On 6/10/2011 10:59 AM, S. Erisman wrote:
The _mm_* function _do_ indeed get compiled down to SSE assembly
instructions.
For reference... Here is what the non-SSE code compiles down too:
rfx_decode_YCbCr_to_RGB():
0: 55 push %ebp
1: 31 d2 xor %edx,%edx
3: 89 e5 mov %esp,%ebp
5: 57 push %edi
6: 56 push %esi
7: 53 push %ebx
8: 83 ec 0c sub $0xc,%esp
b: 90 nop
c: 8d 74 26 00 lea 0x0(%esi,%eiz,1),%esi
10: 8b 5d 0c mov 0xc(%ebp),%ebx
13: 8b 7d 10 mov 0x10(%ebp),%edi
16: 8b 45 08 mov 0x8(%ebp),%eax
19: 0f b7 0c 53 movzwl (%ebx,%edx,2),%ecx
1d: 0f b7 1c 57 movzwl (%edi,%edx,2),%ebx
21: 0f b7 34 50 movzwl (%eax,%edx,2),%esi
25: 89 d8 mov %ebx,%eax
27: 89 df mov %ebx,%edi
29: 66 c1 f8 03 sar $0x3,%ax
2d: 66 83 ee 80 sub $0xffffff80,%si
31: 66 c1 ff 05 sar $0x5,%di
35: 66 89 45 f0 mov %ax,-0x10(%ebp)
39: 89 d8 mov %ebx,%eax
3b: 66 89 7d f2 mov %di,-0xe(%ebp)
3f: 66 c1 f8 02 sar $0x2,%ax
43: 8d 3c 1e lea (%esi,%ebx,1),%edi
46: 8d 04 07 lea (%edi,%eax,1),%eax
49: 31 ff xor %edi,%edi
4b: 66 03 45 f0 add -0x10(%ebp),%ax
4f: 66 03 45 f2 add -0xe(%ebp),%ax
53: 78 0c js 61 <rfx_decode_YCbCr_to_RGB+0x61>
55: 66 3d ff 00 cmp $0xff,%ax
59: bf ff 00 00 00 mov $0xff,%edi
5e: 0f 4e f8 cmovle %eax,%edi
61: 8b 45 08 mov 0x8(%ebp),%eax
64: 66 89 3c 50 mov %di,(%eax,%edx,2)
68: 89 cf mov %ecx,%edi
6a: 89 f0 mov %esi,%eax
6c: 66 c1 ff 02 sar $0x2,%di
70: 66 89 7d ea mov %di,-0x16(%ebp)
74: 89 cf mov %ecx,%edi
76: 66 2b 45 ea sub -0x16(%ebp),%ax
7a: 66 c1 ff 05 sar $0x5,%di
7e: 66 29 f8 sub %di,%ax
81: 89 df mov %ebx,%edi
83: 66 2b 45 f0 sub -0x10(%ebp),%ax
87: 66 d1 ff sar %di
8a: 66 c1 fb 04 sar $0x4,%bx
8e: 66 29 f8 sub %di,%ax
91: 89 cf mov %ecx,%edi
93: 66 2b 45 f2 sub -0xe(%ebp),%ax
97: 66 c1 ff 04 sar $0x4,%di
9b: 66 29 d8 sub %bx,%ax
9e: 31 db xor %ebx,%ebx
a0: 66 29 f8 sub %di,%ax
a3: 78 0c js b1 <rfx_decode_YCbCr_to_RGB+0xb1>
a5: 66 3d ff 00 cmp $0xff,%ax
a9: bb ff 00 00 00 mov $0xff,%ebx
ae: 0f 4e d8 cmovle %eax,%ebx
b1: 8b 45 0c mov 0xc(%ebp),%eax
b4: 01 ce add %ecx,%esi
b6: 66 89 1c 50 mov %bx,(%eax,%edx,2)
ba: 0f b7 5d ea movzwl -0x16(%ebp),%ebx
be: 89 c8 mov %ecx,%eax
c0: 66 d1 f8 sar %ax
c3: 66 c1 f9 06 sar $0x6,%cx
c7: 8d 3c 1e lea (%esi,%ebx,1),%edi
ca: 01 c7 add %eax,%edi
cc: 31 c0 xor %eax,%eax
ce: 66 01 f9 add %di,%cx
d1: 78 0d js e0 <rfx_decode_YCbCr_to_RGB+0xe0>
d3: 66 81 f9 ff 00 cmp $0xff,%cx
d8: b8 ff 00 00 00 mov $0xff,%eax
dd: 0f 4e c1 cmovle %ecx,%eax
e0: 8b 7d 10 mov 0x10(%ebp),%edi
e3: 66 89 04 57 mov %ax,(%edi,%edx,2)
e7: 83 c2 01 add $0x1,%edx
ea: 81 fa 00 10 00 00 cmp $0x1000,%edx
f0: 0f 85 1a ff ff ff jne 10 <rfx_decode_YCbCr_to_RGB+0x10>
f6: 83 c4 0c add $0xc,%esp
f9: 5b pop %ebx
fa: 5e pop %esi
fb: 5f pop %edi
fc: 5d pop %ebp
fd: c3 ret
fe: 66 90 xchg %ax,%ax
Thanks,
Steve
------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
_______________________________________________
Freerdp-devel mailing list
Freerdp-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freerdp-devel
