AuroraMiddleware/glibc
1
0
Fork 0
mirror of https://sourceware.org/git/glibc.git synced 2024-11-25 14:30:06 +00:00
Code Issues Packages Projects Releases Wiki Activity
e41d66e41a
glibc/benchtests/bench-memcpy-walk.c

119 lines
3.7 KiB
C
Raw Normal View History

benchtests: Memory walking benchmark for memcpy This benchmark is an attempt to eliminate cache effects from string benchmarks. The benchmark walks both ways through a large memory area and copies different sizes of memory and alignments one at a time instead of looping around in the same memory area. This is a good metric to have alongside the other memcpy benchmarks, especially for larger sizes where the likelihood of the call being done only once is pretty high. * benchtests/bench-memcpy-walk.c: New file. * benchtests/Makefile (string-benchset): Add it.
2017-10-05 16:19:37 +00:00
/* Measure memcpy function combined throughput for different alignments.
Update copyright dates with scripts/update-copyrights. * All files with FSF copyright notices: Update copyright dates using scripts/update-copyrights. * locale/programs/charmap-kw.h: Regenerated. * locale/programs/locfile-kw.h: Likewise.
2019-01-01 00:11:28 +00:00
Copyright (C) 2017-2019 Free Software Foundation, Inc.
benchtests: Memory walking benchmark for memcpy This benchmark is an attempt to eliminate cache effects from string benchmarks. The benchmark walks both ways through a large memory area and copies different sizes of memory and alignments one at a time instead of looping around in the same memory area. This is a good metric to have alongside the other memcpy benchmarks, especially for larger sizes where the likelihood of the call being done only once is pretty high. * benchtests/bench-memcpy-walk.c: New file. * benchtests/Makefile (string-benchset): Add it.
2017-10-05 16:19:37 +00:00
This file is part of the GNU C Library.
The GNU C Library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
The GNU C Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
/* This microbenchmark measures the throughput of memcpy for various sizes from
1 byte to 32MiB, doubling every iteration and then misaligning by 0-15
bytes. The copies are done from source to destination and then back and the
source walks forward across the array and the destination walks backward by
one byte each, thus measuring misaligned accesses as well. The idea is to
avoid caching effects by copying a different string and far enough from each
other, walking in different directions so that we can measure prefetcher
efficiency (software or hardware) more closely than with a loop copying the
same data over and over, which eventually only gives us L1 cache
performance. */
#ifndef MEMCPY_RESULT
# define MEMCPY_RESULT(dst, len) dst
benchtests: Bump start size since smaller sizes are noisy Numbers for very small sizes (< 128B) are much noisier for non-cached benchmarks like the walk benchmarks, so don't include them. * benchtests/bench-memcpy-walk.c (START_SIZE): Set to 128. * benchtests/bench-memmove-walk.c (START_SIZE): Likewise. * benchtests/bench-memset-walk.c (START_SIZE): Likewise.
2017-11-20 12:26:35 +00:00
# define START_SIZE 128
benchtests: Memory walking benchmark for memcpy This benchmark is an attempt to eliminate cache effects from string benchmarks. The benchmark walks both ways through a large memory area and copies different sizes of memory and alignments one at a time instead of looping around in the same memory area. This is a good metric to have alongside the other memcpy benchmarks, especially for larger sizes where the likelihood of the call being done only once is pretty high. * benchtests/bench-memcpy-walk.c: New file. * benchtests/Makefile (string-benchset): Add it.
2017-10-05 16:19:37 +00:00
# define MIN_PAGE_SIZE (getpagesize () + 32 * 1024 * 1024)
# define TEST_MAIN
# define TEST_NAME "memcpy"
# define TIMEOUT (20 * 60)
# include "bench-string.h"
IMPL (memcpy, 1)
#endif
#include "json-lib.h"
typedef char *(*proto_t) (char *, const char *, size_t);
static void
do_one_test (json_ctx_t *json_ctx, impl_t *impl, char *dst, char *src,
size_t len)
{
benchtests: Fix walking sizes and directions for *-walk benchmarks Make the walking benchmarks walk only backwards since copying both ways is biased in favour of implementations that use non-temporal stores for larger sizes; falkor is one of them. This also fixes up bugs in computation of the result which ended up multiplying the length with the timing result unnecessarily. * benchtests/bench-memcpy-walk.c (do_one_test): Copy only backwards. Fix timing computation. * benchtests/bench-memmove-walk.c (do_one_test): Likewise. * benchtests/bench-memset-walk.c (do_one_test): Walk backwards on memset by N at a time. Fix timing computation.
2017-11-20 12:25:59 +00:00
size_t i = 0;
benchtests: Memory walking benchmark for memcpy This benchmark is an attempt to eliminate cache effects from string benchmarks. The benchmark walks both ways through a large memory area and copies different sizes of memory and alignments one at a time instead of looping around in the same memory area. This is a good metric to have alongside the other memcpy benchmarks, especially for larger sizes where the likelihood of the call being done only once is pretty high. * benchtests/bench-memcpy-walk.c: New file. * benchtests/Makefile (string-benchset): Add it.
2017-10-05 16:19:37 +00:00
timing_t start, stop, cur;
char *dst_end = dst + MIN_PAGE_SIZE - len;
char *src_end = src + MIN_PAGE_SIZE - len;
TIMING_NOW (start);
benchtests: Fix walking sizes and directions for *-walk benchmarks Make the walking benchmarks walk only backwards since copying both ways is biased in favour of implementations that use non-temporal stores for larger sizes; falkor is one of them. This also fixes up bugs in computation of the result which ended up multiplying the length with the timing result unnecessarily. * benchtests/bench-memcpy-walk.c (do_one_test): Copy only backwards. Fix timing computation. * benchtests/bench-memmove-walk.c (do_one_test): Likewise. * benchtests/bench-memset-walk.c (do_one_test): Walk backwards on memset by N at a time. Fix timing computation.
2017-11-20 12:25:59 +00:00
/* Copy the entire buffer backwards, LEN at a time. */
for (; src_end >= src && dst_end >= dst; src_end -= len, dst_end -= len, i++)
CALL (impl, src_end, dst_end, len);
benchtests: Memory walking benchmark for memcpy This benchmark is an attempt to eliminate cache effects from string benchmarks. The benchmark walks both ways through a large memory area and copies different sizes of memory and alignments one at a time instead of looping around in the same memory area. This is a good metric to have alongside the other memcpy benchmarks, especially for larger sizes where the likelihood of the call being done only once is pretty high. * benchtests/bench-memcpy-walk.c: New file. * benchtests/Makefile (string-benchset): Add it.
2017-10-05 16:19:37 +00:00
TIMING_NOW (stop);
TIMING_DIFF (cur, start, stop);
/* Get time taken per function call. */
benchtests: Fix walking sizes and directions for *-walk benchmarks Make the walking benchmarks walk only backwards since copying both ways is biased in favour of implementations that use non-temporal stores for larger sizes; falkor is one of them. This also fixes up bugs in computation of the result which ended up multiplying the length with the timing result unnecessarily. * benchtests/bench-memcpy-walk.c (do_one_test): Copy only backwards. Fix timing computation. * benchtests/bench-memmove-walk.c (do_one_test): Likewise. * benchtests/bench-memset-walk.c (do_one_test): Walk backwards on memset by N at a time. Fix timing computation.
2017-11-20 12:25:59 +00:00
json_element_double (json_ctx, (double) cur / i);
benchtests: Memory walking benchmark for memcpy This benchmark is an attempt to eliminate cache effects from string benchmarks. The benchmark walks both ways through a large memory area and copies different sizes of memory and alignments one at a time instead of looping around in the same memory area. This is a good metric to have alongside the other memcpy benchmarks, especially for larger sizes where the likelihood of the call being done only once is pretty high. * benchtests/bench-memcpy-walk.c: New file. * benchtests/Makefile (string-benchset): Add it.
2017-10-05 16:19:37 +00:00
}
static void
do_test (json_ctx_t *json_ctx, size_t len)
{
json_element_object_begin (json_ctx);
json_attr_uint (json_ctx, "length", (double) len);
json_array_begin (json_ctx, "timings");
FOR_EACH_IMPL (impl, 0)
do_one_test (json_ctx, impl, (char *) buf2, (char *) buf1, len);
json_array_end (json_ctx);
json_element_object_end (json_ctx);
}
int
test_main (void)
{
json_ctx_t json_ctx;
test_init ();
json_init (&json_ctx, 0, stdout);
json_document_begin (&json_ctx);
json_attr_string (&json_ctx, "timing_type", TIMING_TYPE);
json_attr_object_begin (&json_ctx, "functions");
json_attr_object_begin (&json_ctx, "memcpy");
json_attr_string (&json_ctx, "bench-variant", "walk");
json_array_begin (&json_ctx, "ifuncs");
FOR_EACH_IMPL (impl, 0)
json_element_string (&json_ctx, impl->name);
json_array_end (&json_ctx);
json_array_begin (&json_ctx, "results");
benchtests: Move iterator declaration into loop header This is a minor style change to move the definition of I to its usage scope instead of at the top of the function. This is consistent with glibc style guidelines and more importantly it was getting in the way of my testing. * benchtests/bench-memcpy-walk.c (do_test): Move declaration of I into loop header. * benchtests/bench-memmove-walk.c (do_test): Likewise.
2018-05-07 15:24:31 +00:00
for (size_t i = START_SIZE; i <= MIN_PAGE_SIZE; i <<= 1)
benchtests: Memory walking benchmark for memcpy This benchmark is an attempt to eliminate cache effects from string benchmarks. The benchmark walks both ways through a large memory area and copies different sizes of memory and alignments one at a time instead of looping around in the same memory area. This is a good metric to have alongside the other memcpy benchmarks, especially for larger sizes where the likelihood of the call being done only once is pretty high. * benchtests/bench-memcpy-walk.c: New file. * benchtests/Makefile (string-benchset): Add it.
2017-10-05 16:19:37 +00:00
{
Improve string benchtest timing Improve string benchtest timing. Many tests run for 0.01s which is way too short to give accurate results. Other tests take over 40 seconds which is way too long. Significantly increase the iterations of the short running tests. Reduce number of alignment variations in the long running memcpy walk tests so they take less than 5 seconds. As a result most tests take at least 0.1s and all finish within 5 seconds. * benchtests/bench-memcpy-random.c (do_one_test): Use medium iterations. * benchtests/bench-memcpy-walk.c (test_main): Reduce alignment tests. * benchtests/bench-memmem.c (do_one_test): Use small iterations. * benchtests/bench-memmove-walk.c (test_main): Reduce alignment tests. * benchtests/bench-memset-walk.c (test_main): Reduce alignment tests. * benchtests/bench-strcasestr.c (do_one_test): Use small iterations. * benchtests/bench-string.h (INNER_LOOP_ITERS): Increase iterations. (INNER_LOOP_ITERS_MEDIUM): New define. (INNER_LOOP_ITERS_SMALL): New define. * benchtests/bench-strpbrk.c (do_one_test): Use medium iterations. * benchtests/bench-strsep.c (do_one_test): Use small iterations. * benchtests/bench-strspn.c (do_one_test): Use medium iterations. * benchtests/bench-strstr.c (do_one_test): Use small iterations. * benchtests/bench-strtok.c (do_one_test): Use small iterations.
2019-05-21 14:19:06 +00:00
do_test (&json_ctx, i);
do_test (&json_ctx, i + 1);
benchtests: Memory walking benchmark for memcpy This benchmark is an attempt to eliminate cache effects from string benchmarks. The benchmark walks both ways through a large memory area and copies different sizes of memory and alignments one at a time instead of looping around in the same memory area. This is a good metric to have alongside the other memcpy benchmarks, especially for larger sizes where the likelihood of the call being done only once is pretty high. * benchtests/bench-memcpy-walk.c: New file. * benchtests/Makefile (string-benchset): Add it.
2017-10-05 16:19:37 +00:00
}
json_array_end (&json_ctx);
json_attr_object_end (&json_ctx);
json_attr_object_end (&json_ctx);
json_document_end (&json_ctx);
return ret;
}
#include <support/test-driver.c>
Reference in New Issue Copy Permalink