Skip to content

Instantly share code, notes, and snippets.

View tanakamura's full-sized avatar

Takashi Nakamura tanakamura

View GitHub Profile
ostimer: clock_gettime
userland_timer: rdtscp
perf_counter: yes
Intel(R) Core(TM) i5-8250U CPU @ 1.60GHz
==== membw_1t ====
| |MiB/sec
=================================
| simple-long-copy |14401.82962
---------------------------------
| libc-memcpy |18055.36681
ostimer: clock_gettime
userland_timer: rdtscp
use_cpucyle: no
==== membw_1t ====
| |MiB/sec
=================================
| simple-long-copy |16795.94556
---------------------------------
| libc-memcpy |32935.04326
---------------------------------
@tanakamura
tanakamura / tgl.txt
Created December 12, 2020 13:21
membw-release
ostimer: clock_gettime
userland_timer: rdtscp
use_cpucyle: no
==== membw_1t ====
| |MiB/sec
=================================
| simple-long-copy |23061.14028
---------------------------------
| libc-memcpy |30647.25286
---------------------------------
@tanakamura
tanakamura / tgl.txt
Created December 11, 2020 17:43
membw
ostimer: clock_gettime
userland_timer: rdtscp
use_cpucyle: no
==== membw_1t ====
| |MiB/sec
=================================
| simple-long-copy | 8626.01679
---------------------------------
| libc-memcpy |16827.83568
---------------------------------
@tanakamura
tanakamura / cfl-syscall-mitigationsoff.txt
Created December 6, 2020 18:41
syscall-mitigations=off
==== syscall ====
| | MCall/sec
================================
| close_minus1 | 9.58463
--------------------------------
| open_close | 1.10591
--------------------------------
| pipe_close | 0.57343
--------------------------------
| select_0 | 2.15039
@tanakamura
tanakamura / cfl-syscall.txt
Last active December 6, 2020 18:44
syscall
==== syscall ====
| | MCall/sec
================================
| close_minus1 | 3.78879
--------------------------------
| open_close | 0.77823
--------------------------------
| pipe_close | 0.43007
--------------------------------
| select_0 | 1.45407
@tanakamura
tanakamura / cfl-32.txt
Created December 6, 2020 15:48
int-div-latency
-> : divider_bit
| |27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|29|30|31|32
====================================================================================================
| 0 |25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25|25
----------------------------------------------------------------------------------------------------
| 1 |27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27
----------------------------------------------------------------------------------------------------
| 2 |27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27
----------------------------------------------------------------------------------------------------
| 3 |27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27|27
#include <unistd.h>
#include <fcntl.h>
#include <pthread.h>
#include <stdio.h>
static double getsec()
{
struct timespec ts;
clock_gettime(CLOCK_MONOTONIC, &ts);
num_thread = 1
libc-memset : 1024[ B] 109.797060[GB/s]
libc-memset : 2048[ B] 119.469974[GB/s]
libc-memset : 4096[ B] 124.585380[GB/s]
libc-memset : 8192[ B] 127.374256[GB/s]
libc-memset : 16[KB] 126.546911[GB/s]
libc-memset : 32[KB] 125.730241[GB/s]
libc-memset : 64[KB] 120.244104[GB/s]
libc-memset : 128[KB] 118.669685[GB/s]
libc-memset : 256[KB] 112.864779[GB/s]
Ryzen 7 3700x vs i7-6700
============= LATENCY ==============================================================================
instruction | IPC ( rel[%]), CPI ( rel[%])
------------------------------------------+---------------------------------------------------------
m128 addps | 0.33-0.25 ( 33.3[%]), 3.00-4.00 ( -25.0[%])
m128 aesdec | 0.25-0.25 ( 0.0[%]), 4.00-4.00 ( -0.0[%])
m128 aesdeclast | 0.25-0.25 ( -0.0[%]), 4.00-4.00 ( 0.0[%])
m128 aesenc | 0.25-0.25 ( 0.0[%]), 4.00-4.00 ( -0.0[%])
m128 aesenclast | 0.25-0.25 ( -0.0[%]), 4.00-4.00 ( 0.0[%])