Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(564)

Unified Diff: simd/jcqnts2f-64.asm

Issue 1953443002: Update to libjpeg_turbo 1.4.90 (Closed) Base URL: https://chromium.googlesource.com/chromium/deps/libjpeg_turbo.git@master
Patch Set: Created 4 years, 7 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View side-by-side diff with in-line comments
Download patch
« no previous file with comments | « simd/jcqnts2f.asm ('k') | simd/jcqnts2i.asm » ('j') | no next file with comments »
Expand Comments ('e') | Collapse Comments ('c') | Show Comments Hide Comments ('s')
Index: simd/jcqnts2f-64.asm
diff --git a/simd/jcqnts2f-64.asm b/simd/jcqnts2f-64.asm
deleted file mode 100644
index 0bc73bcf9290a612f994bb61d9882b76f305ccb0..0000000000000000000000000000000000000000
--- a/simd/jcqnts2f-64.asm
+++ /dev/null
@@ -1,158 +0,0 @@
-;
-; jcqnts2f-64.asm - sample data conversion and quantization (64-bit SSE & SSE2)
-;
-; Copyright 2009 Pierre Ossman <ossman@cendio.se> for Cendio AB
-; Copyright 2009 D. R. Commander
-;
-; Based on
-; x86 SIMD extension for IJG JPEG library
-; Copyright (C) 1999-2006, MIYASAKA Masaru.
-; For conditions of distribution and use, see copyright notice in jsimdext.inc
-;
-; This file should be assembled with NASM (Netwide Assembler),
-; can *not* be assembled with Microsoft's MASM or any compatible
-; assembler (including Borland's Turbo Assembler).
-; NASM is available from http://nasm.sourceforge.net/ or
-; http://sourceforge.net/project/showfiles.php?group_id=6208
-;
-; [TAB8]
-
-%include "jsimdext.inc"
-%include "jdct.inc"
-
-; --------------------------------------------------------------------------
- SECTION SEG_TEXT
- BITS 64
-;
-; Load data into workspace, applying unsigned->signed conversion
-;
-; GLOBAL(void)
-; jsimd_convsamp_float_sse2 (JSAMPARRAY sample_data, JDIMENSION start_col,
-; FAST_FLOAT * workspace);
-;
-
-; r10 = JSAMPARRAY sample_data
-; r11 = JDIMENSION start_col
-; r12 = FAST_FLOAT * workspace
-
- align 16
- global EXTN(jsimd_convsamp_float_sse2) PRIVATE
-
-EXTN(jsimd_convsamp_float_sse2):
- push rbp
- mov rax,rsp
- mov rbp,rsp
- collect_args
- push rbx
-
- pcmpeqw xmm7,xmm7
- psllw xmm7,7
- packsswb xmm7,xmm7 ; xmm7 = PB_CENTERJSAMPLE (0x808080..)
-
- mov rsi, r10
- mov eax, r11d
- mov rdi, r12
- mov rcx, DCTSIZE/2
-.convloop:
- mov rbx, JSAMPROW [rsi+0*SIZEOF_JSAMPROW] ; (JSAMPLE *)
- mov rdx, JSAMPROW [rsi+1*SIZEOF_JSAMPROW] ; (JSAMPLE *)
-
- movq xmm0, XMM_MMWORD [rbx+rax*SIZEOF_JSAMPLE]
- movq xmm1, XMM_MMWORD [rdx+rax*SIZEOF_JSAMPLE]
-
- psubb xmm0,xmm7 ; xmm0=(01234567)
- psubb xmm1,xmm7 ; xmm1=(89ABCDEF)
-
- punpcklbw xmm0,xmm0 ; xmm0=(*0*1*2*3*4*5*6*7)
- punpcklbw xmm1,xmm1 ; xmm1=(*8*9*A*B*C*D*E*F)
-
- punpcklwd xmm2,xmm0 ; xmm2=(***0***1***2***3)
- punpckhwd xmm0,xmm0 ; xmm0=(***4***5***6***7)
- punpcklwd xmm3,xmm1 ; xmm3=(***8***9***A***B)
- punpckhwd xmm1,xmm1 ; xmm1=(***C***D***E***F)
-
- psrad xmm2,(DWORD_BIT-BYTE_BIT) ; xmm2=(0123)
- psrad xmm0,(DWORD_BIT-BYTE_BIT) ; xmm0=(4567)
- cvtdq2ps xmm2,xmm2 ; xmm2=(0123)
- cvtdq2ps xmm0,xmm0 ; xmm0=(4567)
- psrad xmm3,(DWORD_BIT-BYTE_BIT) ; xmm3=(89AB)
- psrad xmm1,(DWORD_BIT-BYTE_BIT) ; xmm1=(CDEF)
- cvtdq2ps xmm3,xmm3 ; xmm3=(89AB)
- cvtdq2ps xmm1,xmm1 ; xmm1=(CDEF)
-
- movaps XMMWORD [XMMBLOCK(0,0,rdi,SIZEOF_FAST_FLOAT)], xmm2
- movaps XMMWORD [XMMBLOCK(0,1,rdi,SIZEOF_FAST_FLOAT)], xmm0
- movaps XMMWORD [XMMBLOCK(1,0,rdi,SIZEOF_FAST_FLOAT)], xmm3
- movaps XMMWORD [XMMBLOCK(1,1,rdi,SIZEOF_FAST_FLOAT)], xmm1
-
- add rsi, byte 2*SIZEOF_JSAMPROW
- add rdi, byte 2*DCTSIZE*SIZEOF_FAST_FLOAT
- dec rcx
- jnz short .convloop
-
- pop rbx
- uncollect_args
- pop rbp
- ret
-
-
-; --------------------------------------------------------------------------
-;
-; Quantize/descale the coefficients, and store into coef_block
-;
-; GLOBAL(void)
-; jsimd_quantize_float_sse2 (JCOEFPTR coef_block, FAST_FLOAT * divisors,
-; FAST_FLOAT * workspace);
-;
-
-; r10 = JCOEFPTR coef_block
-; r11 = FAST_FLOAT * divisors
-; r12 = FAST_FLOAT * workspace
-
- align 16
- global EXTN(jsimd_quantize_float_sse2) PRIVATE
-
-EXTN(jsimd_quantize_float_sse2):
- push rbp
- mov rax,rsp
- mov rbp,rsp
- collect_args
-
- mov rsi, r12
- mov rdx, r11
- mov rdi, r10
- mov rax, DCTSIZE2/16
-.quantloop:
- movaps xmm0, XMMWORD [XMMBLOCK(0,0,rsi,SIZEOF_FAST_FLOAT)]
- movaps xmm1, XMMWORD [XMMBLOCK(0,1,rsi,SIZEOF_FAST_FLOAT)]
- mulps xmm0, XMMWORD [XMMBLOCK(0,0,rdx,SIZEOF_FAST_FLOAT)]
- mulps xmm1, XMMWORD [XMMBLOCK(0,1,rdx,SIZEOF_FAST_FLOAT)]
- movaps xmm2, XMMWORD [XMMBLOCK(1,0,rsi,SIZEOF_FAST_FLOAT)]
- movaps xmm3, XMMWORD [XMMBLOCK(1,1,rsi,SIZEOF_FAST_FLOAT)]
- mulps xmm2, XMMWORD [XMMBLOCK(1,0,rdx,SIZEOF_FAST_FLOAT)]
- mulps xmm3, XMMWORD [XMMBLOCK(1,1,rdx,SIZEOF_FAST_FLOAT)]
-
- cvtps2dq xmm0,xmm0
- cvtps2dq xmm1,xmm1
- cvtps2dq xmm2,xmm2
- cvtps2dq xmm3,xmm3
-
- packssdw xmm0,xmm1
- packssdw xmm2,xmm3
-
- movdqa XMMWORD [XMMBLOCK(0,0,rdi,SIZEOF_JCOEF)], xmm0
- movdqa XMMWORD [XMMBLOCK(1,0,rdi,SIZEOF_JCOEF)], xmm2
-
- add rsi, byte 16*SIZEOF_FAST_FLOAT
- add rdx, byte 16*SIZEOF_FAST_FLOAT
- add rdi, byte 16*SIZEOF_JCOEF
- dec rax
- jnz short .quantloop
-
- uncollect_args
- pop rbp
- ret
-
-; For some reason, the OS X linker does not honor the request to align the
-; segment unless we do this.
- align 16
« no previous file with comments | « simd/jcqnts2f.asm ('k') | simd/jcqnts2i.asm » ('j') | no next file with comments »

Powered by Google App Engine
This is Rietveld 408576698