Improvements from Matjaz Thaler

ac3dec/downmix_i386.S

@@ -0,0 +1,92 @@
+/* This is basicly gcc generated.
+ * Only the floating point rounding mode loads and saves
+ * are removed in the stream_sample_to_s16 functions.
+ */                                                                            
+
+#ifdef __i386__
+
+	.file	"downmix.c"
+	.version	"01.01"
+gcc2_compiled.:
+.text
+	.align 4
+.globl stream_sample_2ch_to_s16_i386
+	.type	 stream_sample_2ch_to_s16_i386,@function
+stream_sample_2ch_to_s16_i386:
+	pushl %ebp
+	movl %esp,%ebp
+	subl $28,%esp
+	pushl %edi
+	pushl %esi
+	pushl %ebx
+	movl 8(%ebp),%edx
+	movl 12(%ebp),%ebx
+	movl 16(%ebp),%ecx
+	movl $255,%esi
+	.p2align 4,,7
+.L373:
+	flds (%ebx)
+	fistpl -8(%ebp)
+	movl -8(%ebp),%eax
+	movw %ax,(%edx)
+	addl $2,%edx
+	addl $4,%ebx
+	flds (%ecx)
+	fistpl -8(%ebp)
+	movl -8(%ebp),%eax
+	movw %ax,(%edx)
+	addl $4,%ecx
+	addl $2,%edx
+	decl %esi
+	jns .L373
+	popl %ebx
+	popl %esi
+	popl %edi
+	leave
+	ret
+.Lfe6:
+	.size	 stream_sample_2ch_to_s16_i386,.Lfe6-stream_sample_2ch_to_s16_i386
+.section	.rodata
+	.align 4
+.LC46:
+	.long 0x3f350481
+.text
+	.align 4
+.globl stream_sample_1ch_to_s16_i386
+	.type	 stream_sample_1ch_to_s16_i386,@function
+stream_sample_1ch_to_s16_i386:
+	pushl %ebp
+	movl %esp,%ebp
+	subl $16,%esp
+	pushl %esi
+	pushl %ebx
+	movl 8(%ebp),%edx
+	movl 12(%ebp),%ecx
+	flds .LC46
+	movl $255,%ebx
+	.p2align 4,,7
+.L379:
+	flds (%ecx)
+	fmul %st(1),%st
+	fistpl -8(%ebp)
+	movl -8(%ebp),%eax
+	movw %ax,-2(%ebp)
+	addl $4,%ecx
+	flds -2(%ebp)
+	fistpl -8(%ebp)
+	movl -8(%ebp),%eax
+	movw %ax,(%edx)
+	addl $2,%edx
+	movw %ax,(%edx)
+	addl $2,%edx
+	decl %ebx
+	jns .L379
+	fstp %st(0)
+	popl %ebx
+	popl %esi
+	leave
+	ret
+.Lfe7:
+	.size	 stream_sample_1ch_to_s16_i386,.Lfe7-stream_sample_1ch_to_s16_i386
+	.ident	"GCC: (GNU) 2.95.3 19991030 (prerelease)"
+#endif