Test whether removing a cast still hurts performance.
BUG=499241 TEST=none TBR=andrew Review URL: https://codereview.webrtc.org/1206653002 Cr-Commit-Position: refs/heads/master@{#9491}
This commit is contained in:
@ -23,10 +23,8 @@ void WebRtcSpl_CrossCorrelationC(int32_t* cross_correlation,
|
|||||||
for (i = 0; i < dim_cross_correlation; i++) {
|
for (i = 0; i < dim_cross_correlation; i++) {
|
||||||
int32_t corr = 0;
|
int32_t corr = 0;
|
||||||
/* Unrolling doesn't seem to improve performance. */
|
/* Unrolling doesn't seem to improve performance. */
|
||||||
for (j = 0; j < dim_seq; j++) {
|
for (j = 0; j < dim_seq; j++)
|
||||||
// It's not clear why casting |right_shifts| here helps performance.
|
corr += (seq1[j] * seq2[j]) >> right_shifts;
|
||||||
corr += (seq1[j] * seq2[j]) >> (int16_t)right_shifts;
|
|
||||||
}
|
|
||||||
seq2 += step_seq2;
|
seq2 += step_seq2;
|
||||||
*cross_correlation++ = corr;
|
*cross_correlation++ = corr;
|
||||||
}
|
}
|
||||||
|
Reference in New Issue
Block a user