OpenCloudOS-Kernel/lib/find_bit_benchmark.c

198 lines
4.9 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/*
* Test for find_*_bit functions.
*
* Copyright (c) 2017 Cavium.
*/
/*
* find_bit functions are widely used in kernel, so the successful boot
* is good enough test for correctness.
*
* This test is focused on performance of traversing bitmaps. Two typical
* scenarios are reproduced:
* - randomly filled bitmap with approximately equal number of set and
* cleared bits;
* - sparse bitmap with few set bits at random positions.
*/
#include <linux/bitops.h>
#include <linux/kernel.h>
#include <linux/list.h>
#include <linux/module.h>
#include <linux/printk.h>
#include <linux/random.h>
#define BITMAP_LEN (4096UL * 8 * 10)
#define SPARSE 500
static DECLARE_BITMAP(bitmap, BITMAP_LEN) __initdata;
lib: optimize cpumask_next_and() We've measured that we spend ~0.6% of sys cpu time in cpumask_next_and(). It's essentially a joined iteration in search for a non-zero bit, which is currently implemented as a lookup join (find a nonzero bit on the lhs, lookup the rhs to see if it's set there). Implement a direct join (find a nonzero bit on the incrementally built join). Also add generic bitmap benchmarks in the new `test_find_bit` module for new function (see `find_next_and_bit` in [2] and [3] below). For cpumask_next_and, direct benchmarking shows that it's 1.17x to 14x faster with a geometric mean of 2.1 on 32 CPUs [1]. No impact on memory usage. Note that on Arm, the new pure-C implementation still outperforms the old one that uses a mix of C and asm (`find_next_bit`) [3]. [1] Approximate benchmark code: ``` unsigned long src1p[nr_cpumask_longs] = {pattern1}; unsigned long src2p[nr_cpumask_longs] = {pattern2}; for (/*a bunch of repetitions*/) { for (int n = -1; n <= nr_cpu_ids; ++n) { asm volatile("" : "+rm"(src1p)); // prevent any optimization asm volatile("" : "+rm"(src2p)); unsigned long result = cpumask_next_and(n, src1p, src2p); asm volatile("" : "+rm"(result)); } } ``` Results: pattern1 pattern2 time_before/time_after 0x0000ffff 0x0000ffff 1.65 0x0000ffff 0x00005555 2.24 0x0000ffff 0x00001111 2.94 0x0000ffff 0x00000000 14.0 0x00005555 0x0000ffff 1.67 0x00005555 0x00005555 1.71 0x00005555 0x00001111 1.90 0x00005555 0x00000000 6.58 0x00001111 0x0000ffff 1.46 0x00001111 0x00005555 1.49 0x00001111 0x00001111 1.45 0x00001111 0x00000000 3.10 0x00000000 0x0000ffff 1.18 0x00000000 0x00005555 1.18 0x00000000 0x00001111 1.17 0x00000000 0x00000000 1.25 ----------------------------- geo.mean 2.06 [2] test_find_next_bit, X86 (skylake) [ 3913.477422] Start testing find_bit() with random-filled bitmap [ 3913.477847] find_next_bit: 160868 cycles, 16484 iterations [ 3913.477933] find_next_zero_bit: 169542 cycles, 16285 iterations [ 3913.478036] find_last_bit: 201638 cycles, 16483 iterations [ 3913.480214] find_first_bit: 4353244 cycles, 16484 iterations [ 3913.480216] Start testing find_next_and_bit() with random-filled bitmap [ 3913.481074] find_next_and_bit: 89604 cycles, 8216 iterations [ 3913.481075] Start testing find_bit() with sparse bitmap [ 3913.481078] find_next_bit: 2536 cycles, 66 iterations [ 3913.481252] find_next_zero_bit: 344404 cycles, 32703 iterations [ 3913.481255] find_last_bit: 2006 cycles, 66 iterations [ 3913.481265] find_first_bit: 17488 cycles, 66 iterations [ 3913.481266] Start testing find_next_and_bit() with sparse bitmap [ 3913.481272] find_next_and_bit: 764 cycles, 1 iterations [3] test_find_next_bit, arm (v7 odroid XU3). [ 267.206928] Start testing find_bit() with random-filled bitmap [ 267.214752] find_next_bit: 4474 cycles, 16419 iterations [ 267.221850] find_next_zero_bit: 5976 cycles, 16350 iterations [ 267.229294] find_last_bit: 4209 cycles, 16419 iterations [ 267.279131] find_first_bit: 1032991 cycles, 16420 iterations [ 267.286265] Start testing find_next_and_bit() with random-filled bitmap [ 267.302386] find_next_and_bit: 2290 cycles, 8140 iterations [ 267.309422] Start testing find_bit() with sparse bitmap [ 267.316054] find_next_bit: 191 cycles, 66 iterations [ 267.322726] find_next_zero_bit: 8758 cycles, 32703 iterations [ 267.329803] find_last_bit: 84 cycles, 66 iterations [ 267.336169] find_first_bit: 4118 cycles, 66 iterations [ 267.342627] Start testing find_next_and_bit() with sparse bitmap [ 267.356919] find_next_and_bit: 91 cycles, 1 iterations [courbet@google.com: v6] Link: http://lkml.kernel.org/r/20171129095715.23430-1-courbet@google.com [geert@linux-m68k.org: m68k/bitops: always include <asm-generic/bitops/find.h>] Link: http://lkml.kernel.org/r/1512556816-28627-1-git-send-email-geert@linux-m68k.org Link: http://lkml.kernel.org/r/20171128131334.23491-1-courbet@google.com Signed-off-by: Clement Courbet <courbet@google.com> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Yury Norov <ynorov@caviumnetworks.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-02-07 07:38:34 +08:00
static DECLARE_BITMAP(bitmap2, BITMAP_LEN) __initdata;
/*
* This is Schlemiel the Painter's algorithm. It should be called after
* all other tests for the same bitmap because it sets all bits of bitmap to 1.
*/
static int __init test_find_first_bit(void *bitmap, unsigned long len)
{
unsigned long i, cnt;
ktime_t time;
time = ktime_get();
for (cnt = i = 0; i < len; cnt++) {
i = find_first_bit(bitmap, len);
__clear_bit(i, bitmap);
}
time = ktime_get() - time;
pr_err("find_first_bit: %18llu ns, %6ld iterations\n", time, cnt);
return 0;
}
static int __init test_find_first_and_bit(void *bitmap, const void *bitmap2, unsigned long len)
{
static DECLARE_BITMAP(cp, BITMAP_LEN) __initdata;
unsigned long i, cnt;
ktime_t time;
bitmap_copy(cp, bitmap, BITMAP_LEN);
time = ktime_get();
for (cnt = i = 0; i < len; cnt++) {
i = find_first_and_bit(cp, bitmap2, len);
__clear_bit(i, cp);
}
time = ktime_get() - time;
pr_err("find_first_and_bit: %18llu ns, %6ld iterations\n", time, cnt);
return 0;
}
static int __init test_find_next_bit(const void *bitmap, unsigned long len)
{
unsigned long i, cnt;
ktime_t time;
time = ktime_get();
for (cnt = i = 0; i < BITMAP_LEN; cnt++)
i = find_next_bit(bitmap, BITMAP_LEN, i) + 1;
time = ktime_get() - time;
pr_err("find_next_bit: %18llu ns, %6ld iterations\n", time, cnt);
return 0;
}
static int __init test_find_next_zero_bit(const void *bitmap, unsigned long len)
{
unsigned long i, cnt;
ktime_t time;
time = ktime_get();
for (cnt = i = 0; i < BITMAP_LEN; cnt++)
i = find_next_zero_bit(bitmap, len, i) + 1;
time = ktime_get() - time;
pr_err("find_next_zero_bit: %18llu ns, %6ld iterations\n", time, cnt);
return 0;
}
static int __init test_find_last_bit(const void *bitmap, unsigned long len)
{
unsigned long l, cnt = 0;
ktime_t time;
time = ktime_get();
do {
cnt++;
l = find_last_bit(bitmap, len);
if (l >= len)
break;
len = l;
} while (len);
time = ktime_get() - time;
pr_err("find_last_bit: %18llu ns, %6ld iterations\n", time, cnt);
return 0;
}
static int __init test_find_nth_bit(const unsigned long *bitmap, unsigned long len)
{
unsigned long l, n, w = bitmap_weight(bitmap, len);
ktime_t time;
time = ktime_get();
for (n = 0; n < w; n++) {
l = find_nth_bit(bitmap, len, n);
WARN_ON(l >= len);
}
time = ktime_get() - time;
pr_err("find_nth_bit: %18llu ns, %6ld iterations\n", time, w);
return 0;
}
lib: optimize cpumask_next_and() We've measured that we spend ~0.6% of sys cpu time in cpumask_next_and(). It's essentially a joined iteration in search for a non-zero bit, which is currently implemented as a lookup join (find a nonzero bit on the lhs, lookup the rhs to see if it's set there). Implement a direct join (find a nonzero bit on the incrementally built join). Also add generic bitmap benchmarks in the new `test_find_bit` module for new function (see `find_next_and_bit` in [2] and [3] below). For cpumask_next_and, direct benchmarking shows that it's 1.17x to 14x faster with a geometric mean of 2.1 on 32 CPUs [1]. No impact on memory usage. Note that on Arm, the new pure-C implementation still outperforms the old one that uses a mix of C and asm (`find_next_bit`) [3]. [1] Approximate benchmark code: ``` unsigned long src1p[nr_cpumask_longs] = {pattern1}; unsigned long src2p[nr_cpumask_longs] = {pattern2}; for (/*a bunch of repetitions*/) { for (int n = -1; n <= nr_cpu_ids; ++n) { asm volatile("" : "+rm"(src1p)); // prevent any optimization asm volatile("" : "+rm"(src2p)); unsigned long result = cpumask_next_and(n, src1p, src2p); asm volatile("" : "+rm"(result)); } } ``` Results: pattern1 pattern2 time_before/time_after 0x0000ffff 0x0000ffff 1.65 0x0000ffff 0x00005555 2.24 0x0000ffff 0x00001111 2.94 0x0000ffff 0x00000000 14.0 0x00005555 0x0000ffff 1.67 0x00005555 0x00005555 1.71 0x00005555 0x00001111 1.90 0x00005555 0x00000000 6.58 0x00001111 0x0000ffff 1.46 0x00001111 0x00005555 1.49 0x00001111 0x00001111 1.45 0x00001111 0x00000000 3.10 0x00000000 0x0000ffff 1.18 0x00000000 0x00005555 1.18 0x00000000 0x00001111 1.17 0x00000000 0x00000000 1.25 ----------------------------- geo.mean 2.06 [2] test_find_next_bit, X86 (skylake) [ 3913.477422] Start testing find_bit() with random-filled bitmap [ 3913.477847] find_next_bit: 160868 cycles, 16484 iterations [ 3913.477933] find_next_zero_bit: 169542 cycles, 16285 iterations [ 3913.478036] find_last_bit: 201638 cycles, 16483 iterations [ 3913.480214] find_first_bit: 4353244 cycles, 16484 iterations [ 3913.480216] Start testing find_next_and_bit() with random-filled bitmap [ 3913.481074] find_next_and_bit: 89604 cycles, 8216 iterations [ 3913.481075] Start testing find_bit() with sparse bitmap [ 3913.481078] find_next_bit: 2536 cycles, 66 iterations [ 3913.481252] find_next_zero_bit: 344404 cycles, 32703 iterations [ 3913.481255] find_last_bit: 2006 cycles, 66 iterations [ 3913.481265] find_first_bit: 17488 cycles, 66 iterations [ 3913.481266] Start testing find_next_and_bit() with sparse bitmap [ 3913.481272] find_next_and_bit: 764 cycles, 1 iterations [3] test_find_next_bit, arm (v7 odroid XU3). [ 267.206928] Start testing find_bit() with random-filled bitmap [ 267.214752] find_next_bit: 4474 cycles, 16419 iterations [ 267.221850] find_next_zero_bit: 5976 cycles, 16350 iterations [ 267.229294] find_last_bit: 4209 cycles, 16419 iterations [ 267.279131] find_first_bit: 1032991 cycles, 16420 iterations [ 267.286265] Start testing find_next_and_bit() with random-filled bitmap [ 267.302386] find_next_and_bit: 2290 cycles, 8140 iterations [ 267.309422] Start testing find_bit() with sparse bitmap [ 267.316054] find_next_bit: 191 cycles, 66 iterations [ 267.322726] find_next_zero_bit: 8758 cycles, 32703 iterations [ 267.329803] find_last_bit: 84 cycles, 66 iterations [ 267.336169] find_first_bit: 4118 cycles, 66 iterations [ 267.342627] Start testing find_next_and_bit() with sparse bitmap [ 267.356919] find_next_and_bit: 91 cycles, 1 iterations [courbet@google.com: v6] Link: http://lkml.kernel.org/r/20171129095715.23430-1-courbet@google.com [geert@linux-m68k.org: m68k/bitops: always include <asm-generic/bitops/find.h>] Link: http://lkml.kernel.org/r/1512556816-28627-1-git-send-email-geert@linux-m68k.org Link: http://lkml.kernel.org/r/20171128131334.23491-1-courbet@google.com Signed-off-by: Clement Courbet <courbet@google.com> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Yury Norov <ynorov@caviumnetworks.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-02-07 07:38:34 +08:00
static int __init test_find_next_and_bit(const void *bitmap,
const void *bitmap2, unsigned long len)
{
unsigned long i, cnt;
ktime_t time;
lib: optimize cpumask_next_and() We've measured that we spend ~0.6% of sys cpu time in cpumask_next_and(). It's essentially a joined iteration in search for a non-zero bit, which is currently implemented as a lookup join (find a nonzero bit on the lhs, lookup the rhs to see if it's set there). Implement a direct join (find a nonzero bit on the incrementally built join). Also add generic bitmap benchmarks in the new `test_find_bit` module for new function (see `find_next_and_bit` in [2] and [3] below). For cpumask_next_and, direct benchmarking shows that it's 1.17x to 14x faster with a geometric mean of 2.1 on 32 CPUs [1]. No impact on memory usage. Note that on Arm, the new pure-C implementation still outperforms the old one that uses a mix of C and asm (`find_next_bit`) [3]. [1] Approximate benchmark code: ``` unsigned long src1p[nr_cpumask_longs] = {pattern1}; unsigned long src2p[nr_cpumask_longs] = {pattern2}; for (/*a bunch of repetitions*/) { for (int n = -1; n <= nr_cpu_ids; ++n) { asm volatile("" : "+rm"(src1p)); // prevent any optimization asm volatile("" : "+rm"(src2p)); unsigned long result = cpumask_next_and(n, src1p, src2p); asm volatile("" : "+rm"(result)); } } ``` Results: pattern1 pattern2 time_before/time_after 0x0000ffff 0x0000ffff 1.65 0x0000ffff 0x00005555 2.24 0x0000ffff 0x00001111 2.94 0x0000ffff 0x00000000 14.0 0x00005555 0x0000ffff 1.67 0x00005555 0x00005555 1.71 0x00005555 0x00001111 1.90 0x00005555 0x00000000 6.58 0x00001111 0x0000ffff 1.46 0x00001111 0x00005555 1.49 0x00001111 0x00001111 1.45 0x00001111 0x00000000 3.10 0x00000000 0x0000ffff 1.18 0x00000000 0x00005555 1.18 0x00000000 0x00001111 1.17 0x00000000 0x00000000 1.25 ----------------------------- geo.mean 2.06 [2] test_find_next_bit, X86 (skylake) [ 3913.477422] Start testing find_bit() with random-filled bitmap [ 3913.477847] find_next_bit: 160868 cycles, 16484 iterations [ 3913.477933] find_next_zero_bit: 169542 cycles, 16285 iterations [ 3913.478036] find_last_bit: 201638 cycles, 16483 iterations [ 3913.480214] find_first_bit: 4353244 cycles, 16484 iterations [ 3913.480216] Start testing find_next_and_bit() with random-filled bitmap [ 3913.481074] find_next_and_bit: 89604 cycles, 8216 iterations [ 3913.481075] Start testing find_bit() with sparse bitmap [ 3913.481078] find_next_bit: 2536 cycles, 66 iterations [ 3913.481252] find_next_zero_bit: 344404 cycles, 32703 iterations [ 3913.481255] find_last_bit: 2006 cycles, 66 iterations [ 3913.481265] find_first_bit: 17488 cycles, 66 iterations [ 3913.481266] Start testing find_next_and_bit() with sparse bitmap [ 3913.481272] find_next_and_bit: 764 cycles, 1 iterations [3] test_find_next_bit, arm (v7 odroid XU3). [ 267.206928] Start testing find_bit() with random-filled bitmap [ 267.214752] find_next_bit: 4474 cycles, 16419 iterations [ 267.221850] find_next_zero_bit: 5976 cycles, 16350 iterations [ 267.229294] find_last_bit: 4209 cycles, 16419 iterations [ 267.279131] find_first_bit: 1032991 cycles, 16420 iterations [ 267.286265] Start testing find_next_and_bit() with random-filled bitmap [ 267.302386] find_next_and_bit: 2290 cycles, 8140 iterations [ 267.309422] Start testing find_bit() with sparse bitmap [ 267.316054] find_next_bit: 191 cycles, 66 iterations [ 267.322726] find_next_zero_bit: 8758 cycles, 32703 iterations [ 267.329803] find_last_bit: 84 cycles, 66 iterations [ 267.336169] find_first_bit: 4118 cycles, 66 iterations [ 267.342627] Start testing find_next_and_bit() with sparse bitmap [ 267.356919] find_next_and_bit: 91 cycles, 1 iterations [courbet@google.com: v6] Link: http://lkml.kernel.org/r/20171129095715.23430-1-courbet@google.com [geert@linux-m68k.org: m68k/bitops: always include <asm-generic/bitops/find.h>] Link: http://lkml.kernel.org/r/1512556816-28627-1-git-send-email-geert@linux-m68k.org Link: http://lkml.kernel.org/r/20171128131334.23491-1-courbet@google.com Signed-off-by: Clement Courbet <courbet@google.com> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Yury Norov <ynorov@caviumnetworks.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-02-07 07:38:34 +08:00
time = ktime_get();
lib: optimize cpumask_next_and() We've measured that we spend ~0.6% of sys cpu time in cpumask_next_and(). It's essentially a joined iteration in search for a non-zero bit, which is currently implemented as a lookup join (find a nonzero bit on the lhs, lookup the rhs to see if it's set there). Implement a direct join (find a nonzero bit on the incrementally built join). Also add generic bitmap benchmarks in the new `test_find_bit` module for new function (see `find_next_and_bit` in [2] and [3] below). For cpumask_next_and, direct benchmarking shows that it's 1.17x to 14x faster with a geometric mean of 2.1 on 32 CPUs [1]. No impact on memory usage. Note that on Arm, the new pure-C implementation still outperforms the old one that uses a mix of C and asm (`find_next_bit`) [3]. [1] Approximate benchmark code: ``` unsigned long src1p[nr_cpumask_longs] = {pattern1}; unsigned long src2p[nr_cpumask_longs] = {pattern2}; for (/*a bunch of repetitions*/) { for (int n = -1; n <= nr_cpu_ids; ++n) { asm volatile("" : "+rm"(src1p)); // prevent any optimization asm volatile("" : "+rm"(src2p)); unsigned long result = cpumask_next_and(n, src1p, src2p); asm volatile("" : "+rm"(result)); } } ``` Results: pattern1 pattern2 time_before/time_after 0x0000ffff 0x0000ffff 1.65 0x0000ffff 0x00005555 2.24 0x0000ffff 0x00001111 2.94 0x0000ffff 0x00000000 14.0 0x00005555 0x0000ffff 1.67 0x00005555 0x00005555 1.71 0x00005555 0x00001111 1.90 0x00005555 0x00000000 6.58 0x00001111 0x0000ffff 1.46 0x00001111 0x00005555 1.49 0x00001111 0x00001111 1.45 0x00001111 0x00000000 3.10 0x00000000 0x0000ffff 1.18 0x00000000 0x00005555 1.18 0x00000000 0x00001111 1.17 0x00000000 0x00000000 1.25 ----------------------------- geo.mean 2.06 [2] test_find_next_bit, X86 (skylake) [ 3913.477422] Start testing find_bit() with random-filled bitmap [ 3913.477847] find_next_bit: 160868 cycles, 16484 iterations [ 3913.477933] find_next_zero_bit: 169542 cycles, 16285 iterations [ 3913.478036] find_last_bit: 201638 cycles, 16483 iterations [ 3913.480214] find_first_bit: 4353244 cycles, 16484 iterations [ 3913.480216] Start testing find_next_and_bit() with random-filled bitmap [ 3913.481074] find_next_and_bit: 89604 cycles, 8216 iterations [ 3913.481075] Start testing find_bit() with sparse bitmap [ 3913.481078] find_next_bit: 2536 cycles, 66 iterations [ 3913.481252] find_next_zero_bit: 344404 cycles, 32703 iterations [ 3913.481255] find_last_bit: 2006 cycles, 66 iterations [ 3913.481265] find_first_bit: 17488 cycles, 66 iterations [ 3913.481266] Start testing find_next_and_bit() with sparse bitmap [ 3913.481272] find_next_and_bit: 764 cycles, 1 iterations [3] test_find_next_bit, arm (v7 odroid XU3). [ 267.206928] Start testing find_bit() with random-filled bitmap [ 267.214752] find_next_bit: 4474 cycles, 16419 iterations [ 267.221850] find_next_zero_bit: 5976 cycles, 16350 iterations [ 267.229294] find_last_bit: 4209 cycles, 16419 iterations [ 267.279131] find_first_bit: 1032991 cycles, 16420 iterations [ 267.286265] Start testing find_next_and_bit() with random-filled bitmap [ 267.302386] find_next_and_bit: 2290 cycles, 8140 iterations [ 267.309422] Start testing find_bit() with sparse bitmap [ 267.316054] find_next_bit: 191 cycles, 66 iterations [ 267.322726] find_next_zero_bit: 8758 cycles, 32703 iterations [ 267.329803] find_last_bit: 84 cycles, 66 iterations [ 267.336169] find_first_bit: 4118 cycles, 66 iterations [ 267.342627] Start testing find_next_and_bit() with sparse bitmap [ 267.356919] find_next_and_bit: 91 cycles, 1 iterations [courbet@google.com: v6] Link: http://lkml.kernel.org/r/20171129095715.23430-1-courbet@google.com [geert@linux-m68k.org: m68k/bitops: always include <asm-generic/bitops/find.h>] Link: http://lkml.kernel.org/r/1512556816-28627-1-git-send-email-geert@linux-m68k.org Link: http://lkml.kernel.org/r/20171128131334.23491-1-courbet@google.com Signed-off-by: Clement Courbet <courbet@google.com> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Yury Norov <ynorov@caviumnetworks.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-02-07 07:38:34 +08:00
for (cnt = i = 0; i < BITMAP_LEN; cnt++)
i = find_next_and_bit(bitmap, bitmap2, BITMAP_LEN, i + 1);
time = ktime_get() - time;
pr_err("find_next_and_bit: %18llu ns, %6ld iterations\n", time, cnt);
lib: optimize cpumask_next_and() We've measured that we spend ~0.6% of sys cpu time in cpumask_next_and(). It's essentially a joined iteration in search for a non-zero bit, which is currently implemented as a lookup join (find a nonzero bit on the lhs, lookup the rhs to see if it's set there). Implement a direct join (find a nonzero bit on the incrementally built join). Also add generic bitmap benchmarks in the new `test_find_bit` module for new function (see `find_next_and_bit` in [2] and [3] below). For cpumask_next_and, direct benchmarking shows that it's 1.17x to 14x faster with a geometric mean of 2.1 on 32 CPUs [1]. No impact on memory usage. Note that on Arm, the new pure-C implementation still outperforms the old one that uses a mix of C and asm (`find_next_bit`) [3]. [1] Approximate benchmark code: ``` unsigned long src1p[nr_cpumask_longs] = {pattern1}; unsigned long src2p[nr_cpumask_longs] = {pattern2}; for (/*a bunch of repetitions*/) { for (int n = -1; n <= nr_cpu_ids; ++n) { asm volatile("" : "+rm"(src1p)); // prevent any optimization asm volatile("" : "+rm"(src2p)); unsigned long result = cpumask_next_and(n, src1p, src2p); asm volatile("" : "+rm"(result)); } } ``` Results: pattern1 pattern2 time_before/time_after 0x0000ffff 0x0000ffff 1.65 0x0000ffff 0x00005555 2.24 0x0000ffff 0x00001111 2.94 0x0000ffff 0x00000000 14.0 0x00005555 0x0000ffff 1.67 0x00005555 0x00005555 1.71 0x00005555 0x00001111 1.90 0x00005555 0x00000000 6.58 0x00001111 0x0000ffff 1.46 0x00001111 0x00005555 1.49 0x00001111 0x00001111 1.45 0x00001111 0x00000000 3.10 0x00000000 0x0000ffff 1.18 0x00000000 0x00005555 1.18 0x00000000 0x00001111 1.17 0x00000000 0x00000000 1.25 ----------------------------- geo.mean 2.06 [2] test_find_next_bit, X86 (skylake) [ 3913.477422] Start testing find_bit() with random-filled bitmap [ 3913.477847] find_next_bit: 160868 cycles, 16484 iterations [ 3913.477933] find_next_zero_bit: 169542 cycles, 16285 iterations [ 3913.478036] find_last_bit: 201638 cycles, 16483 iterations [ 3913.480214] find_first_bit: 4353244 cycles, 16484 iterations [ 3913.480216] Start testing find_next_and_bit() with random-filled bitmap [ 3913.481074] find_next_and_bit: 89604 cycles, 8216 iterations [ 3913.481075] Start testing find_bit() with sparse bitmap [ 3913.481078] find_next_bit: 2536 cycles, 66 iterations [ 3913.481252] find_next_zero_bit: 344404 cycles, 32703 iterations [ 3913.481255] find_last_bit: 2006 cycles, 66 iterations [ 3913.481265] find_first_bit: 17488 cycles, 66 iterations [ 3913.481266] Start testing find_next_and_bit() with sparse bitmap [ 3913.481272] find_next_and_bit: 764 cycles, 1 iterations [3] test_find_next_bit, arm (v7 odroid XU3). [ 267.206928] Start testing find_bit() with random-filled bitmap [ 267.214752] find_next_bit: 4474 cycles, 16419 iterations [ 267.221850] find_next_zero_bit: 5976 cycles, 16350 iterations [ 267.229294] find_last_bit: 4209 cycles, 16419 iterations [ 267.279131] find_first_bit: 1032991 cycles, 16420 iterations [ 267.286265] Start testing find_next_and_bit() with random-filled bitmap [ 267.302386] find_next_and_bit: 2290 cycles, 8140 iterations [ 267.309422] Start testing find_bit() with sparse bitmap [ 267.316054] find_next_bit: 191 cycles, 66 iterations [ 267.322726] find_next_zero_bit: 8758 cycles, 32703 iterations [ 267.329803] find_last_bit: 84 cycles, 66 iterations [ 267.336169] find_first_bit: 4118 cycles, 66 iterations [ 267.342627] Start testing find_next_and_bit() with sparse bitmap [ 267.356919] find_next_and_bit: 91 cycles, 1 iterations [courbet@google.com: v6] Link: http://lkml.kernel.org/r/20171129095715.23430-1-courbet@google.com [geert@linux-m68k.org: m68k/bitops: always include <asm-generic/bitops/find.h>] Link: http://lkml.kernel.org/r/1512556816-28627-1-git-send-email-geert@linux-m68k.org Link: http://lkml.kernel.org/r/20171128131334.23491-1-courbet@google.com Signed-off-by: Clement Courbet <courbet@google.com> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Yury Norov <ynorov@caviumnetworks.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-02-07 07:38:34 +08:00
return 0;
}
static int __init find_bit_test(void)
{
unsigned long nbits = BITMAP_LEN / SPARSE;
pr_err("\nStart testing find_bit() with random-filled bitmap\n");
get_random_bytes(bitmap, sizeof(bitmap));
lib: optimize cpumask_next_and() We've measured that we spend ~0.6% of sys cpu time in cpumask_next_and(). It's essentially a joined iteration in search for a non-zero bit, which is currently implemented as a lookup join (find a nonzero bit on the lhs, lookup the rhs to see if it's set there). Implement a direct join (find a nonzero bit on the incrementally built join). Also add generic bitmap benchmarks in the new `test_find_bit` module for new function (see `find_next_and_bit` in [2] and [3] below). For cpumask_next_and, direct benchmarking shows that it's 1.17x to 14x faster with a geometric mean of 2.1 on 32 CPUs [1]. No impact on memory usage. Note that on Arm, the new pure-C implementation still outperforms the old one that uses a mix of C and asm (`find_next_bit`) [3]. [1] Approximate benchmark code: ``` unsigned long src1p[nr_cpumask_longs] = {pattern1}; unsigned long src2p[nr_cpumask_longs] = {pattern2}; for (/*a bunch of repetitions*/) { for (int n = -1; n <= nr_cpu_ids; ++n) { asm volatile("" : "+rm"(src1p)); // prevent any optimization asm volatile("" : "+rm"(src2p)); unsigned long result = cpumask_next_and(n, src1p, src2p); asm volatile("" : "+rm"(result)); } } ``` Results: pattern1 pattern2 time_before/time_after 0x0000ffff 0x0000ffff 1.65 0x0000ffff 0x00005555 2.24 0x0000ffff 0x00001111 2.94 0x0000ffff 0x00000000 14.0 0x00005555 0x0000ffff 1.67 0x00005555 0x00005555 1.71 0x00005555 0x00001111 1.90 0x00005555 0x00000000 6.58 0x00001111 0x0000ffff 1.46 0x00001111 0x00005555 1.49 0x00001111 0x00001111 1.45 0x00001111 0x00000000 3.10 0x00000000 0x0000ffff 1.18 0x00000000 0x00005555 1.18 0x00000000 0x00001111 1.17 0x00000000 0x00000000 1.25 ----------------------------- geo.mean 2.06 [2] test_find_next_bit, X86 (skylake) [ 3913.477422] Start testing find_bit() with random-filled bitmap [ 3913.477847] find_next_bit: 160868 cycles, 16484 iterations [ 3913.477933] find_next_zero_bit: 169542 cycles, 16285 iterations [ 3913.478036] find_last_bit: 201638 cycles, 16483 iterations [ 3913.480214] find_first_bit: 4353244 cycles, 16484 iterations [ 3913.480216] Start testing find_next_and_bit() with random-filled bitmap [ 3913.481074] find_next_and_bit: 89604 cycles, 8216 iterations [ 3913.481075] Start testing find_bit() with sparse bitmap [ 3913.481078] find_next_bit: 2536 cycles, 66 iterations [ 3913.481252] find_next_zero_bit: 344404 cycles, 32703 iterations [ 3913.481255] find_last_bit: 2006 cycles, 66 iterations [ 3913.481265] find_first_bit: 17488 cycles, 66 iterations [ 3913.481266] Start testing find_next_and_bit() with sparse bitmap [ 3913.481272] find_next_and_bit: 764 cycles, 1 iterations [3] test_find_next_bit, arm (v7 odroid XU3). [ 267.206928] Start testing find_bit() with random-filled bitmap [ 267.214752] find_next_bit: 4474 cycles, 16419 iterations [ 267.221850] find_next_zero_bit: 5976 cycles, 16350 iterations [ 267.229294] find_last_bit: 4209 cycles, 16419 iterations [ 267.279131] find_first_bit: 1032991 cycles, 16420 iterations [ 267.286265] Start testing find_next_and_bit() with random-filled bitmap [ 267.302386] find_next_and_bit: 2290 cycles, 8140 iterations [ 267.309422] Start testing find_bit() with sparse bitmap [ 267.316054] find_next_bit: 191 cycles, 66 iterations [ 267.322726] find_next_zero_bit: 8758 cycles, 32703 iterations [ 267.329803] find_last_bit: 84 cycles, 66 iterations [ 267.336169] find_first_bit: 4118 cycles, 66 iterations [ 267.342627] Start testing find_next_and_bit() with sparse bitmap [ 267.356919] find_next_and_bit: 91 cycles, 1 iterations [courbet@google.com: v6] Link: http://lkml.kernel.org/r/20171129095715.23430-1-courbet@google.com [geert@linux-m68k.org: m68k/bitops: always include <asm-generic/bitops/find.h>] Link: http://lkml.kernel.org/r/1512556816-28627-1-git-send-email-geert@linux-m68k.org Link: http://lkml.kernel.org/r/20171128131334.23491-1-courbet@google.com Signed-off-by: Clement Courbet <courbet@google.com> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Yury Norov <ynorov@caviumnetworks.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-02-07 07:38:34 +08:00
get_random_bytes(bitmap2, sizeof(bitmap2));
test_find_next_bit(bitmap, BITMAP_LEN);
test_find_next_zero_bit(bitmap, BITMAP_LEN);
test_find_last_bit(bitmap, BITMAP_LEN);
test_find_nth_bit(bitmap, BITMAP_LEN / 10);
/*
* test_find_first_bit() may take some time, so
* traverse only part of bitmap to avoid soft lockup.
*/
test_find_first_bit(bitmap, BITMAP_LEN / 10);
test_find_first_and_bit(bitmap, bitmap2, BITMAP_LEN / 2);
lib: optimize cpumask_next_and() We've measured that we spend ~0.6% of sys cpu time in cpumask_next_and(). It's essentially a joined iteration in search for a non-zero bit, which is currently implemented as a lookup join (find a nonzero bit on the lhs, lookup the rhs to see if it's set there). Implement a direct join (find a nonzero bit on the incrementally built join). Also add generic bitmap benchmarks in the new `test_find_bit` module for new function (see `find_next_and_bit` in [2] and [3] below). For cpumask_next_and, direct benchmarking shows that it's 1.17x to 14x faster with a geometric mean of 2.1 on 32 CPUs [1]. No impact on memory usage. Note that on Arm, the new pure-C implementation still outperforms the old one that uses a mix of C and asm (`find_next_bit`) [3]. [1] Approximate benchmark code: ``` unsigned long src1p[nr_cpumask_longs] = {pattern1}; unsigned long src2p[nr_cpumask_longs] = {pattern2}; for (/*a bunch of repetitions*/) { for (int n = -1; n <= nr_cpu_ids; ++n) { asm volatile("" : "+rm"(src1p)); // prevent any optimization asm volatile("" : "+rm"(src2p)); unsigned long result = cpumask_next_and(n, src1p, src2p); asm volatile("" : "+rm"(result)); } } ``` Results: pattern1 pattern2 time_before/time_after 0x0000ffff 0x0000ffff 1.65 0x0000ffff 0x00005555 2.24 0x0000ffff 0x00001111 2.94 0x0000ffff 0x00000000 14.0 0x00005555 0x0000ffff 1.67 0x00005555 0x00005555 1.71 0x00005555 0x00001111 1.90 0x00005555 0x00000000 6.58 0x00001111 0x0000ffff 1.46 0x00001111 0x00005555 1.49 0x00001111 0x00001111 1.45 0x00001111 0x00000000 3.10 0x00000000 0x0000ffff 1.18 0x00000000 0x00005555 1.18 0x00000000 0x00001111 1.17 0x00000000 0x00000000 1.25 ----------------------------- geo.mean 2.06 [2] test_find_next_bit, X86 (skylake) [ 3913.477422] Start testing find_bit() with random-filled bitmap [ 3913.477847] find_next_bit: 160868 cycles, 16484 iterations [ 3913.477933] find_next_zero_bit: 169542 cycles, 16285 iterations [ 3913.478036] find_last_bit: 201638 cycles, 16483 iterations [ 3913.480214] find_first_bit: 4353244 cycles, 16484 iterations [ 3913.480216] Start testing find_next_and_bit() with random-filled bitmap [ 3913.481074] find_next_and_bit: 89604 cycles, 8216 iterations [ 3913.481075] Start testing find_bit() with sparse bitmap [ 3913.481078] find_next_bit: 2536 cycles, 66 iterations [ 3913.481252] find_next_zero_bit: 344404 cycles, 32703 iterations [ 3913.481255] find_last_bit: 2006 cycles, 66 iterations [ 3913.481265] find_first_bit: 17488 cycles, 66 iterations [ 3913.481266] Start testing find_next_and_bit() with sparse bitmap [ 3913.481272] find_next_and_bit: 764 cycles, 1 iterations [3] test_find_next_bit, arm (v7 odroid XU3). [ 267.206928] Start testing find_bit() with random-filled bitmap [ 267.214752] find_next_bit: 4474 cycles, 16419 iterations [ 267.221850] find_next_zero_bit: 5976 cycles, 16350 iterations [ 267.229294] find_last_bit: 4209 cycles, 16419 iterations [ 267.279131] find_first_bit: 1032991 cycles, 16420 iterations [ 267.286265] Start testing find_next_and_bit() with random-filled bitmap [ 267.302386] find_next_and_bit: 2290 cycles, 8140 iterations [ 267.309422] Start testing find_bit() with sparse bitmap [ 267.316054] find_next_bit: 191 cycles, 66 iterations [ 267.322726] find_next_zero_bit: 8758 cycles, 32703 iterations [ 267.329803] find_last_bit: 84 cycles, 66 iterations [ 267.336169] find_first_bit: 4118 cycles, 66 iterations [ 267.342627] Start testing find_next_and_bit() with sparse bitmap [ 267.356919] find_next_and_bit: 91 cycles, 1 iterations [courbet@google.com: v6] Link: http://lkml.kernel.org/r/20171129095715.23430-1-courbet@google.com [geert@linux-m68k.org: m68k/bitops: always include <asm-generic/bitops/find.h>] Link: http://lkml.kernel.org/r/1512556816-28627-1-git-send-email-geert@linux-m68k.org Link: http://lkml.kernel.org/r/20171128131334.23491-1-courbet@google.com Signed-off-by: Clement Courbet <courbet@google.com> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Yury Norov <ynorov@caviumnetworks.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-02-07 07:38:34 +08:00
test_find_next_and_bit(bitmap, bitmap2, BITMAP_LEN);
pr_err("\nStart testing find_bit() with sparse bitmap\n");
bitmap_zero(bitmap, BITMAP_LEN);
lib: optimize cpumask_next_and() We've measured that we spend ~0.6% of sys cpu time in cpumask_next_and(). It's essentially a joined iteration in search for a non-zero bit, which is currently implemented as a lookup join (find a nonzero bit on the lhs, lookup the rhs to see if it's set there). Implement a direct join (find a nonzero bit on the incrementally built join). Also add generic bitmap benchmarks in the new `test_find_bit` module for new function (see `find_next_and_bit` in [2] and [3] below). For cpumask_next_and, direct benchmarking shows that it's 1.17x to 14x faster with a geometric mean of 2.1 on 32 CPUs [1]. No impact on memory usage. Note that on Arm, the new pure-C implementation still outperforms the old one that uses a mix of C and asm (`find_next_bit`) [3]. [1] Approximate benchmark code: ``` unsigned long src1p[nr_cpumask_longs] = {pattern1}; unsigned long src2p[nr_cpumask_longs] = {pattern2}; for (/*a bunch of repetitions*/) { for (int n = -1; n <= nr_cpu_ids; ++n) { asm volatile("" : "+rm"(src1p)); // prevent any optimization asm volatile("" : "+rm"(src2p)); unsigned long result = cpumask_next_and(n, src1p, src2p); asm volatile("" : "+rm"(result)); } } ``` Results: pattern1 pattern2 time_before/time_after 0x0000ffff 0x0000ffff 1.65 0x0000ffff 0x00005555 2.24 0x0000ffff 0x00001111 2.94 0x0000ffff 0x00000000 14.0 0x00005555 0x0000ffff 1.67 0x00005555 0x00005555 1.71 0x00005555 0x00001111 1.90 0x00005555 0x00000000 6.58 0x00001111 0x0000ffff 1.46 0x00001111 0x00005555 1.49 0x00001111 0x00001111 1.45 0x00001111 0x00000000 3.10 0x00000000 0x0000ffff 1.18 0x00000000 0x00005555 1.18 0x00000000 0x00001111 1.17 0x00000000 0x00000000 1.25 ----------------------------- geo.mean 2.06 [2] test_find_next_bit, X86 (skylake) [ 3913.477422] Start testing find_bit() with random-filled bitmap [ 3913.477847] find_next_bit: 160868 cycles, 16484 iterations [ 3913.477933] find_next_zero_bit: 169542 cycles, 16285 iterations [ 3913.478036] find_last_bit: 201638 cycles, 16483 iterations [ 3913.480214] find_first_bit: 4353244 cycles, 16484 iterations [ 3913.480216] Start testing find_next_and_bit() with random-filled bitmap [ 3913.481074] find_next_and_bit: 89604 cycles, 8216 iterations [ 3913.481075] Start testing find_bit() with sparse bitmap [ 3913.481078] find_next_bit: 2536 cycles, 66 iterations [ 3913.481252] find_next_zero_bit: 344404 cycles, 32703 iterations [ 3913.481255] find_last_bit: 2006 cycles, 66 iterations [ 3913.481265] find_first_bit: 17488 cycles, 66 iterations [ 3913.481266] Start testing find_next_and_bit() with sparse bitmap [ 3913.481272] find_next_and_bit: 764 cycles, 1 iterations [3] test_find_next_bit, arm (v7 odroid XU3). [ 267.206928] Start testing find_bit() with random-filled bitmap [ 267.214752] find_next_bit: 4474 cycles, 16419 iterations [ 267.221850] find_next_zero_bit: 5976 cycles, 16350 iterations [ 267.229294] find_last_bit: 4209 cycles, 16419 iterations [ 267.279131] find_first_bit: 1032991 cycles, 16420 iterations [ 267.286265] Start testing find_next_and_bit() with random-filled bitmap [ 267.302386] find_next_and_bit: 2290 cycles, 8140 iterations [ 267.309422] Start testing find_bit() with sparse bitmap [ 267.316054] find_next_bit: 191 cycles, 66 iterations [ 267.322726] find_next_zero_bit: 8758 cycles, 32703 iterations [ 267.329803] find_last_bit: 84 cycles, 66 iterations [ 267.336169] find_first_bit: 4118 cycles, 66 iterations [ 267.342627] Start testing find_next_and_bit() with sparse bitmap [ 267.356919] find_next_and_bit: 91 cycles, 1 iterations [courbet@google.com: v6] Link: http://lkml.kernel.org/r/20171129095715.23430-1-courbet@google.com [geert@linux-m68k.org: m68k/bitops: always include <asm-generic/bitops/find.h>] Link: http://lkml.kernel.org/r/1512556816-28627-1-git-send-email-geert@linux-m68k.org Link: http://lkml.kernel.org/r/20171128131334.23491-1-courbet@google.com Signed-off-by: Clement Courbet <courbet@google.com> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Yury Norov <ynorov@caviumnetworks.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-02-07 07:38:34 +08:00
bitmap_zero(bitmap2, BITMAP_LEN);
lib: optimize cpumask_next_and() We've measured that we spend ~0.6% of sys cpu time in cpumask_next_and(). It's essentially a joined iteration in search for a non-zero bit, which is currently implemented as a lookup join (find a nonzero bit on the lhs, lookup the rhs to see if it's set there). Implement a direct join (find a nonzero bit on the incrementally built join). Also add generic bitmap benchmarks in the new `test_find_bit` module for new function (see `find_next_and_bit` in [2] and [3] below). For cpumask_next_and, direct benchmarking shows that it's 1.17x to 14x faster with a geometric mean of 2.1 on 32 CPUs [1]. No impact on memory usage. Note that on Arm, the new pure-C implementation still outperforms the old one that uses a mix of C and asm (`find_next_bit`) [3]. [1] Approximate benchmark code: ``` unsigned long src1p[nr_cpumask_longs] = {pattern1}; unsigned long src2p[nr_cpumask_longs] = {pattern2}; for (/*a bunch of repetitions*/) { for (int n = -1; n <= nr_cpu_ids; ++n) { asm volatile("" : "+rm"(src1p)); // prevent any optimization asm volatile("" : "+rm"(src2p)); unsigned long result = cpumask_next_and(n, src1p, src2p); asm volatile("" : "+rm"(result)); } } ``` Results: pattern1 pattern2 time_before/time_after 0x0000ffff 0x0000ffff 1.65 0x0000ffff 0x00005555 2.24 0x0000ffff 0x00001111 2.94 0x0000ffff 0x00000000 14.0 0x00005555 0x0000ffff 1.67 0x00005555 0x00005555 1.71 0x00005555 0x00001111 1.90 0x00005555 0x00000000 6.58 0x00001111 0x0000ffff 1.46 0x00001111 0x00005555 1.49 0x00001111 0x00001111 1.45 0x00001111 0x00000000 3.10 0x00000000 0x0000ffff 1.18 0x00000000 0x00005555 1.18 0x00000000 0x00001111 1.17 0x00000000 0x00000000 1.25 ----------------------------- geo.mean 2.06 [2] test_find_next_bit, X86 (skylake) [ 3913.477422] Start testing find_bit() with random-filled bitmap [ 3913.477847] find_next_bit: 160868 cycles, 16484 iterations [ 3913.477933] find_next_zero_bit: 169542 cycles, 16285 iterations [ 3913.478036] find_last_bit: 201638 cycles, 16483 iterations [ 3913.480214] find_first_bit: 4353244 cycles, 16484 iterations [ 3913.480216] Start testing find_next_and_bit() with random-filled bitmap [ 3913.481074] find_next_and_bit: 89604 cycles, 8216 iterations [ 3913.481075] Start testing find_bit() with sparse bitmap [ 3913.481078] find_next_bit: 2536 cycles, 66 iterations [ 3913.481252] find_next_zero_bit: 344404 cycles, 32703 iterations [ 3913.481255] find_last_bit: 2006 cycles, 66 iterations [ 3913.481265] find_first_bit: 17488 cycles, 66 iterations [ 3913.481266] Start testing find_next_and_bit() with sparse bitmap [ 3913.481272] find_next_and_bit: 764 cycles, 1 iterations [3] test_find_next_bit, arm (v7 odroid XU3). [ 267.206928] Start testing find_bit() with random-filled bitmap [ 267.214752] find_next_bit: 4474 cycles, 16419 iterations [ 267.221850] find_next_zero_bit: 5976 cycles, 16350 iterations [ 267.229294] find_last_bit: 4209 cycles, 16419 iterations [ 267.279131] find_first_bit: 1032991 cycles, 16420 iterations [ 267.286265] Start testing find_next_and_bit() with random-filled bitmap [ 267.302386] find_next_and_bit: 2290 cycles, 8140 iterations [ 267.309422] Start testing find_bit() with sparse bitmap [ 267.316054] find_next_bit: 191 cycles, 66 iterations [ 267.322726] find_next_zero_bit: 8758 cycles, 32703 iterations [ 267.329803] find_last_bit: 84 cycles, 66 iterations [ 267.336169] find_first_bit: 4118 cycles, 66 iterations [ 267.342627] Start testing find_next_and_bit() with sparse bitmap [ 267.356919] find_next_and_bit: 91 cycles, 1 iterations [courbet@google.com: v6] Link: http://lkml.kernel.org/r/20171129095715.23430-1-courbet@google.com [geert@linux-m68k.org: m68k/bitops: always include <asm-generic/bitops/find.h>] Link: http://lkml.kernel.org/r/1512556816-28627-1-git-send-email-geert@linux-m68k.org Link: http://lkml.kernel.org/r/20171128131334.23491-1-courbet@google.com Signed-off-by: Clement Courbet <courbet@google.com> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Yury Norov <ynorov@caviumnetworks.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-02-07 07:38:34 +08:00
while (nbits--) {
__set_bit(get_random_u32_below(BITMAP_LEN), bitmap);
__set_bit(get_random_u32_below(BITMAP_LEN), bitmap2);
lib: optimize cpumask_next_and() We've measured that we spend ~0.6% of sys cpu time in cpumask_next_and(). It's essentially a joined iteration in search for a non-zero bit, which is currently implemented as a lookup join (find a nonzero bit on the lhs, lookup the rhs to see if it's set there). Implement a direct join (find a nonzero bit on the incrementally built join). Also add generic bitmap benchmarks in the new `test_find_bit` module for new function (see `find_next_and_bit` in [2] and [3] below). For cpumask_next_and, direct benchmarking shows that it's 1.17x to 14x faster with a geometric mean of 2.1 on 32 CPUs [1]. No impact on memory usage. Note that on Arm, the new pure-C implementation still outperforms the old one that uses a mix of C and asm (`find_next_bit`) [3]. [1] Approximate benchmark code: ``` unsigned long src1p[nr_cpumask_longs] = {pattern1}; unsigned long src2p[nr_cpumask_longs] = {pattern2}; for (/*a bunch of repetitions*/) { for (int n = -1; n <= nr_cpu_ids; ++n) { asm volatile("" : "+rm"(src1p)); // prevent any optimization asm volatile("" : "+rm"(src2p)); unsigned long result = cpumask_next_and(n, src1p, src2p); asm volatile("" : "+rm"(result)); } } ``` Results: pattern1 pattern2 time_before/time_after 0x0000ffff 0x0000ffff 1.65 0x0000ffff 0x00005555 2.24 0x0000ffff 0x00001111 2.94 0x0000ffff 0x00000000 14.0 0x00005555 0x0000ffff 1.67 0x00005555 0x00005555 1.71 0x00005555 0x00001111 1.90 0x00005555 0x00000000 6.58 0x00001111 0x0000ffff 1.46 0x00001111 0x00005555 1.49 0x00001111 0x00001111 1.45 0x00001111 0x00000000 3.10 0x00000000 0x0000ffff 1.18 0x00000000 0x00005555 1.18 0x00000000 0x00001111 1.17 0x00000000 0x00000000 1.25 ----------------------------- geo.mean 2.06 [2] test_find_next_bit, X86 (skylake) [ 3913.477422] Start testing find_bit() with random-filled bitmap [ 3913.477847] find_next_bit: 160868 cycles, 16484 iterations [ 3913.477933] find_next_zero_bit: 169542 cycles, 16285 iterations [ 3913.478036] find_last_bit: 201638 cycles, 16483 iterations [ 3913.480214] find_first_bit: 4353244 cycles, 16484 iterations [ 3913.480216] Start testing find_next_and_bit() with random-filled bitmap [ 3913.481074] find_next_and_bit: 89604 cycles, 8216 iterations [ 3913.481075] Start testing find_bit() with sparse bitmap [ 3913.481078] find_next_bit: 2536 cycles, 66 iterations [ 3913.481252] find_next_zero_bit: 344404 cycles, 32703 iterations [ 3913.481255] find_last_bit: 2006 cycles, 66 iterations [ 3913.481265] find_first_bit: 17488 cycles, 66 iterations [ 3913.481266] Start testing find_next_and_bit() with sparse bitmap [ 3913.481272] find_next_and_bit: 764 cycles, 1 iterations [3] test_find_next_bit, arm (v7 odroid XU3). [ 267.206928] Start testing find_bit() with random-filled bitmap [ 267.214752] find_next_bit: 4474 cycles, 16419 iterations [ 267.221850] find_next_zero_bit: 5976 cycles, 16350 iterations [ 267.229294] find_last_bit: 4209 cycles, 16419 iterations [ 267.279131] find_first_bit: 1032991 cycles, 16420 iterations [ 267.286265] Start testing find_next_and_bit() with random-filled bitmap [ 267.302386] find_next_and_bit: 2290 cycles, 8140 iterations [ 267.309422] Start testing find_bit() with sparse bitmap [ 267.316054] find_next_bit: 191 cycles, 66 iterations [ 267.322726] find_next_zero_bit: 8758 cycles, 32703 iterations [ 267.329803] find_last_bit: 84 cycles, 66 iterations [ 267.336169] find_first_bit: 4118 cycles, 66 iterations [ 267.342627] Start testing find_next_and_bit() with sparse bitmap [ 267.356919] find_next_and_bit: 91 cycles, 1 iterations [courbet@google.com: v6] Link: http://lkml.kernel.org/r/20171129095715.23430-1-courbet@google.com [geert@linux-m68k.org: m68k/bitops: always include <asm-generic/bitops/find.h>] Link: http://lkml.kernel.org/r/1512556816-28627-1-git-send-email-geert@linux-m68k.org Link: http://lkml.kernel.org/r/20171128131334.23491-1-courbet@google.com Signed-off-by: Clement Courbet <courbet@google.com> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Yury Norov <ynorov@caviumnetworks.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-02-07 07:38:34 +08:00
}
test_find_next_bit(bitmap, BITMAP_LEN);
test_find_next_zero_bit(bitmap, BITMAP_LEN);
test_find_last_bit(bitmap, BITMAP_LEN);
test_find_nth_bit(bitmap, BITMAP_LEN);
test_find_first_bit(bitmap, BITMAP_LEN);
test_find_first_and_bit(bitmap, bitmap2, BITMAP_LEN);
lib: optimize cpumask_next_and() We've measured that we spend ~0.6% of sys cpu time in cpumask_next_and(). It's essentially a joined iteration in search for a non-zero bit, which is currently implemented as a lookup join (find a nonzero bit on the lhs, lookup the rhs to see if it's set there). Implement a direct join (find a nonzero bit on the incrementally built join). Also add generic bitmap benchmarks in the new `test_find_bit` module for new function (see `find_next_and_bit` in [2] and [3] below). For cpumask_next_and, direct benchmarking shows that it's 1.17x to 14x faster with a geometric mean of 2.1 on 32 CPUs [1]. No impact on memory usage. Note that on Arm, the new pure-C implementation still outperforms the old one that uses a mix of C and asm (`find_next_bit`) [3]. [1] Approximate benchmark code: ``` unsigned long src1p[nr_cpumask_longs] = {pattern1}; unsigned long src2p[nr_cpumask_longs] = {pattern2}; for (/*a bunch of repetitions*/) { for (int n = -1; n <= nr_cpu_ids; ++n) { asm volatile("" : "+rm"(src1p)); // prevent any optimization asm volatile("" : "+rm"(src2p)); unsigned long result = cpumask_next_and(n, src1p, src2p); asm volatile("" : "+rm"(result)); } } ``` Results: pattern1 pattern2 time_before/time_after 0x0000ffff 0x0000ffff 1.65 0x0000ffff 0x00005555 2.24 0x0000ffff 0x00001111 2.94 0x0000ffff 0x00000000 14.0 0x00005555 0x0000ffff 1.67 0x00005555 0x00005555 1.71 0x00005555 0x00001111 1.90 0x00005555 0x00000000 6.58 0x00001111 0x0000ffff 1.46 0x00001111 0x00005555 1.49 0x00001111 0x00001111 1.45 0x00001111 0x00000000 3.10 0x00000000 0x0000ffff 1.18 0x00000000 0x00005555 1.18 0x00000000 0x00001111 1.17 0x00000000 0x00000000 1.25 ----------------------------- geo.mean 2.06 [2] test_find_next_bit, X86 (skylake) [ 3913.477422] Start testing find_bit() with random-filled bitmap [ 3913.477847] find_next_bit: 160868 cycles, 16484 iterations [ 3913.477933] find_next_zero_bit: 169542 cycles, 16285 iterations [ 3913.478036] find_last_bit: 201638 cycles, 16483 iterations [ 3913.480214] find_first_bit: 4353244 cycles, 16484 iterations [ 3913.480216] Start testing find_next_and_bit() with random-filled bitmap [ 3913.481074] find_next_and_bit: 89604 cycles, 8216 iterations [ 3913.481075] Start testing find_bit() with sparse bitmap [ 3913.481078] find_next_bit: 2536 cycles, 66 iterations [ 3913.481252] find_next_zero_bit: 344404 cycles, 32703 iterations [ 3913.481255] find_last_bit: 2006 cycles, 66 iterations [ 3913.481265] find_first_bit: 17488 cycles, 66 iterations [ 3913.481266] Start testing find_next_and_bit() with sparse bitmap [ 3913.481272] find_next_and_bit: 764 cycles, 1 iterations [3] test_find_next_bit, arm (v7 odroid XU3). [ 267.206928] Start testing find_bit() with random-filled bitmap [ 267.214752] find_next_bit: 4474 cycles, 16419 iterations [ 267.221850] find_next_zero_bit: 5976 cycles, 16350 iterations [ 267.229294] find_last_bit: 4209 cycles, 16419 iterations [ 267.279131] find_first_bit: 1032991 cycles, 16420 iterations [ 267.286265] Start testing find_next_and_bit() with random-filled bitmap [ 267.302386] find_next_and_bit: 2290 cycles, 8140 iterations [ 267.309422] Start testing find_bit() with sparse bitmap [ 267.316054] find_next_bit: 191 cycles, 66 iterations [ 267.322726] find_next_zero_bit: 8758 cycles, 32703 iterations [ 267.329803] find_last_bit: 84 cycles, 66 iterations [ 267.336169] find_first_bit: 4118 cycles, 66 iterations [ 267.342627] Start testing find_next_and_bit() with sparse bitmap [ 267.356919] find_next_and_bit: 91 cycles, 1 iterations [courbet@google.com: v6] Link: http://lkml.kernel.org/r/20171129095715.23430-1-courbet@google.com [geert@linux-m68k.org: m68k/bitops: always include <asm-generic/bitops/find.h>] Link: http://lkml.kernel.org/r/1512556816-28627-1-git-send-email-geert@linux-m68k.org Link: http://lkml.kernel.org/r/20171128131334.23491-1-courbet@google.com Signed-off-by: Clement Courbet <courbet@google.com> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Yury Norov <ynorov@caviumnetworks.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-02-07 07:38:34 +08:00
test_find_next_and_bit(bitmap, bitmap2, BITMAP_LEN);
/*
* Everything is OK. Return error just to let user run benchmark
* again without annoying rmmod.
*/
return -EINVAL;
}
module_init(find_bit_test);
MODULE_LICENSE("GPL");