - notice, this list of conditions and the following disclaimer in the - documentation and/or other materials provided with the distribution. - - - Neither the name of the Xiph.org Foundation nor the names of its - contributors may be used to endorse or promote products derived from - this software without specific prior written permission. - - THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS - ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT - LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR - A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR - CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, - EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, - PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR - PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF - LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING - NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS - SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -*/ - -#include - -#define OVERRIDE_INNER_PRODUCT_SINGLE -static inline float inner_product_single(const float *a, const float *b, unsigned int len) -{ - int i; - float ret; - __m128 sum = _mm_setzero_ps(); - for (i=0;i -#define OVERRIDE_INNER_PRODUCT_DOUBLE - -static inline double inner_product_double(const float *a, const float *b, unsigned int len) -{ - int i; - double ret; - __m128d sum = _mm_setzero_pd(); - __m128 t; - for (i=0;i -# else -# ifdef HAVE_ALLOCA_H -# include -# else -# include -# endif -# endif -#endif - -/** - * @def ALIGN(stack, size) - * - * Aligns the stack to a 'size' boundary - * - * @param stack Stack - * @param size New size boundary - */ - -/** - * @def PUSH(stack, size, type) - * - * Allocates 'size' elements of type 'type' on the stack - * - * @param stack Stack - * @param size Number of elements - * @param type Type of element - */ - -/** - * @def VARDECL(var) - * - * Declare variable on stack - * - * @param var Variable to declare - */ - -/** - * @def ALLOC(var, size, type) - * - * Allocate 'size' elements of 'type' on stack - * - * @param var Name of variable to allocate - * @param size Number of elements - * @param type Type of element - */ - -#ifdef ENABLE_VALGRIND - -#include - -#define ALIGN(stack, size) ((stack) += ((size) - (long)(stack)) & ((size) - 1)) - -#define PUSH(stack, size, type) (VALGRIND_MAKE_NOACCESS(stack, 1000),ALIGN((stack),sizeof(type)),VALGRIND_MAKE_WRITABLE(stack, ((size)*sizeof(type))),(stack)+=((size)*sizeof(type)),(type*)((stack)-((size)*sizeof(type)))) - -#else - -#define ALIGN(stack, size) ((stack) += ((size) - (long)(stack)) & ((size) - 1)) - -#define PUSH(stack, size, type) (ALIGN((stack),sizeof(type)),(stack)+=((size)*sizeof(type)),(type*)((stack)-((size)*sizeof(type)))) - -#endif - -#if defined(VAR_ARRAYS) -#define VARDECL(var) -#define ALLOC(var, size, type) type var[size] -#elif defined(USE_ALLOCA) -#define VARDECL(var) var -#define ALLOC(var, size, type) var = alloca(sizeof(type)*(size)) -#else -#define VARDECL(var) var -#define ALLOC(var, size, type) var = PUSH(stack, size, type) -#endif - - -#endif diff --git a/mednafen/sound/Fir_Resampler.cpp b/mednafen/sound/Fir_Resampler.cpp deleted file mode 100644 index 4e0a463..0000000 --- a/mednafen/sound/Fir_Resampler.cpp +++ /dev/null @@ -1,199 +0,0 @@ -// Game_Music_Emu 0.5.2. http://www.slack.net/~ant/ - -#include "Fir_Resampler.h" - -#include -#include -#include -#include - -/* Copyright (C) 2004-2006 Shay Green. This module is free software; you -can redistribute it and/or modify it under the terms of the GNU Lesser -General Public License as published by the Free Software Foundation; either -version 2.1 of the License, or (at your option) any later version. This -module is distributed in the hope that it will be useful, but WITHOUT ANY -WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS -FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more -details. You should have received a copy of the GNU Lesser General Public -License along with this module; if not, write to the Free Software Foundation, -Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ - -#include "blargg_source.h" - -#undef PI -#define PI 3.1415926535897932384626433832795029 - -static void gen_sinc( double rolloff, int width, double offset, double spacing, double scale, - int count, short* out ) -{ - double const maxh = 256; - double const step = PI / maxh * spacing; - double const to_w = maxh * 2 / width; - double const pow_a_n = pow( rolloff, maxh ); - scale /= maxh * 2; - - double angle = (count / 2 - 1 + offset) * -step; - while ( count-- ) - { - *out++ = 0; - double w = angle * to_w; - if ( fabs( w ) < PI ) - { - double rolloff_cos_a = rolloff * cos( angle ); - double num = 1 - rolloff_cos_a - - pow_a_n * cos( maxh * angle ) + - pow_a_n * rolloff * cos( (maxh - 1) * angle ); - double den = 1 - rolloff_cos_a - rolloff_cos_a + rolloff * rolloff; - double sinc = scale * num / den - scale; - - out [-1] = (short) (cos( w ) * sinc + sinc); - } - angle += step; - } -} - -Fir_Resampler_::Fir_Resampler_( int width, sample_t* impulses_ ) : - width_( width ), - write_offset( width * stereo - stereo ), - impulses( impulses_ ) -{ - write_pos = 0; - res = 1; - imp_phase = 0; - skip_bits = 0; - step = stereo; - ratio_ = 1.0; -} - -Fir_Resampler_::~Fir_Resampler_() { } - -void Fir_Resampler_::clear() -{ - imp_phase = 0; - if ( buf.size() ) - { - write_pos = &buf [write_offset]; - memset( buf.begin(), 0, write_offset * sizeof buf [0] ); - } -} - -blargg_err_t Fir_Resampler_::buffer_size( int new_size ) -{ - RETURN_ERR( buf.resize( new_size + write_offset ) ); - clear(); - return 0; -} - -double Fir_Resampler_::time_ratio( double new_factor, double rolloff, double gain ) -{ - ratio_ = new_factor; - - double fstep = 0.0; - { - double least_error = 2; - double pos = 0; - res = -1; - for ( int r = 1; r <= max_res; r++ ) - { - pos += ratio_; - double nearest = floor( pos + 0.5 ); - double error = fabs( pos - nearest ); - if ( error < least_error ) - { - res = r; - fstep = nearest / res; - least_error = error; - } - } - } - - skip_bits = 0; - - step = stereo * (int) floor( fstep ); - - ratio_ = fstep; - fstep = fmod( fstep, 1.0 ); - - double filter = (ratio_ < 1.0) ? 1.0 : 1.0 / ratio_; - double pos = 0.0; - input_per_cycle = 0; - for ( int i = 0; i < res; i++ ) - { - gen_sinc( rolloff, int (width_ * filter + 1) & ~1, pos, filter, - double (0x7FFF * gain * filter), - (int) width_, impulses + i * width_ ); - - pos += fstep; - input_per_cycle += step; - if ( pos >= 0.9999999 ) - { - pos -= 1.0; - skip_bits |= 1 << i; - input_per_cycle++; - } - } - - clear(); - - return ratio_; -} - -int Fir_Resampler_::input_needed( blargg_long output_count ) const -{ - blargg_long input_count = 0; - - unsigned long skip = skip_bits >> imp_phase; - int remain = res - imp_phase; - while ( (output_count -= 2) > 0 ) - { - input_count += step + (skip & 1) * stereo; - skip >>= 1; - if ( !--remain ) - { - skip = skip_bits; - remain = res; - } - output_count -= 2; - } - - long input_extra = input_count - (write_pos - &buf [(width_ - 1) * stereo]); - if ( input_extra < 0 ) - input_extra = 0; - return input_extra; -} - -int Fir_Resampler_::avail_( blargg_long input_count ) const -{ - int cycle_count = input_count / input_per_cycle; - int output_count = cycle_count * res * stereo; - input_count -= cycle_count * input_per_cycle; - - blargg_ulong skip = skip_bits >> imp_phase; - int remain = res - imp_phase; - while ( input_count >= 0 ) - { - input_count -= step + (skip & 1) * stereo; - skip >>= 1; - if ( !--remain ) - { - skip = skip_bits; - remain = res; - } - output_count += 2; - } - return output_count; -} - -int Fir_Resampler_::skip_input( long count ) -{ - int remain = write_pos - buf.begin(); - int max_count = remain - width_ * stereo; - if ( count > max_count ) - count = max_count; - - remain -= count; - write_pos = &buf [remain]; - memmove( buf.begin(), &buf [count], remain * sizeof buf [0] ); - - return count; -} diff --git a/mednafen/sound/WAVRecord.cpp b/mednafen/sound/WAVRecord.cpp deleted file mode 100644 index 58885f6..0000000 --- a/mednafen/sound/WAVRecord.cpp +++ /dev/null @@ -1,99 +0,0 @@ -/* Mednafen - Multi-system Emulator - * - * This program is free software; you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation; either version 2 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA - */ - -#include "../mednafen.h" -#include "WAVRecord.h" -#include - -WAVRecord::WAVRecord(const char *path, double SoundRate_arg, uint32 SoundChan_arg) : wavfile(path, FileWrapper::MODE_WRITE_SAFE) -{ - Finished = false; - PCMBytesWritten = 0; - - SoundRate = SoundRate_arg; - SoundChan = SoundChan_arg; - - memset(&raw_headers, 0, sizeof(raw_headers)); - - MDFN_en32msb(&raw_headers[0x00], 0x52494646); // "RIFF" - // @ 0x04 = total file size - 8 bytes - MDFN_en32msb(&raw_headers[0x08], 0x57415645); // "WAVE" - - - MDFN_en32msb(&raw_headers[0x0C], 0x666d7420); // "fmt " - MDFN_en32lsb(&raw_headers[0x10], 16); - MDFN_en16lsb(&raw_headers[0x14], 1); // PCM format - MDFN_en16lsb(&raw_headers[0x16], SoundChan); // Number of sound channels - MDFN_en32lsb(&raw_headers[0x18], SoundRate); // Sampling rate - MDFN_en32lsb(&raw_headers[0x1C], SoundRate * SoundChan * sizeof(int16)); //Byte rate - MDFN_en16lsb(&raw_headers[0x20], SoundChan * sizeof(int16)); // Block("audio frame" in Mednafen) alignment - MDFN_en16lsb(&raw_headers[0x22], sizeof(int16) * 8); // Bits per sample. - - MDFN_en32msb(&raw_headers[0x24], 0x64617461); // "data" - // @ 0x28 = bytes of PCM data following - - wavfile.write(raw_headers, sizeof(raw_headers)); -} - -void WAVRecord::WriteSound(const int16 *SoundBuf, uint32 NumSoundFrames) -{ - uint32 NumSoundSamples = NumSoundFrames * SoundChan; - - while(NumSoundSamples > 0) - { - int16 swap_buf[256]; - uint32 s_this_time = std::min((uint32)NumSoundSamples, (uint32)256); - - for(uint32 i = 0; i < s_this_time; i++) - MDFN_en16lsb((uint8 *)&swap_buf[i], SoundBuf[i]); - - wavfile.write(swap_buf, s_this_time * sizeof(int16)); - PCMBytesWritten += s_this_time * sizeof(int16); - NumSoundSamples -= s_this_time; - SoundBuf += s_this_time; - } - -} - - -void WAVRecord::Finish(void) -{ - if(Finished) - return; - - MDFN_en32lsb(&raw_headers[0x04], std::min(wavfile.tell() - 8, (int64)0xFFFFFFFFLL)); - - MDFN_en32lsb(&raw_headers[0x28], std::min(PCMBytesWritten, (int64)0xFFFFFFFFLL)); - - wavfile.seek(0, SEEK_SET); - wavfile.write(raw_headers, sizeof(raw_headers)); - wavfile.close(); - - Finished = true; -} - -WAVRecord::~WAVRecord() -{ - try - { - Finish(); - } - catch(std::exception &e) - { - MDFND_PrintError(e.what()); - } -} diff --git a/mednafen/sound/WAVRecord.h b/mednafen/sound/WAVRecord.h deleted file mode 100644 index 6defbfc..0000000 --- a/mednafen/sound/WAVRecord.h +++ /dev/null @@ -1,31 +0,0 @@ -#ifndef __MDFN_WAVRECORD_H -#define __MDFN_WAVERECORD_H - -#include "../mednafen.h" -#include "../FileWrapper.h" - -class WAVRecord -{ - public: - - WAVRecord(const char *path, double SoundRate, uint32 SoundChan); - - void WriteSound(const int16 *SoundBuf, uint32 NumSoundFrames); - - void Finish(); - - ~WAVRecord(); - - private: - - FileWrapper wavfile; - bool Finished; - - uint8 raw_headers[0x2C]; - int64 PCMBytesWritten; - uint32 SoundRate; - uint32 SoundChan; -}; - - -#endif diff --git a/mednafen/sound/sound.cpp b/mednafen/sound/sound.cpp deleted file mode 100644 index afcb039..0000000 --- a/mednafen/sound/sound.cpp +++ /dev/null @@ -1,41 +0,0 @@ -#include "../mednafen.h" -#include "../sound.h" - -#if 0 - -MDFN_SoundBuffer::MDFN_SoundBuffer(void) -{ - Alloced = 0; - FramesInBuffer = 0; - Channels = 0; - buffer = NULL; -} - -void MDFN_SoundBuffer::Clear(void) -{ - FramesInBuffer = 0; -} - -void MDFN_SoundBuffer::SetChannels(int ch) -{ - assert(ch == 1 || ch == 2); - - if(buffer && ch != Channels) - { - // If going from mono to stereo, half the - if(ch == 2) - Alloced >>= 1; - else - Alloced <<= 1; - } - Channels == ch; -} - -int16 *MDFN_SoundBuffer::BeginWrite(int32 frames) -{ - assert(Channels); - - if - -} -#endif