-
-
Save unnonouno/f92dac1959599db28620 to your computer and use it in GitHub Desktop.
sse inner product
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
vmovss (%rax), %xmm1 | |
vmulss (%rdx), %xmm1, %xmm1 | |
vaddss %xmm1, %xmm0, %xmm0 | |
addq $4, %rdx | |
addq $4, %rax | |
decq %rcx | |
jne LBB0_2 |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
float dot(const vector<float>& x, | |
const vector<float>& y) { | |
float s = 0; | |
for (size_t i = 0, size = x.size(); i < size; ++i) { | |
s += x[i] * y[i]; | |
} | |
return s; | |
} | |
float dot_sse(const vector<float>& x, | |
const vector<float>& y) { | |
float s1 = 0, s2 = 0, s3 = 0, s4 = 0; | |
for (size_t i = 0, size = x.size(); i < size; i += 4) { | |
s1 += x[i] * y[i]; | |
s2 += x[i + 1] * y[i + 1]; | |
s3 += x[i + 2] * y[i + 2]; | |
s4 += x[i + 3] * y[i + 3]; | |
} | |
return s1 + s2 + s3 + s4; | |
} |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
vmovss (%rax,%rsi,4), %xmm1 | |
vinsertps $16, 12(%rax,%rsi,4), %xmm1, %xmm1 | |
vinsertps $32, 8(%rax,%rsi,4), %xmm1, %xmm1 | |
vinsertps $48, 4(%rax,%rsi,4), %xmm1, %xmm1 | |
vmovss (%rdx,%rsi,4), %xmm2 | |
vinsertps $16, 12(%rdx,%rsi,4), %xmm2, %xmm2 | |
vinsertps $32, 8(%rdx,%rsi,4), %xmm2, %xmm2 | |
vinsertps $48, 4(%rdx,%rsi,4), %xmm2, %xmm2 | |
vmulps %xmm2, %xmm1, %xmm1 | |
vaddps %xmm1, %xmm0, %xmm0 | |
addq $4, %rsi | |
cmpq %rcx, %rsi | |
jb LBB2_2 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment