#include "stdafx.h" #pragma hdrstop /*************************************************************************** * * INTEL Corporation Proprietary Information * * * Copyright (c) 1996 Intel Corporation. * All rights reserved. * *************************************************************************** */ /* * jfdctfst.c * * Copyright (C) 1994-1996, Thomas G. Lane. * This file is part of the Independent JPEG Group's software. * For conditions of distribution and use, see the accompanying README file. * * This file contains a fast, not so accurate integer implementation of the * forward DCT (Discrete Cosine Transform). * * A 2-D DCT can be done by 1-D DCT on each row followed by 1-D DCT * on each column. Direct algorithms are also available, but they are * much more complex and seem not to be any faster when reduced to code. * * This implementation is based on Arai, Agui, and Nakajima's algorithm for * scaled DCT. Their original paper (Trans. IEICE E-71(11):1095) is in * Japanese, but the algorithm is described in the Pennebaker & Mitchell * JPEG textbook (see REFERENCES section in file README). The following code * is based directly on figure 4-8 in P&M. * While an 8-point DCT cannot be done in less than 11 multiplies, it is * possible to arrange the computation so that many of the multiplies are * simple scalings of the final outputs. These multiplies can then be * folded into the multiplications or divisions by the JPEG quantization * table entries. The AA&N method leaves only 5 multiplies and 29 adds * to be done in the DCT itself. * The primary disadvantage of this method is that with fixed-point math, * accuracy is lost due to imprecise representation of the scaled * quantization values. The smaller the quantization table entry, the less * precise the scaled value, so this implementation does worse with high- * quality-setting files than with low-quality ones. */ #define JPEG_INTERNALS #include "jinclude.h" #include "jpeglib.h" #include "jdct.h" /* Private declarations for DCT subsystem */ #ifdef DCT_IFAST_SUPPORTED /* * This module is specialized to the case DCTSIZE = 8. */ #if DCTSIZE != 8 Sorry, this code only copes with 8x8 DCTs. /* deliberate syntax err */ #endif /* Scaling decisions are generally the same as in the LL&M algorithm; * see jfdctint.c for more details. However, we choose to descale * (right shift) multiplication products as soon as they are formed, * rather than carrying additional fractional bits into subsequent additions. * This compromises accuracy slightly, but it lets us save a few shifts. * More importantly, 16-bit arithmetic is then adequate (for 8-bit samples) * everywhere except in the multiplications proper; this saves a good deal * of work on 16-bit-int machines. * * Again to save a few shifts, the intermediate results between pass 1 and * pass 2 are not upscaled, but are represented only to integral precision. * * A final compromise is to represent the multiplicative constants to only * 8 fractional bits, rather than 13. This saves some shifting work on some * machines, and may also reduce the cost of multiplication (since there * are fewer one-bits in the constants). */ #define CONST_BITS 8 /* Some C compilers fail to reduce "FIX(constant)" at compile time, thus * causing a lot of useless floating-point operations at run time. * To get around this we use the following pre-calculated constants. * If you change CONST_BITS you may want to add appropriate values. * (With a reasonable C compiler, you can just rely on the FIX() macro...) */ #if CONST_BITS == 8 #define FIX_0_382683433 98 /* FIX(0.382683433) */ #define FIX_0_541196100 139 /* FIX(0.541196100) */ #define FIX_0_707106781 181 /* FIX(0.707106781) */ #define FIX_1_306562965 334 /* FIX(1.306562965) */ #else #define FIX_0_382683433 FIX(0.382683433) #define FIX_0_541196100 FIX(0.541196100) #define FIX_0_707106781 FIX(0.707106781) #define FIX_1_306562965 FIX(1.306562965) #endif /* We can gain a little more speed, with a further compromise in accuracy, * by omitting the addition in a descaling shift. This yields an incorrectly * rounded result half the time... */ // The assembly version makes this compromise. //#ifndef USE_ACCURATE_ROUNDING //#undef DESCALE //#define DESCALE(x,n) RIGHT_SHIFT(x, n) //#endif #define DCTWIDTH 32 #define DATASIZE 4 /* Multiply a DCTELEM variable by an INT32 constant, and immediately * descale to yield a DCTELEM result. */ #define MULTIPLY(var,const) ((DCTELEM) DESCALE((var) * (const), CONST_BITS)) #if _MSC_FULL_VER >= 13008827 && defined(_M_IX86) #pragma warning(push) #pragma warning(disable:4731) // EBP modified with inline asm #endif /* * Perform the forward DCT on one block of samples. */ GLOBAL(void) pfdct8x8aan (DCTELEM * data) { DCTELEM tmp4, tmp6, tmp7; int counter; __asm{ /* Pass 1: process rows. */ // dataptr = data; mov esi, [data] mov counter, 8 // for (ctr = DCTSIZE-1; ctr >= 0; ctr--) { // tmp0 = dataptr[0] + dataptr[7]; // tmp7 = dataptr[0] - dataptr[7]; // tmp1 = dataptr[1] + dataptr[6]; // tmp6 = dataptr[1] - dataptr[6]; // tmp2 = dataptr[2] + dataptr[5]; // tmp5 = dataptr[2] - dataptr[5]; // tmp3 = dataptr[3] + dataptr[4]; // tmp4 = dataptr[3] - dataptr[4]; StartRow: mov eax, [esi][DATASIZE*0] mov ebx, [esi][DATASIZE*7] mov edx, eax add eax, ebx ; eax = tmp0 sub edx, ebx ; edx = tmp7 mov ebx, [esi][DATASIZE*3] mov ecx, [esi][DATASIZE*4] mov edi, ebx add ebx, ecx ; ebx = tmp3 sub edi, ecx ; edi = tmp4 mov tmp4, edi mov tmp7, edx /* Even part */ // tmp10 = tmp0 + tmp3; // tmp13 = tmp0 - tmp3; // tmp11 = tmp1 + tmp2; // tmp12 = tmp1 - tmp2; mov ecx, eax add eax, ebx ; eax = tmp10 sub ecx, ebx ; ecx = tmp13 mov edx, [esi][DATASIZE*1] mov edi, [esi][DATASIZE*6] mov ebx, edx add edx, edi ; edx = tmp1 sub ebx, edi ; ebx = tmp6 mov tmp6, ebx push ebp mov edi, [esi][DATASIZE*2] mov ebp, [esi][DATASIZE*5] mov ebx, edi add edi, ebp ; edi = tmp2 sub ebx, ebp ; ebx = tmp5 mov ebp, edx add edx, edi ; edx = tmp11 sub ebp, edi ; ebp = tmp12 // dataptr[0] = tmp10 + tmp11; /* phase 3 */ // dataptr[4] = tmp10 - tmp11; mov edi, eax add eax, edx ; eax = tmp10 + tmp11 sub edi, edx ; edi = tmp10 - tmp11 add ebp, ecx ; ebp = tmp12 + tmp13 // z1 = MULTIPLY(tmp12 + tmp13, FIX_0_707106781); /* c4 */ imul ebp, FIX_0_707106781 ; ebp = z1 sar ebp, 8 mov [esi][DATASIZE*0], eax // dataptr[2] = tmp13 + z1; /* phase 5 */ // dataptr[6] = tmp13 - z1; mov eax, ecx add ecx, ebp sub eax, ebp pop ebp mov [esi][DATASIZE*4], edi mov [esi][DATASIZE*2], ecx mov [esi][DATASIZE*6], eax mov edi, tmp4 /* Odd part */ // tmp10 = tmp4 + tmp5; /* phase 2 */ // tmp11 = tmp5 + tmp6; // tmp12 = tmp6 + tmp7; mov ecx, tmp6 mov edx, tmp7 add edi, ebx ; edi = tmp10 add ebx, ecx ; ebx = tmp11 // z3 = MULTIPLY(tmp11, FIX_0_707106781); /* c4 */ // z11 = tmp7 + z3; /* phase 5 */ // z13 = tmp7 - z3; imul ebx, FIX_0_707106781 ; ebx = z3 sar ebx, 8 add ecx, edx ; ecx = tmp12 mov eax, edx add edx, ebx ; edx = z11 sub eax, ebx ; eax = z13 mov ebx, edi /* The rotator is modified from fig 4-8 to avoid extra negations. */ // z5 = MULTIPLY(tmp10 - tmp12, FIX_0_382683433); /* c6 */ // z2 = MULTIPLY(tmp10, FIX_0_541196100) + z5; /* c2-c6 */ // z4 = MULTIPLY(tmp12, FIX_1_306562965) + z5; /* c2+c6 */ imul ebx, FIX_0_541196100 sar ebx, 8 sub edi, ecx ; edi = tmp10 - tmp12 imul edi, FIX_0_382683433 ; edi = z5 sar edi, 8 add esi, 32 imul ecx, FIX_1_306562965 sar ecx, 8 add ebx, edi ; ebx = z2 add ecx, edi ; ecx = z4 mov edi, eax // dataptr[5] = z13 + z2; /* phase 6 */ // dataptr[3] = z13 - z2; // dataptr[1] = z11 + z4; // dataptr[7] = z11 - z4; add eax, ebx ; eax = z13 + z2 sub edi, ebx ; edi = z13 - z2 mov [esi][DATASIZE*5-32], eax mov ebx, edx mov [esi][DATASIZE*3-32], edi add edx, ecx ; edx = z11 + z4 mov [esi][DATASIZE*1-32], edx sub ebx, ecx ; ebx = z11 - z4 mov ecx, counter mov [esi][DATASIZE*7-32], ebx dec ecx mov counter, ecx jnz StartRow // dataptr += DCTSIZE; /* advance pointer to next row */ // } /* Pass 2: process columns.*/ // dataptr = data; mov esi, [data] mov counter, 8 // for (ctr = DCTSIZE-1; ctr >= 0; ctr--) { // tmp0 = dataptr[DCTSIZE*0] + dataptr[DCTSIZE*7]; // tmp7 = dataptr[DCTSIZE*0] - dataptr[DCTSIZE*7]; // tmp1 = dataptr[DCTSIZE*1] + dataptr[DCTSIZE*6]; // tmp6 = dataptr[DCTSIZE*1] - dataptr[DCTSIZE*6]; // tmp2 = dataptr[DCTSIZE*2] + dataptr[DCTSIZE*5]; // tmp5 = dataptr[DCTSIZE*2] - dataptr[DCTSIZE*5]; // tmp3 = dataptr[DCTSIZE*3] + dataptr[DCTSIZE*4]; // tmp4 = dataptr[DCTSIZE*3] - dataptr[DCTSIZE*4]; StartCol: mov eax, [esi][DCTWIDTH*0] mov ebx, [esi][DCTWIDTH*7] mov edx, eax add eax, ebx ; eax = tmp0 sub edx, ebx ; edx = tmp7 mov ebx, [esi][DCTWIDTH*3] mov ecx, [esi][DCTWIDTH*4] mov edi, ebx add ebx, ecx ; ebx = tmp3 sub edi, ecx ; edi = tmp4 mov tmp4, edi mov tmp7, edx /* Even part */ // tmp10 = tmp0 + tmp3; // tmp13 = tmp0 - tmp3; // tmp11 = tmp1 + tmp2; // tmp12 = tmp1 - tmp2; mov ecx, eax add eax, ebx ; eax = tmp10 sub ecx, ebx ; ecx = tmp13 mov edx, [esi][DCTWIDTH*1] mov edi, [esi][DCTWIDTH*6] mov ebx, edx add edx, edi ; edx = tmp1 sub ebx, edi ; ebx = tmp6 mov tmp6, ebx push ebp mov edi, [esi][DCTWIDTH*2] mov ebp, [esi][DCTWIDTH*5] mov ebx, edi add edi, ebp ; edi = tmp2 sub ebx, ebp ; ebx = tmp5 mov ebp, edx add edx, edi ; edx = tmp11 sub ebp, edi ; ebp = tmp12 // dataptr[DCTSIZE*0] = tmp10 + tmp11; /* phase 3 */ // dataptr[DCTSIZE*4] = tmp10 - tmp11; mov edi, eax add eax, edx ; eax = tmp10 + tmp11 sub edi, edx ; edi = tmp10 - tmp11 add ebp, ecx ; ebp = tmp12 + tmp13 // z1 = MULTIPLY(tmp12 + tmp13, FIX_0_707106781); /* c4 */ imul ebp, FIX_0_707106781 ; ebp = z1 sar ebp, 8 mov [esi][DCTWIDTH*0], eax // dataptr[DCTSIZE*2] = tmp13 + z1; /* phase 5 */ // dataptr[DCTSIZE*6] = tmp13 - z1; mov eax, ecx add ecx, ebp sub eax, ebp pop ebp mov [esi][DCTWIDTH*4], edi mov [esi][DCTWIDTH*2], ecx mov [esi][DCTWIDTH*6], eax mov edi, tmp4 /* Odd part */ // tmp10 = tmp4 + tmp5; /* phase 2 */ // tmp11 = tmp5 + tmp6; // tmp12 = tmp6 + tmp7; mov ecx, tmp6 mov edx, tmp7 add edi, ebx ; edi = tmp10 add ebx, ecx ; ebx = tmp11 // z3 = MULTIPLY(tmp11, FIX_0_707106781); /* c4 */ // z11 = tmp7 + z3; /* phase 5 */ // z13 = tmp7 - z3; imul ebx, FIX_0_707106781 ; ebx = z3 sar ebx, 8 add ecx, edx ; ecx = tmp12 mov eax, edx add edx, ebx ; edx = z11 sub eax, ebx ; eax = z13 mov ebx, edi /* The rotator is modified from fig 4-8 to avoid extra negations. */ // z5 = MULTIPLY(tmp10 - tmp12, FIX_0_382683433); /* c6 */ // z2 = MULTIPLY(tmp10, FIX_0_541196100) + z5; /* c2-c6 */ // z4 = MULTIPLY(tmp12, FIX_1_306562965) + z5; /* c2+c6 */ imul ebx, FIX_0_541196100 sar ebx, 8 sub edi, ecx ; edi = tmp10 - tmp12 imul edi, FIX_0_382683433 ; edi = z5 sar edi, 8 add esi, 4 imul ecx, FIX_1_306562965 sar ecx, 8 add ebx, edi ; ebx = z2 add ecx, edi ; ecx = z4 mov edi, eax // dataptr[DCTSIZE*5] = z13 + z2; /* phase 6 */ // dataptr[DCTSIZE*3] = z13 - z2; // dataptr[DCTSIZE*1] = z11 + z4; // dataptr[DCTSIZE*7] = z11 - z4; add eax, ebx ; eax = z13 + z2 sub edi, ebx ; edi = z13 - z2 mov [esi][DCTWIDTH*5-4], eax mov ebx, edx mov [esi][DCTWIDTH*3-4], edi add edx, ecx ; edx = z11 + z4 mov [esi][DCTWIDTH*1-4], edx sub ebx, ecx ; ebx = z11 - z4 mov ecx, counter mov [esi][DCTWIDTH*7-4], ebx dec ecx mov counter, ecx jnz StartCol } //end asm // dataptr++; /* advance pointer to next column */ // } } #if _MSC_FULL_VER >= 13008827 #pragma warning(pop) #endif #endif /* DCT_ISLOW_SUPPORTED */