[Photo] Kunpeng 916-ARM64 architecture source code gcc compilation complete record

Reference: Kunpeng Software Stack Official Document

1. Preparation stage

View system architecture

[root@pc-deeplearning-1 ~]# hostnamectl 
   Static hostname: pc-deeplearning-1
         Icon name: computer-vm
           Chassis: vm
        Machine ID: 65a539173aca43bca524c9829d298907
           Boot ID: 0c07a5693c01472ba997bd83cb43f05b
    Virtualization: kvm
  Operating System: CentOS Linux 7 (AltArch)
       CPE OS Name: cpe:/o:centos:centos:7
            Kernel: Linux 4.14.0-115.el7a.0.1.aarch64
      Architecture: arm64
[root@pc-deeplearning-1 ~]#

View system version

[root@pc-deeplearning-1 ~]# cat /etc/redhat-release 
CentOS Linux release 7.6.1810 (AltArch) 
[root@pc-deeplearning-1 ~]#

View the number of virtual machine cpu cores

[root@pc-deeplearning-1 lib64]# lscpu
Architecture:          aarch64
Byte Order:            Little Endian
CPU(s):                32
On-line CPU(s) list:   0-31
Thread(s) per core:    1
Core(s) per socket:    16
Socket(s):             2
NUMA node(s):          1
Model:                 2
BogoMIPS:              100.00
NUMA node0 CPU(s):     0-31
Flags:                 fp asimd evtstrm aes pmull sha1 sha2 crc32 cpuid
[root@pc-deeplearning-1 lib64]#

View virtual machine memory capacity

[root@pc-deeplearning-1 lib64]# free -h
              total        used        free      shared  buff/cache   available
Mem:            31G        659M         12G        668M         19G         24G
Swap:            0B          0B          0B
[root@pc-deeplearning-1 lib64]#

2. Install cmake

2.1 Check whether cmake is installed

rpm -qa |grep cmake

If it is not installed, you need to install it.

2.2 Install cmake

[root@pc-deeplearning-1 src]# yum repolist
[root@pc-deeplearning-1 src]# yum provides cmake
Loaded plugins: fastestmirror, langpacks
Loading mirror speeds from cached hostfile
 * base: mirrors.huaweicloud.com
 * extras: mirrors.huaweicloud.com
 * updates: mirrors.huaweicloud.com
cmake-2.8.12.2-2.el7.aarch64 : Cross-platform make system
Repo        : base



[root@pc-deeplearning-1 src]# 
[root@pc-deeplearning-1 src]# yum -y install cmake
Complete!
[root@pc-deeplearning-1 src]#

2.3 View cmake version

[root@pc-deeplearning-1 src]# cmake -version
cmake version 2.8.12.2
[root@pc-deeplearning-1 src]#

3. Install dependency packages

3.1 Install the following dependencies

yum -y install bison* ncurses*

yum -y install bzip2 wget

3.2 Check whether the dependent installation package is installed successfully

[root@pc-deeplearning-1 src]# rpm -qa | grep bison*
libisofs-1.2.8-4.el7.aarch64
bison-3.0.4-2.el7.aarch64
[root@pc-deeplearning-1 src]# 
[root@pc-deeplearning-1 src]# rpm -qa | grep ncurses*
ncurses-libs-5.9-14.20130511.el7_4.aarch64
ncurses-base-5.9-14.20130511.el7_4.noarch
ncurses-5.9-14.20130511.el7_4.aarch64
[root@pc-deeplearning-1 src]#

[root@pc-deeplearning-1 src]# rpm -qa | grep bzip2
bzip2-1.0.6-13.el7.aarch64
bzip2-libs-1.0.6-13.el7.aarch64
[root@pc-deeplearning-1 src]# 
[root@pc-deeplearning-1 src]# rpm -qa | grep wget
wget-1.14-18.el7.aarch64
[root@pc-deeplearning-1 src]#

4. gcc download, compile and install

4.1 Check the version of gcc

[root@pc-deeplearning-1 src]# gcc --version
gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-36)
Copyright (C) 2015 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

[root@pc-deeplearning-1 src]#

The version is too low, need to download an updated version

4.2 download gcc7.3

[root@pc-deeplearning-1 software]# wget https://mirrors.tuna.tsinghua.edu.cn/gnu/gcc/gcc-7.3.0/gcc-7.3.0.tar.gz

2020-11-15 17:49:07 (440 KB/s) - ‘gcc-7.3.0.tar.gz’ saved [112201917/112201917]

[root@pc-deeplearning-1 software]#

4.3 Unzip the gcc7.3 installation package

[root@pc-deeplearning-1 software]# tar -zxf gcc-7.3.0.tar.gz
[root@pc-deeplearning-1 software]#

4.4 In the "gcc-7.3.0" directory, check whether the gcc dependency package has been downloaded and installed

[root@pc-deeplearning-1 software]# cd gcc-7.3.0/
[root@pc-deeplearning-1 gcc-7.3.0]# ./contrib/download_prerequisites 

gmp-6.1.0.tar.bz2: OK
mpfr-3.1.4.tar.bz2: OK
mpc-1.0.3.tar.gz: OK
isl-0.16.1.tar.bz2: OK
All prerequisites downloaded successfully.
[root@pc-deeplearning-1 gcc-7.3.0]#

You can see that isl, gmp, mpc, mpfr have been downloaded

4.5 Compile gcc.

The "-j" parameter can use multi-core CPU to speed up the compilation speed. In this operation, a 32-core CPU is used, so here is "-j32".

(Note: First, it was operated in accordance with the official documentation of Kunpeng Software Stack, make -j16, and then the speed was very slow; after listening to the guidance of the great god, Ctrl-C interrupted and then made -j32, the speed was very fast)

4.5.1 The number of CPU cores can be queried through the following commands:

[root@pc-deeplearning-1 gcc-7.3.0]# cat /proc/cpuinfo| grep "processor"| wc -l
32
[root@pc-deeplearning-1 gcc-7.3.0]#

4.5.2 View the current directory

[root@pc-deeplearning-1 gcc-7.3.0]# pwd
/opt/software/gcc-7.3.0
[root@pc-deeplearning-1 gcc-7.3.0]#

4.5.3 Create gcc compilation directory

[root@pc-deeplearning-1 gcc-7.3.0]# mkdir gcc-build-7.3.0

[root@pc-deeplearning-1 gcc-7.3.0]# cd gcc-build-7.3.0

[root@pc-deeplearning-1 gcc-build-7.3.0]# pwd
/opt/software/gcc-7.3.0/gcc-build-7.3.0
[root@pc-deeplearning-1 gcc-build-7.3.0]#

[root@pc-deeplearning-1 gcc-build-7.3.0]# ../configure --enable-checking=release --enable-language=c,c++ --disable-multilib --prefix=/usr

4.5.4 Because of 32vcpu, use make -j32 to compile and run

[root@pc-deeplearning-1 gcc-build-7.3.0]# make -j32

4.5.5 Perform installation

[root@pc-deeplearning-1 gcc-build-7.3.0]# make install

4.5.6 Make sure that "libstdc++.so" is in the "/usr/lib64" directory

[root@pc-deeplearning-1 gcc-build-7.3.0]# cd /usr/lib64

Query the location of "libstdc++.so"

[root@pc-deeplearning-1 lib64]# ll |grep libstdc++.so
lrwxrwxrwx.  1 root root       19 Nov 15 19:48 libstdc++.so -> libstdc++.so.6.0.24
lrwxrwxrwx.  1 root root       19 Nov 15 19:48 libstdc++.so.6 -> libstdc++.so.6.0.24
-rwxr-xr-x.  1 root root  1057320 Oct 30  2018 libstdc++.so.6.0.19
-rwxr-xr-x.  1 root root 11844176 Nov 15 19:48 libstdc++.so.6.0.24
-rw-r--r--.  1 root root     2385 Nov 15 19:48 libstdc++.so.6.0.24-gdb.py
lrwxrwxrwx.  1 root root       19 Mar 20  2020 libstdc++.so.6.old -> libstdc++.so.6.0.19
[root@pc-deeplearning-1 lib64]#

Make sure the soft connection exists

[root@pc-deeplearning-1 lib64]# ll |grep libstdc++.so.6.0.24
lrwxrwxrwx.  1 root root       19 Nov 15 19:48 libstdc++.so -> libstdc++.so.6.0.24
lrwxrwxrwx.  1 root root       19 Nov 15 19:48 libstdc++.so.6 -> libstdc++.so.6.0.24
-rwxr-xr-x.  1 root root 11844176 Nov 15 19:48 libstdc++.so.6.0.24
-rw-r--r--.  1 root root     2385 Nov 15 19:48 libstdc++.so.6.0.24-gdb.py
[root@pc-deeplearning-1 lib64]#

5. Check the gcc version

[root@pc-deeplearning-1 lib64]# gcc -v
Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/usr/libexec/gcc/aarch64-unknown-linux-gnu/7.3.0/lto-wrapper
Target: aarch64-unknown-linux-gnu
Configured with: ../configure --enable-checking=release --enable-language=c,c++ --disable-multilib --prefix=/usr
Thread model: posix
gcc version 7.3.0 (GCC) 
[root@pc-deeplearning-1 lib64]#

This concludes this article!

Contact information:

WeChat public account:
Insert picture description here
WeChat:

If you think this article is helpful to you, please click like and then go!
More exciting, stay tuned!
Welcome to follow this blogger!