potok/opencv
1
0
Fork 0
mirror of https://github.com/opencv/opencv.git synced 2026-07-31 00:03:03 +04:00
Code Packages Releases Wiki Activity

core: vectorize dotProd_32s

Browse Source 
Use 4x FMA chains to sum on SIMD 128 FP64 targets. On
x86 this showed about 1.4x improvement.

For PPC, do a full multiply (32x32->64b), convert to DP
then accumulate. This may be slightly less precise for
some inputs. But is 1.5x faster than the above which
is about 1.5x than the FMA above for ~2.5x speedup.
This commit is contained in:
Paul E. Murphy
2019-08-20 11:26:38 -05:00
parent 7295983964
commit 33fb253a66
2 changed files with 38 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/core/include/opencv2/core/hal/intrin_vsx.hpp
+9
View File
@@ -1051,6 +1051,15 @@ inline v_float64x2 v_cvt_f64(const v_float32x4& a)
inline v_float64x2 v_cvt_f64_high(const v_float32x4& a)
{ return v_float64x2(vec_cvfo(vec_mergel(a.val, a.val))); }
// The altivec intrinsic is missing for this 2.06 insn
inline v_float64x2 v_cvt_f64(const v_int64x2& a)
{
vec_double2 out;
__asm__ ("xvcvsxddp %x0,%x1" : "=wa"(out) : "wa"(a.val));
return v_float64x2(out);
}
////////////// Lookup table access ////////////////////
inline v_int8x16 v_lut(const schar* tab, const int* idx)