Apologies if this is the wrong list.
I'm afraid I'm not much of an assembly programmer, but I was just
wondering if this generated code was 'correct', because from descriptions
of SSE that I've read, it looks like it's inefficient.
The C code:
float *vector_add4f(float va[4], const float vb[4])
{
va[0] += vb[0];
va[1] += vb[1];
va[2] += vb[2];
va[3] += vb[3];
return va;
}
Now unless my understanding is totally off, the processor should be able
to do those four additions in one by using the SSE extensions. The standard
code generated (without SSE) is as expected:
88: 55 push %ebp
89: 89 e5 mov%esp,%ebp
8b: 8b 45 08mov0x8(%ebp),%eax
8e: 8b 55 0cmov0xc(%ebp),%edx
91: d9 00 flds (%eax)
93: d8 02 fadds (%edx)
95: d9 18 fstps (%eax)
97: d9 40 04flds 0x4(%eax)
9a: d8 42 04fadds 0x4(%edx)
9d: d9 58 04fstps 0x4(%eax)
a0: d9 40 08flds 0x8(%eax)
a3: d8 42 08fadds 0x8(%edx)
a6: d9 58 08fstps 0x8(%eax)
a9: d9 40 0cflds 0xc(%eax)
ac: d8 42 0cfadds 0xc(%edx)
af: d9 58 0cfstps 0xc(%eax)
b2: c9 leave
b3: c3 ret
Using -march=pentium3 -mtune=pentium3m -mfpmath=sse, the following
is generated:
140: 55 push %ebp
141: 89 e5 mov%esp,%ebp
143: 8b 4d 08mov0x8(%ebp),%ecx
146: 8b 45 08mov0x8(%ebp),%eax
149: 8b 55 0cmov0xc(%ebp),%edx
14c: f3 0f 10 00 movss (%eax),%xmm0
150: f3 0f 58 02 addss (%edx),%xmm0
154: f3 0f 11 01 movss %xmm0,(%ecx)
158: 8b 4d 08mov0x8(%ebp),%ecx
15b: 83 c1 04add$0x4,%ecx
15e: 8b 45 08mov0x8(%ebp),%eax
161: 83 c0 04add$0x4,%eax
164: 8b 55 0cmov0xc(%ebp),%edx
167: 83 c2 04add$0x4,%edx
16a: f3 0f 10 00 movss (%eax),%xmm0
16e: f3 0f 58 02 addss (%edx),%xmm0
172: f3 0f 11 01 movss %xmm0,(%ecx)
176: 8b 4d 08mov0x8(%ebp),%ecx
179: 83 c1 08add$0x8,%ecx
17c: 8b 45 08mov0x8(%ebp),%eax
17f: 83 c0 08add$0x8,%eax
182: 8b 55 0cmov0xc(%ebp),%edx
185: 83 c2 08add$0x8,%edx
188: f3 0f 10 00 movss (%eax),%xmm0
18c: f3 0f 58 02 addss (%edx),%xmm0
190: f3 0f 11 01 movss %xmm0,(%ecx)
194: 8b 4d 08mov0x8(%ebp),%ecx
197: 83 c1 0cadd$0xc,%ecx
19a: 8b 45 08mov0x8(%ebp),%eax
19d: 83 c0 0cadd$0xc,%eax
1a0: 8b 55 0cmov0xc(%ebp),%edx
1a3: 83 c2 0cadd$0xc,%edx
1a6: f3 0f 10 00 movss (%eax),%xmm0
1aa: f3 0f 58 02 addss (%edx),%xmm0
1ae: f3 0f 11 01 movss %xmm0,(%ecx)
1b2: 8b 45 08mov0x8(%ebp),%eax
1b5: 5d pop%ebp
1b6: c3 ret
1b7: 89 f6 mov%esi,%esi
1b9: 8d bc 27 00 00 00 00lea0x0(%edi),%edi
Now, uh, isn't that four additions? Do I need to do something gcc-specific
to get it to use the 'add-packed-single' instruction to turn those four
additions into one?
MC
