On Fri, Mar 15, 2024 at 12:41:49PM -0500, Nathan Bossart wrote:
> I've also attached the results of running this benchmark on my machine at
> HEAD, after applying 0001, and after applying both 0001 and 0002. 0001
> appears to work pretty well. When there is a small "tail," it regresses a
> small amount, but overall, it seems to improve more cases than it harms.
> 0002 does regress searches on smaller arrays quite a bit, since it
> postpones the SIMD optimizations until the arrays are longer. It might be
> possible to mitigate by using 2 registers when the "tail" is long enough,
> but I have yet to try that.
The attached 0003 is a sketch of what such mitigation might look like. It
appears to help with the regressions nicely. I omitted the benchmarking
patch in v3 to appease cfbot.
--
Nathan Bossart
Amazon Web Services: https://aws.amazon.com
>From 3817435d200af5da954d505ae66245662dea064c Mon Sep 17 00:00:00 2001
From: Nathan Bossart <nat...@postgresql.org>
Date: Fri, 15 Mar 2024 12:26:26 -0500
Subject: [PATCH v3 1/3] pg_lfind32: process "tail" with SIMD intructions
---
src/include/port/pg_lfind.h | 16 ++++++++++++++--
1 file changed, 14 insertions(+), 2 deletions(-)
diff --git a/src/include/port/pg_lfind.h b/src/include/port/pg_lfind.h
index b8dfa66eef..9d21284724 100644
--- a/src/include/port/pg_lfind.h
+++ b/src/include/port/pg_lfind.h
@@ -103,7 +103,7 @@ pg_lfind32(uint32 key, uint32 *base, uint32 nelem)
const uint32 nelem_per_iteration = 4 * nelem_per_vector;
/* round down to multiple of elements per iteration */
- const uint32 tail_idx = nelem & ~(nelem_per_iteration - 1);
+ uint32 tail_idx = nelem & ~(nelem_per_iteration - 1);
#if defined(USE_ASSERT_CHECKING)
bool assert_result = false;
@@ -117,9 +117,11 @@ pg_lfind32(uint32 key, uint32 *base, uint32 nelem)
break;
}
}
+ i = 0;
#endif
- for (i = 0; i < tail_idx; i += nelem_per_iteration)
+retry:
+ for (; i < tail_idx; i += nelem_per_iteration)
{
Vector32 vals1,
vals2,
@@ -157,6 +159,16 @@ pg_lfind32(uint32 key, uint32 *base, uint32 nelem)
return true;
}
}
+
+ if (i == nelem)
+ return false;
+ else if (tail_idx > 0)
+ {
+ tail_idx = nelem;
+ i = nelem - nelem_per_iteration;
+ goto retry;
+ }
+
#endif /* ! USE_NO_SIMD */
/* Process the remaining elements one at a time. */
--
2.25.1
>From a867e342db08aae501374c75c0d8f17473a6cbc9 Mon Sep 17 00:00:00 2001
From: Nathan Bossart <nat...@postgresql.org>
Date: Fri, 15 Mar 2024 12:26:52 -0500
Subject: [PATCH v3 2/3] add avx2 support in simd.h
---
src/include/port/simd.h | 58 ++++++++++++++++++++++++++++++++---------
1 file changed, 45 insertions(+), 13 deletions(-)
diff --git a/src/include/port/simd.h b/src/include/port/simd.h
index 597496f2fb..767127b85c 100644
--- a/src/include/port/simd.h
+++ b/src/include/port/simd.h
@@ -18,7 +18,15 @@
#ifndef SIMD_H
#define SIMD_H
-#if (defined(__x86_64__) || defined(_M_AMD64))
+#if defined(__AVX2__)
+
+#include <immintrin.h>
+#define USE_AVX2
+typedef __m256i Vector8;
+typedef __m256i Vector32;
+
+#elif (defined(__x86_64__) || defined(_M_AMD64))
+
/*
* SSE2 instructions are part of the spec for the 64-bit x86 ISA. We assume
* that compilers targeting this architecture understand SSE2 intrinsics.
@@ -107,7 +115,9 @@ static inline Vector32 vector32_eq(const Vector32 v1, const Vector32 v2);
static inline void
vector8_load(Vector8 *v, const uint8 *s)
{
-#if defined(USE_SSE2)
+#if defined(USE_AVX2)
+ *v = _mm256_loadu_si256((const __m256i *) s);
+#elif defined(USE_SSE2)
*v = _mm_loadu_si128((const __m128i *) s);
#elif defined(USE_NEON)
*v = vld1q_u8(s);
@@ -120,7 +130,9 @@ vector8_load(Vector8 *v, const uint8 *s)
static inline void
vector32_load(Vector32 *v, const uint32 *s)
{
-#ifdef USE_SSE2
+#if defined(USE_AVX2)
+ *v = _mm256_loadu_si256((const __m256i *) s);
+#elif defined(USE_SSE2)
*v = _mm_loadu_si128((const __m128i *) s);
#elif defined(USE_NEON)
*v = vld1q_u32(s);
@@ -134,7 +146,9 @@ vector32_load(Vector32 *v, const uint32 *s)
static inline Vector8
vector8_broadcast(const uint8 c)
{
-#if defined(USE_SSE2)
+#if defined(USE_AVX2)
+ return _mm256_set1_epi8(c);
+#elif defined(USE_SSE2)
return _mm_set1_epi8(c);
#elif defined(USE_NEON)
return vdupq_n_u8(c);
@@ -147,7 +161,9 @@ vector8_broadcast(const uint8 c)
static inline Vector32
vector32_broadcast(const uint32 c)
{
-#ifdef USE_SSE2
+#if defined(USE_AVX2)
+ return _mm256_set1_epi32(c);
+#elif defined(USE_SSE2)
return _mm_set1_epi32(c);
#elif defined(USE_NEON)
return vdupq_n_u32(c);
@@ -270,7 +286,9 @@ vector8_has_le(const Vector8 v, const uint8 c)
static inline bool
vector8_is_highbit_set(const Vector8 v)
{
-#ifdef USE_SSE2
+#if defined(USE_AVX2)
+ return _mm256_movemask_epi8(v) != 0;
+#elif defined(USE_SSE2)
return _mm_movemask_epi8(v) != 0;
#elif defined(USE_NEON)
return vmaxvq_u8(v) > 0x7F;
@@ -308,7 +326,9 @@ vector32_is_highbit_set(const Vector32 v)
static inline uint32
vector8_highbit_mask(const Vector8 v)
{
-#ifdef USE_SSE2
+#if defined(USE_AVX2)
+ return (uint32) _mm256_movemask_epi8(v);
+#elif defined(USE_SSE2)
return (uint32) _mm_movemask_epi8(v);
#elif defined(USE_NEON)
/*
@@ -337,7 +357,9 @@ vector8_highbit_mask(const Vector8 v)
static inline Vector8
vector8_or(const Vector8 v1, const Vector8 v2)
{
-#ifdef USE_SSE2
+#if defined(USE_AVX2)
+ return _mm256_or_si256(v1, v2);
+#elif defined(USE_SSE2)
return _mm_or_si128(v1, v2);
#elif defined(USE_NEON)
return vorrq_u8(v1, v2);
@@ -350,7 +372,9 @@ vector8_or(const Vector8 v1, const Vector8 v2)
static inline Vector32
vector32_or(const Vector32 v1, const Vector32 v2)
{
-#ifdef USE_SSE2
+#if defined(USE_AVX2)
+ return _mm256_or_si256(v1, v2);
+#elif defined(USE_SSE2)
return _mm_or_si128(v1, v2);
#elif defined(USE_NEON)
return vorrq_u32(v1, v2);
@@ -368,7 +392,9 @@ vector32_or(const Vector32 v1, const Vector32 v2)
static inline Vector8
vector8_ssub(const Vector8 v1, const Vector8 v2)
{
-#ifdef USE_SSE2
+#if defined(USE_AVX2)
+ return _mm256_subs_epu8(v1, v2);
+#elif defined(USE_SSE2)
return _mm_subs_epu8(v1, v2);
#elif defined(USE_NEON)
return vqsubq_u8(v1, v2);
@@ -384,7 +410,9 @@ vector8_ssub(const Vector8 v1, const Vector8 v2)
static inline Vector8
vector8_eq(const Vector8 v1, const Vector8 v2)
{
-#ifdef USE_SSE2
+#if defined(USE_AVX2)
+ return _mm256_cmpeq_epi8(v1, v2);
+#elif defined(USE_SSE2)
return _mm_cmpeq_epi8(v1, v2);
#elif defined(USE_NEON)
return vceqq_u8(v1, v2);
@@ -396,7 +424,9 @@ vector8_eq(const Vector8 v1, const Vector8 v2)
static inline Vector32
vector32_eq(const Vector32 v1, const Vector32 v2)
{
-#ifdef USE_SSE2
+#if defined(USE_AVX2)
+ return _mm256_cmpeq_epi32(v1, v2);
+#elif defined(USE_SSE2)
return _mm_cmpeq_epi32(v1, v2);
#elif defined(USE_NEON)
return vceqq_u32(v1, v2);
@@ -411,7 +441,9 @@ vector32_eq(const Vector32 v1, const Vector32 v2)
static inline Vector8
vector8_min(const Vector8 v1, const Vector8 v2)
{
-#ifdef USE_SSE2
+#if defined(USE_AVX2)
+ return _mm256_min_epu8(v1, v2);
+#elif defined(USE_SSE2)
return _mm_min_epu8(v1, v2);
#elif defined(USE_NEON)
return vminq_u8(v1, v2);
--
2.25.1
>From edd188759e4b937089c5bc2259a401cea1f9331f Mon Sep 17 00:00:00 2001
From: Nathan Bossart <nat...@postgresql.org>
Date: Fri, 15 Mar 2024 14:27:52 -0500
Subject: [PATCH v3 3/3] optimize pg_lfind32() by processing "tail" with fewer
vectors
---
src/include/port/pg_lfind.h | 29 +++++++++++++++++++++++++++--
1 file changed, 27 insertions(+), 2 deletions(-)
diff --git a/src/include/port/pg_lfind.h b/src/include/port/pg_lfind.h
index 9d21284724..9c6cce0b69 100644
--- a/src/include/port/pg_lfind.h
+++ b/src/include/port/pg_lfind.h
@@ -100,7 +100,7 @@ pg_lfind32(uint32 key, uint32 *base, uint32 nelem)
*/
const Vector32 keys = vector32_broadcast(key); /* load copies of key */
const uint32 nelem_per_vector = sizeof(Vector32) / sizeof(uint32);
- const uint32 nelem_per_iteration = 4 * nelem_per_vector;
+ uint32 nelem_per_iteration = 4 * nelem_per_vector;
/* round down to multiple of elements per iteration */
uint32 tail_idx = nelem & ~(nelem_per_iteration - 1);
@@ -120,7 +120,6 @@ pg_lfind32(uint32 key, uint32 *base, uint32 nelem)
i = 0;
#endif
-retry:
for (; i < tail_idx; i += nelem_per_iteration)
{
Vector32 vals1,
@@ -160,6 +159,32 @@ retry:
}
}
+retry:
+ nelem_per_iteration = 2 * nelem_per_vector;
+ tail_idx = nelem & ~(nelem_per_iteration - 1);
+ for (; i < tail_idx; i += nelem_per_iteration)
+ {
+ Vector32 vals1,
+ vals2,
+ result1,
+ result2,
+ result;
+
+ vector32_load(&vals1, &base[i]);
+ vector32_load(&vals2, &base[i + nelem_per_vector]);
+
+ result1 = vector32_eq(keys, vals1);
+ result2 = vector32_eq(keys, vals2);
+
+ result = vector32_or(result1, result2);
+
+ if (vector32_is_highbit_set(result))
+ {
+ Assert(assert_result == true);
+ return true;
+ }
+ }
+
if (i == nelem)
return false;
else if (tail_idx > 0)
--
2.25.1
