Hi.
Em seg., 9 de mar. de 2026 às 14:02, Bryan Green <[email protected]>
escreveu:
> I performed a micro-benchmark on my dual epyc (zen 2) server and version 1
> wins for small values of n.
>
> 20 runs:
>
> n version min median mean max stddev noise%
> -----------------------------------------------------------------------
> n=1 version1 2.440 2.440 2.450 2.550 0.024 4.5%
> n=1 version2 4.260 4.280 4.277 4.290 0.007 0.7%
>
> n=2 version1 2.740 2.750 2.757 2.880 0.029 5.1%
> n=2 version2 3.970 3.980 3.980 4.020 0.010 1.3%
>
> n=4 version1 4.580 4.595 4.649 4.910 0.094 7.2%
> n=4 version2 5.780 5.815 5.809 5.820 0.013 0.7%
>
> But, micro-benchmarks always make me nervous, so I looked at the actual
> instruction cost for my
> platform given the version 1 and version 2 code.
>
> If we count cpu cycles using the AMD Zen 2 instruction latency/throughput
> tables: version 1 (loop body)
> has a critical path of ~5-6 cycles per iteration. version 2 (loop body)
> has ~3-4 cycles per iteration.
>
> The problem for version 2 is that the call to memcpy is ~24-30 cycles due
> to the stub + function call + return
> and branch predictor pressure on first call. This probably results in
> ~2.5 ns per iteration cost for version 2.
>
> So, no I wouldn't call it an optimization. But, it will be interesting to
> hear other opinions on this.
>
I made dirty and quick tests with two versions:
gcc 15.2.0
gcc -O2 memcpy1.c -o memcpy1
The first test was with keys 10000000 and 10000000 loops:
version1: on memcpy call
done in 1873 nanoseconds
version2: inlined memcpy
not finish
The second test was with keys 4 and 10000000 loops:
version1: one memcpy call
version2: inlined memcpy call
version1: done in 1519 nanoseconds
version2: done in 104981851 nanoseconds
(1.44692e-05 times faster)
version1: done in 1979 nanoseconds
version2: done in 110568901 nanoseconds
(1.78983e-05 times faster)
version1: done in 1814 nanoseconds
version2: done in 108555484 nanoseconds
(1.67103e-05 times faster)
version1: done in 1631 nanoseconds
version2: done in 109867919 nanoseconds
(1.48451e-05 times faster)
version1: done in 1269 nanoseconds
version2: done in 111639106 nanoseconds
(1.1367e-05 times faster)
Unless I'm doing something wrong, one call memcpy wins!
memcpy1.c attached.
best regards,
Ranier Vilela
#include <stdlib.h>
#include <string.h>
#include <stdint.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdio.h>
#include <time.h>
#include <immintrin.h>
/* Closer approximation of ScanKeyData - has function pointer and Datum */
typedef void (*RegProcedure)(void);
typedef uintptr_t Datum;
typedef struct ScanKeyData
{
int sk_flags;
int sk_attno;
RegProcedure sk_func;
Datum sk_argument;
} ScanKeyData;
/* */
const ScanKeyData * version1(int n, const ScanKeyData * key)
{
ScanKeyData *idxkey = (ScanKeyData *) malloc(n * sizeof(ScanKeyData));
memcpy(&idxkey, &key, n * sizeof(ScanKeyData));
for (int i = 0; i < n; i++)
{
idxkey[i].sk_attno = i + 1;
}
return idxkey;
}
/* */
const ScanKeyData * version2(int n, const ScanKeyData *key)
{
ScanKeyData *idxkey = (ScanKeyData *) malloc(n * sizeof(ScanKeyData));
for (int i = 0; i < n; i++)
{
memcpy(&idxkey[i], &key[i], sizeof(ScanKeyData));
idxkey[i].sk_attno = i + 1;
}
return idxkey;
}
#define NANOSEC_PER_SEC 1000000000
// Returns difference in nanoseconds
int64_t
get_clock_diff(struct timespec *t1, struct timespec *t2)
{
int64_t nanosec = (t1->tv_sec - t2->tv_sec) * NANOSEC_PER_SEC;
nanosec += (t1->tv_nsec - t2->tv_nsec);
return nanosec;
}
//#define NKEYS 10000000 version2 does not finish
#define NKEYS 4
#define LOOPS 10000000
void test1(int n)
{
ScanKeyData *keys;
ScanKeyData *idx;
keys = (ScanKeyData *) malloc(NKEYS * sizeof(ScanKeyData));
memset(keys, 0, NKEYS * sizeof(ScanKeyData));
for(int i = 0; i < n; i++)
{
idx = version1(NKEYS, keys);
free(idx);
}
free(keys);
}
void test2(int n)
{
ScanKeyData *keys;
ScanKeyData *idx;
keys = (ScanKeyData *) malloc(NKEYS * sizeof(ScanKeyData));
memset(keys, 0, NKEYS * sizeof(ScanKeyData));
for(int i = 0; i < n; i++)
{
idx = version2(NKEYS, keys);
free(idx);
}
free(keys);
}
int main(void)
{
struct timespec start,end;
int64_t version1_time, version2_time;
clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &start);
test1(LOOPS);
clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &end);
version1_time = get_clock_diff(&end, &start);
printf("version1: done in %lld nanoseconds\n", version1_time);
clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &start);
test2(LOOPS);
clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &end);
version2_time = get_clock_diff(&end, &start);
printf("version2: done in %lld nanoseconds\n", version2_time);
printf("(%g times faster)\n", (double) version1_time / version2_time);
return 0;
}
