public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc r12-5420] rs6000: Add optimizations for _mm_sad_epu8
@ 2021-11-19 20:02 Paul Clarke
0 siblings, 0 replies; only message in thread
From: Paul Clarke @ 2021-11-19 20:02 UTC (permalink / raw)
To: gcc-cvs
https://gcc.gnu.org/g:fc6c6f64ecff376902e7e1ef295f2d8518407ab5
commit r12-5420-gfc6c6f64ecff376902e7e1ef295f2d8518407ab5
Author: Paul A. Clarke <pc@us.ibm.com>
Date: Fri Oct 22 12:09:43 2021 -0500
rs6000: Add optimizations for _mm_sad_epu8
Power9 ISA added `vabsdub` instruction which is realized in the
`vec_absd` instrinsic.
Use `vec_absd` for `_mm_sad_epu8` compatibility intrinsic, when
`_ARCH_PWR9`.
Also, the realization of `vec_sum2s` on little-endian includes
two rotates in order to position the input and output to match
the semantics of `vec_sum2s`:
- Rotate the second input vector left 12 bytes. In the current usage,
that vector is `{0}`, so this shift is unnecessary, but is currently
not eliminated under optimization.
- Rotate the vector produced by the `vsum2sws` instruction left 4 bytes.
The two words within each doubleword of this (rotated) result must then
be explicitly swapped to match the semantics of `_mm_sad_epu8`,
effectively reversing this rotate. So, this rotate (and a susequent
swap) are unnecessary, but not currently removed under optimization.
Using `__builtin_altivec_vsum2sws` retains both rotates, so is not an
option for removing the rotates.
For little-endian, use the `vsum2sws` instruction directly, and
eliminate the explicit rotate (swap).
2021-11-19 Paul A. Clarke <pc@us.ibm.com>
gcc
* config/rs6000/emmintrin.h (_mm_sad_epu8): Use vec_absd when
_ARCH_PWR9, optimize vec_sum2s when LE.
Diff:
---
gcc/config/rs6000/emmintrin.h | 24 +++++++++++++++++-------
1 file changed, 17 insertions(+), 7 deletions(-)
diff --git a/gcc/config/rs6000/emmintrin.h b/gcc/config/rs6000/emmintrin.h
index 32ad72b4cc3..4125b12fa09 100644
--- a/gcc/config/rs6000/emmintrin.h
+++ b/gcc/config/rs6000/emmintrin.h
@@ -2189,27 +2189,37 @@ extern __inline __m128i __attribute__((__gnu_inline__, __always_inline__, __arti
_mm_sad_epu8 (__m128i __A, __m128i __B)
{
__v16qu a, b;
- __v16qu vmin, vmax, vabsdiff;
+ __v16qu vabsdiff;
__v4si vsum;
const __v4su zero = { 0, 0, 0, 0 };
__v4si result;
a = (__v16qu) __A;
b = (__v16qu) __B;
- vmin = vec_min (a, b);
- vmax = vec_max (a, b);
+#ifndef _ARCH_PWR9
+ __v16qu vmin = vec_min (a, b);
+ __v16qu vmax = vec_max (a, b);
vabsdiff = vec_sub (vmax, vmin);
+#else
+ vabsdiff = vec_absd (a, b);
+#endif
/* Sum four groups of bytes into integers. */
vsum = (__vector signed int) vec_sum4s (vabsdiff, zero);
+#ifdef __LITTLE_ENDIAN__
+ /* Sum across four integers with two integer results. */
+ asm ("vsum2sws %0,%1,%2" : "=v" (result) : "v" (vsum), "v" (zero));
+ /* Note: vec_sum2s could be used here, but on little-endian, vector
+ shifts are added that are not needed for this use-case.
+ A vector shift to correctly position the 32-bit integer results
+ (currently at [0] and [2]) to [1] and [3] would then need to be
+ swapped back again since the desired results are two 64-bit
+ integers ([1]|[0] and [3]|[2]). Thus, no shift is performed. */
+#else
/* Sum across four integers with two integer results. */
result = vec_sum2s (vsum, (__vector signed int) zero);
/* Rotate the sums into the correct position. */
-#ifdef __LITTLE_ENDIAN__
- result = vec_sld (result, result, 4);
-#else
result = vec_sld (result, result, 6);
#endif
- /* Rotate the sums into the correct position. */
return (__m128i) result;
}
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2021-11-19 20:02 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-11-19 20:02 [gcc r12-5420] rs6000: Add optimizations for _mm_sad_epu8 Paul Clarke
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).