You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 

177 lines
9.8 KiB

;* ======================================================================== *;
;* TEXAS INSTRUMENTS, INC. *;
;* *;
;* IMGLIB DSP Image/Video Processing Library *;
;* *;
;* Release: Revision 1.04b *;
;* CVS Revision: 1.3 Fri Oct 25 00:29:08 2002 (UTC) *;
;* Snapshot date: 23-Oct-2003 *;
;* *;
;* This library contains proprietary intellectual property of Texas *;
;* Instruments, Inc. The library and its source code are protected by *;
;* various copyrights, and portions may also be protected by patents or *;
;* other legal protections. *;
;* *;
;* This software is licensed for use with Texas Instruments TMS320 *;
;* family DSPs. This license was provided to you prior to installing *;
;* the software. You may review this license by consulting the file *;
;* TI_license.PDF which accompanies the files in this library. *;
;* ------------------------------------------------------------------------ *;
;* Copyright (C) 2003 Texas Instruments, Incorporated. *;
;* All Rights Reserved. *;
;* ======================================================================== *;
;* ======================================================================== *;
;* Assembler compatibility shim for assembling 4.30 and later code on *;
;* tools prior to 4.30. *;
;* ======================================================================== *;
.if $isdefed(".ASSEMBLER_VERSION")
.asg .ASSEMBLER_VERSION, $asmver
.else
.asg 0, $asmver
.endif
.if ($asmver < 430)
.asg B, CALL ; Function Call
.asg B, RET ; Return from a Function
.asg B, CALLRET ; Function call with Call / Ret chaining.
.if .TMS320C6400
.asg BNOP, CALLNOP ; C64x BNOP as a Fn. Call
.asg BNOP, RETNOP ; C64x BNOP as a Fn. Return
.asg BNOP, CRNOP ; C64x Fn call w/, Call/Ret chaining via BNOP.
.endif
.asg , .asmfunc ; .func equivalent for hand-assembly code
.asg , .endasmfunc ; .endfunc equivalent for hand-assembly code
.endif
;* ======================================================================== *;
;* End of assembler compatibility shim. *;
;* ======================================================================== *;
* ========================================================================= *
* NAME *
* IMG_idct_8x8 -- Wrapper for idct_8x8_12q4, Little Endian. *
* *
* REVISION DATE *
* 24-Oct-2002 *
* *
* USAGE *
* This routine is C callable, and has the following C prototype: *
* *
* void IMG_idct_8x8(short idct_data[], unsigned num_idcts) *
* *
* DESCRIPTION *
* This function provides a wrapper around the IMG_idct_8x8_12q4 *
* function. It reads 12Q0 values in from the input array, and *
* writes them back as 12Q4. Once all the values have been adjusted, *
* the function then branches directly to IMG_idct_8x8_12q4. *
* *
* It is strongly recommended that you adapt your code to call *
* IMG_idct_8x8_12q4 directly. Calling this wrapper adds an extra *
* 16 cycles per block of processing overhead, as well as 116 *
* bytes of additional code. *
* *
* ASSUMPTIONS *
* This is a LITTLE ENDIAN implementation. *
* The input array must be aligned on a double-word boundary. *
* *
* MEMORY NOTE *
* No bank conflicts occur. *
* *
* NOTES *
* This code locks out interrupts for its entire duration. *
* It is, however, fully interrupt tolerant. *
* *
* CYCLES *
* cycles = 16 + 16 * num_idcts, for num_idcts > 0 *
* *
* CODESIZE *
* 116 bytes *
* ------------------------------------------------------------------------- *
* Copyright (c) 2003 Texas Instruments, Incorporated. *
* All Rights Reserved. *
* ========================================================================= *
.global _IMG_idct_8x8_12q4
.global _IMG_idct_8x8
.sect ".text:_idct_wrap"
.global _IMG_idct_8x8
_IMG_idct_8x8:
* ===================== SYMBOLIC REGISTER ASSIGNMENTS ===================== *
.asg A5, A_ptr
.asg B5, B_ptr
.asg A1, A_i
.asg B7, B_msk
.asg B17, B_x76
.asg B16, B_x54
.asg A9, A_x32
.asg A8, A_x10
.asg B19, B_y76
.asg B18, B_y54
.asg A17, A_y32
.asg A16, A_y10
.asg B29, B_z76
.asg B28, B_z54
.asg A27, A_z32
.asg A26, A_z10
.asg B0, B_p
.asg B20, B_idct_8x8_12q4
* ========================================================================= *
* =========================== PIPE LOOP PROLOG ============================ *
MV A4, A_ptr
|| ADD A4, 8, B_ptr
B .S1 loop
|| LDDW .D1T1 *A_ptr++[2],A_x32:A_x10 ;[ 1,1]
|| LDDW .D2T2 *B_ptr++[2],B_x76:B_x54 ;[ 1,1]
|| MVKL .S2 0xFFF0FFF0, B_msk
MVKH .S2 0xFFF0FFF0, B_msk
|| SHL .S1X B4, 3, A_i
LDDW .D1T1 *A_ptr++[2],A_x32:A_x10 ;[ 1,2]
|| LDDW .D2T2 *B_ptr++[2],B_x76:B_x54 ;[ 1,2]
|| MVKL .S2 _IMG_idct_8x8_12q4, B_idct_8x8_12q4
||[A_i] BDEC .S1 loop, A_i
MVKH .S2 _IMG_idct_8x8_12q4, B_idct_8x8_12q4
|| MVK .L2 2, B_p
; ===== 2 prolog stages collapsed
* =========================== PIPE LOOP KERNEL ============================ *
loop:
AND .L1X A_y32, B_msk, A_z32 ;[ 9,1]
|| AND .L2 B_y76, B_msk, B_z76 ;[ 9,1]
|| ROTL .M1 A_x32, 4, A_y32 ;[ 7,2]
|| SHL .S2 B_x54, 4, B_y54 ;[ 7,2]
||[ A_i]BDEC .S1 loop, A_i ;[ 5,3]
|| LDDW .D1T1 *A_ptr++[2],A_x32:A_x10 ;[ 1,5]
|| LDDW .D2T2 *B_ptr++[2],B_x76:B_x54 ;[ 1,5]
[ B_p]SUB .L2 B_p, 1, B_p ;[10,1]
||[!B_p]STDW .D1T1 A_z32:A_z10, *-A_ptr[10] ;[10,1]
||[!B_p]STDW .D2T2 B_z76:B_z54, *-B_ptr[10] ;[10,1]
|| AND .L1X A_y10, B_msk, A_z10 ;[ 8,2]
|| AND .S2 B_y54, B_msk, B_z54 ;[ 8,2]
|| SHL .S1 A_x10, 4, A_y10 ;[ 6,3]
|| ROTL .M2 B_x76, 4, B_y76 ;[ 6,3]
* ========================================================================= *
; ===== epilog collapsed completely
CRNOP .S2 B_idct_8x8_12q4, 5
* ========================================================================= *
* End of file: img_idct_8x8.asm *
* ------------------------------------------------------------------------- *
* Copyright (c) 2003 Texas Instruments, Incorporated. *
* All Rights Reserved. *
* ========================================================================= *