高性能 ：OpenAI Triton Open-source GPU programming Language LINUX 环境配置
目录
- 配置triton环境
- cuda
- build-essential
- 带有pip的python环境
- 直接安装pip
- anaconda
- 安装 triton 环境
- pip install triton
- pip install torch
- 运行test示例
- vector-add.py
- launch.json
配置triton环境
cuda
- wget http://developer.download.nvidia.com/compute/cuda/11.0.2/local_installers/cuda_11.0.2_450.51.05_linux.run
- sudo sh cuda_11.0.2_450.51.05_linux.run
build-essential
sudo apt-get install build-essential:不安装可能会有报错，RuntimeError: Failed to find C compiler. Please specify via CC environment variable.
(tritonenv) cifi@cifi-666666:~/Downloads$ sudo apt-get install build-essential
Reading package lists... Done
Building dependency tree
Reading state information... Done
The following additional packages will be installed:cpp-9 dpkg-dev fakeroot g++ g++-9 gcc gcc-10-base gcc-9 gcc-9-base libalgorithm-diff-perl libalgorithm-diff-xs-perl libalgorithm-merge-perl libasan5 libatomic1 libc-dev-bin libc6 libc6-dbg libc6-devlibcc1-0 libcrypt-dev libfakeroot libgcc-9-dev libgcc-s1 libgomp1 libitm1 liblsan0 libquadmath0 libstdc++-9-dev libstdc++6 libtsan0 libubsan1 linux-libc-dev make manpages-dev
Suggested packages:gcc-9-locales debian-keyring g++-multilib g++-9-multilib gcc-9-doc gcc-multilib autoconf automake libtool flex bison gcc-doc gcc-9-multilib glibc-doc libstdc++-9-doc make-doc
The following NEW packages will be installed:build-essential dpkg-dev fakeroot g++ g++-9 gcc gcc-9 libalgorithm-diff-perl libalgorithm-diff-xs-perl libalgorithm-merge-perl libasan5 libc-dev-bin libc6-dev libcrypt-dev libfakeroot libgcc-9-devlibitm1 liblsan0 libquadmath0 libstdc++-9-dev libtsan0 libubsan1 linux-libc-dev make manpages-dev
The following packages will be upgraded:cpp-9 gcc-10-base gcc-9-base libatomic1 libc6 libc6-dbg libcc1-0 libgcc-s1 libgomp1 libstdc++6
10 upgraded, 25 newly installed, 0 to remove and 395 not upgraded.
Need to get 65.6 MB of archives.
After this operation, 157 MB of additional disk space will be used.
Do you want to continue? [Y/n] y
Get:1 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libc6-dbg amd64 2.31-0ubuntu9.17 [20.2 MB]
Get:2 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libatomic1 amd64 10.5.0-1ubuntu1~20.04 [9,284 B]
Get:3 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 gcc-10-base amd64 10.5.0-1ubuntu1~20.04 [20.8 kB]
Get:4 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libstdc++6 amd64 10.5.0-1ubuntu1~20.04 [501 kB]
Get:5 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libgomp1 amd64 10.5.0-1ubuntu1~20.04 [102 kB]
Get:6 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libcc1-0 amd64 10.5.0-1ubuntu1~20.04 [48.8 kB]
Get:7 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libgcc-s1 amd64 10.5.0-1ubuntu1~20.04 [41.8 kB]
Get:8 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libc6 amd64 2.31-0ubuntu9.17 [2,721 kB]
Get:9 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libc-dev-bin amd64 2.31-0ubuntu9.17 [71.8 kB]
Get:10 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 linux-libc-dev amd64 5.4.0-205.225 [1,125 kB]
Get:11 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal/main amd64 libcrypt-dev amd64 1:4.4.10-10ubuntu4 [104 kB]
Get:12 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libc6-dev amd64 2.31-0ubuntu9.17 [2,521 kB]
Get:13 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 cpp-9 amd64 9.4.0-1ubuntu1~20.04.2 [7,502 kB]
Get:14 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 gcc-9-base amd64 9.4.0-1ubuntu1~20.04.2 [18.9 kB]
Get:15 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libitm1 amd64 10.5.0-1ubuntu1~20.04 [26.2 kB]
Get:16 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libasan5 amd64 9.4.0-1ubuntu1~20.04.2 [2,752 kB]
Get:17 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 liblsan0 amd64 10.5.0-1ubuntu1~20.04 [835 kB]
Get:18 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libtsan0 amd64 10.5.0-1ubuntu1~20.04 [2,016 kB]
Get:19 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libubsan1 amd64 10.5.0-1ubuntu1~20.04 [785 kB]
Get:20 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libquadmath0 amd64 10.5.0-1ubuntu1~20.04 [146 kB]
Get:21 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libgcc-9-dev amd64 9.4.0-1ubuntu1~20.04.2 [2,359 kB]
Get:22 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 gcc-9 amd64 9.4.0-1ubuntu1~20.04.2 [8,276 kB]
Get:23 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal/main amd64 gcc amd64 4:9.3.0-1ubuntu2 [5,208 B]
Get:24 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 libstdc++-9-dev amd64 9.4.0-1ubuntu1~20.04.2 [1,722 kB]
Get:25 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 g++-9 amd64 9.4.0-1ubuntu1~20.04.2 [8,421 kB]
Get:26 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal/main amd64 g++ amd64 4:9.3.0-1ubuntu2 [1,604 B]
Get:27 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal/main amd64 make amd64 4.2.1-1.2 [162 kB]
Get:28 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 dpkg-dev all 1.19.7ubuntu3.2 [679 kB]
Get:29 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal-updates/main amd64 build-essential amd64 12.8ubuntu1.1 [4,664 B]
Get:30 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal/main amd64 libfakeroot amd64 1.24-1 [25.7 kB]
Get:31 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal/main amd64 fakeroot amd64 1.24-1 [62.6 kB]
Get:32 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal/main amd64 libalgorithm-diff-perl all 1.19.03-2 [46.6 kB]
Get:33 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal/main amd64 libalgorithm-diff-xs-perl amd64 0.04-6 [11.3 kB]
Get:34 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal/main amd64 libalgorithm-merge-perl all 0.08-3 [12.0 kB]
Get:35 https://mirrors.tuna.tsinghua.edu.cn/ubuntu focal/main amd64 manpages-dev all 5.05-1 [2,266 kB]
Fetched 65.6 MB in 8s (8,219 kB/s)
Extracting templates from packages: 100%
Preconfiguring packages ...
(Reading database ... 147792 files and directories currently installed.)
Preparing to unpack .../libc6-dbg_2.31-0ubuntu9.17_amd64.deb ...
Unpacking libc6-dbg:amd64 (2.31-0ubuntu9.17) over (2.31-0ubuntu9.9) ...
Preparing to unpack .../libatomic1_10.5.0-1ubuntu1~20.04_amd64.deb ...
Unpacking libatomic1:amd64 (10.5.0-1ubuntu1~20.04) over (10.3.0-1ubuntu1~20.04) ...
Preparing to unpack .../gcc-10-base_10.5.0-1ubuntu1~20.04_amd64.deb ...
Unpacking gcc-10-base:amd64 (10.5.0-1ubuntu1~20.04) over (10.3.0-1ubuntu1~20.04) ...
Setting up gcc-10-base:amd64 (10.5.0-1ubuntu1~20.04) ...
(Reading database ... 147790 files and directories currently installed.)
Preparing to unpack .../libstdc++6_10.5.0-1ubuntu1~20.04_amd64.deb ...
Unpacking libstdc++6:amd64 (10.5.0-1ubuntu1~20.04) over (10.3.0-1ubuntu1~20.04) ...
Setting up libstdc++6:amd64 (10.5.0-1ubuntu1~20.04) ...
(Reading database ... 147790 files and directories currently installed.)
Preparing to unpack .../libgomp1_10.5.0-1ubuntu1~20.04_amd64.deb ...
Unpacking libgomp1:amd64 (10.5.0-1ubuntu1~20.04) over (10.3.0-1ubuntu1~20.04) ...
Preparing to unpack .../libcc1-0_10.5.0-1ubuntu1~20.04_amd64.deb ...
Unpacking libcc1-0:amd64 (10.5.0-1ubuntu1~20.04) over (10.3.0-1ubuntu1~20.04) ...
Preparing to unpack .../libgcc-s1_10.5.0-1ubuntu1~20.04_amd64.deb ...
Unpacking libgcc-s1:amd64 (10.5.0-1ubuntu1~20.04) over (10.3.0-1ubuntu1~20.04) ...
Setting up libgcc-s1:amd64 (10.5.0-1ubuntu1~20.04) ...
(Reading database ... 147790 files and directories currently installed.)
Preparing to unpack .../libc6_2.31-0ubuntu9.17_amd64.deb ...
Unpacking libc6:amd64 (2.31-0ubuntu9.17) over (2.31-0ubuntu9.9) ...
Setting up libc6:amd64 (2.31-0ubuntu9.17) ...
Selecting previously unselected package libc-dev-bin.
(Reading database ... 147790 files and directories currently installed.)
Preparing to unpack .../00-libc-dev-bin_2.31-0ubuntu9.17_amd64.deb ...
Unpacking libc-dev-bin (2.31-0ubuntu9.17) ...
Selecting previously unselected package linux-libc-dev:amd64.
Preparing to unpack .../01-linux-libc-dev_5.4.0-205.225_amd64.deb ...
Unpacking linux-libc-dev:amd64 (5.4.0-205.225) ...
Selecting previously unselected package libcrypt-dev:amd64.
Preparing to unpack .../02-libcrypt-dev_1%3a4.4.10-10ubuntu4_amd64.deb ...
Unpacking libcrypt-dev:amd64 (1:4.4.10-10ubuntu4) ...
Selecting previously unselected package libc6-dev:amd64.
Preparing to unpack .../03-libc6-dev_2.31-0ubuntu9.17_amd64.deb ...
Unpacking libc6-dev:amd64 (2.31-0ubuntu9.17) ...
Preparing to unpack .../04-cpp-9_9.4.0-1ubuntu1~20.04.2_amd64.deb ...
Unpacking cpp-9 (9.4.0-1ubuntu1~20.04.2) over (9.4.0-1ubuntu1~20.04.1) ...
Preparing to unpack .../05-gcc-9-base_9.4.0-1ubuntu1~20.04.2_amd64.deb ...
Unpacking gcc-9-base:amd64 (9.4.0-1ubuntu1~20.04.2) over (9.4.0-1ubuntu1~20.04.1) ...
Selecting previously unselected package libitm1:amd64.
Preparing to unpack .../06-libitm1_10.5.0-1ubuntu1~20.04_amd64.deb ...
Unpacking libitm1:amd64 (10.5.0-1ubuntu1~20.04) ...
Selecting previously unselected package libasan5:amd64.
Preparing to unpack .../07-libasan5_9.4.0-1ubuntu1~20.04.2_amd64.deb ...
Unpacking libasan5:amd64 (9.4.0-1ubuntu1~20.04.2) ...
Selecting previously unselected package liblsan0:amd64.
Preparing to unpack .../08-liblsan0_10.5.0-1ubuntu1~20.04_amd64.deb ...
Unpacking liblsan0:amd64 (10.5.0-1ubuntu1~20.04) ...
Selecting previously unselected package libtsan0:amd64.
Preparing to unpack .../09-libtsan0_10.5.0-1ubuntu1~20.04_amd64.deb ...
Unpacking libtsan0:amd64 (10.5.0-1ubuntu1~20.04) ...
Selecting previously unselected package libubsan1:amd64.
Preparing to unpack .../10-libubsan1_10.5.0-1ubuntu1~20.04_amd64.deb ...
Unpacking libubsan1:amd64 (10.5.0-1ubuntu1~20.04) ...
Selecting previously unselected package libquadmath0:amd64.
Preparing to unpack .../11-libquadmath0_10.5.0-1ubuntu1~20.04_amd64.deb ...
Unpacking libquadmath0:amd64 (10.5.0-1ubuntu1~20.04) ...
Selecting previously unselected package libgcc-9-dev:amd64.
Preparing to unpack .../12-libgcc-9-dev_9.4.0-1ubuntu1~20.04.2_amd64.deb ...
Unpacking libgcc-9-dev:amd64 (9.4.0-1ubuntu1~20.04.2) ...
Selecting previously unselected package gcc-9.
Preparing to unpack .../13-gcc-9_9.4.0-1ubuntu1~20.04.2_amd64.deb ...
Unpacking gcc-9 (9.4.0-1ubuntu1~20.04.2) ...
Selecting previously unselected package gcc.
Preparing to unpack .../14-gcc_4%3a9.3.0-1ubuntu2_amd64.deb ...
Unpacking gcc (4:9.3.0-1ubuntu2) ...
Selecting previously unselected package libstdc++-9-dev:amd64.
Preparing to unpack .../15-libstdc++-9-dev_9.4.0-1ubuntu1~20.04.2_amd64.deb ...
Unpacking libstdc++-9-dev:amd64 (9.4.0-1ubuntu1~20.04.2) ...
Selecting previously unselected package g++-9.
Preparing to unpack .../16-g++-9_9.4.0-1ubuntu1~20.04.2_amd64.deb ...
Unpacking g++-9 (9.4.0-1ubuntu1~20.04.2) ...
Selecting previously unselected package g++.
Preparing to unpack .../17-g++_4%3a9.3.0-1ubuntu2_amd64.deb ...
Unpacking g++ (4:9.3.0-1ubuntu2) ...
Selecting previously unselected package make.
Preparing to unpack .../18-make_4.2.1-1.2_amd64.deb ...
Unpacking make (4.2.1-1.2) ...
Selecting previously unselected package dpkg-dev.
Preparing to unpack .../19-dpkg-dev_1.19.7ubuntu3.2_all.deb ...
Unpacking dpkg-dev (1.19.7ubuntu3.2) ...
Selecting previously unselected package build-essential.
Preparing to unpack .../20-build-essential_12.8ubuntu1.1_amd64.deb ...
Unpacking build-essential (12.8ubuntu1.1) ...
Selecting previously unselected package libfakeroot:amd64.
Preparing to unpack .../21-libfakeroot_1.24-1_amd64.deb ...
Unpacking libfakeroot:amd64 (1.24-1) ...
Selecting previously unselected package fakeroot.
Preparing to unpack .../22-fakeroot_1.24-1_amd64.deb ...
Unpacking fakeroot (1.24-1) ...
Selecting previously unselected package libalgorithm-diff-perl.
Preparing to unpack .../23-libalgorithm-diff-perl_1.19.03-2_all.deb ...
Unpacking libalgorithm-diff-perl (1.19.03-2) ...
Selecting previously unselected package libalgorithm-diff-xs-perl.
Preparing to unpack .../24-libalgorithm-diff-xs-perl_0.04-6_amd64.deb ...
Unpacking libalgorithm-diff-xs-perl (0.04-6) ...
Selecting previously unselected package libalgorithm-merge-perl.
Preparing to unpack .../25-libalgorithm-merge-perl_0.08-3_all.deb ...
Unpacking libalgorithm-merge-perl (0.08-3) ...
Selecting previously unselected package manpages-dev.
Preparing to unpack .../26-manpages-dev_5.05-1_all.deb ...
Unpacking manpages-dev (5.05-1) ...
Setting up manpages-dev (5.05-1) ...
Setting up libalgorithm-diff-perl (1.19.03-2) ...
Setting up linux-libc-dev:amd64 (5.4.0-205.225) ...
Setting up libgomp1:amd64 (10.5.0-1ubuntu1~20.04) ...
Setting up libfakeroot:amd64 (1.24-1) ...
Setting up libc6-dbg:amd64 (2.31-0ubuntu9.17) ...
Setting up fakeroot (1.24-1) ...
update-alternatives: using /usr/bin/fakeroot-sysv to provide /usr/bin/fakeroot (fakeroot) in auto mode
Setting up make (4.2.1-1.2) ...
Setting up libquadmath0:amd64 (10.5.0-1ubuntu1~20.04) ...
Setting up libatomic1:amd64 (10.5.0-1ubuntu1~20.04) ...
Setting up libubsan1:amd64 (10.5.0-1ubuntu1~20.04) ...
Setting up libcrypt-dev:amd64 (1:4.4.10-10ubuntu4) ...
Setting up libc-dev-bin (2.31-0ubuntu9.17) ...
Setting up libalgorithm-diff-xs-perl (0.04-6) ...
Setting up libcc1-0:amd64 (10.5.0-1ubuntu1~20.04) ...
Setting up liblsan0:amd64 (10.5.0-1ubuntu1~20.04) ...
Setting up libitm1:amd64 (10.5.0-1ubuntu1~20.04) ...
Setting up gcc-9-base:amd64 (9.4.0-1ubuntu1~20.04.2) ...
Setting up libalgorithm-merge-perl (0.08-3) ...
Setting up libtsan0:amd64 (10.5.0-1ubuntu1~20.04) ...
Setting up dpkg-dev (1.19.7ubuntu3.2) ...
Setting up libasan5:amd64 (9.4.0-1ubuntu1~20.04.2) ...
Setting up cpp-9 (9.4.0-1ubuntu1~20.04.2) ...
Setting up libc6-dev:amd64 (2.31-0ubuntu9.17) ...
Setting up libgcc-9-dev:amd64 (9.4.0-1ubuntu1~20.04.2) ...
Setting up gcc-9 (9.4.0-1ubuntu1~20.04.2) ...
Setting up libstdc++-9-dev:amd64 (9.4.0-1ubuntu1~20.04.2) ...
Setting up gcc (4:9.3.0-1ubuntu2) ...
Setting up g++-9 (9.4.0-1ubuntu1~20.04.2) ...
Setting up g++ (4:9.3.0-1ubuntu2) ...
update-alternatives: using /usr/bin/g++ to provide /usr/bin/c++ (c++) in auto mode
Setting up build-essential (12.8ubuntu1.1) ...
Processing triggers for man-db (2.9.1-1) ...
Processing triggers for libc-bin (2.31-0ubuntu9.9) ...带有pip的python环境
直接安装pip
sudo apt install python3-pip
anaconda
wget https://repo.anaconda.com/archive/Anaconda3-2024.02-1-Linux-x86_64.sh，从网页https://www.anaconda.com/download/success下载速度可能更快bash Anaconda3-2024.02-1-Linux-x86_64.shconda create -n tritonenv python=3.8conda activate tritonenv
安装 triton 环境
pip install triton
pip install tritonhttps://triton-lang.org/main/getting-started/installation.html
(base) cifi@cifi-666666:~/Downloads$ conda activate tritonenv
(tritonenv) cifi@cifi-666666:~/Downloads$ pip install triton
Collecting tritonDownloading triton-3.1.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (1.3 kB)
Collecting filelock (from triton)Downloading filelock-3.16.1-py3-none-any.whl.metadata (2.9 kB)
Downloading triton-3.1.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (209.4 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 209.4/209.4 MB 8.1 kB/s eta 0:00:00
Downloading filelock-3.16.1-py3-none-any.whl (16 kB)
Installing collected packages: filelock, triton
Successfully installed filelock-3.16.1 triton-3.1.0
pip install torch
pip install torch
(tritonenv) cifi@cifi-666666:~/Downloads$ pip install torch
Collecting torchDownloading torch-2.4.1-cp38-cp38-manylinux1_x86_64.whl.metadata (26 kB)
Requirement already satisfied: filelock in /home/cifi/anaconda3/envs/tritonenv/lib/python3.8/site-packages (from torch) (3.16.1)
Collecting typing-extensions>=4.8.0 (from torch)Downloading typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
Collecting sympy (from torch)Downloading sympy-1.13.3-py3-none-any.whl.metadata (12 kB)
Collecting networkx (from torch)Downloading networkx-3.1-py3-none-any.whl.metadata (5.3 kB)
Collecting jinja2 (from torch)Downloading jinja2-3.1.5-py3-none-any.whl.metadata (2.6 kB)
Collecting fsspec (from torch)Downloading fsspec-2025.2.0-py3-none-any.whl.metadata (11 kB)
Collecting nvidia-cuda-nvrtc-cu12==12.1.105 (from torch)Downloading nvidia_cuda_nvrtc_cu12-12.1.105-py3-none-manylinux1_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cuda-runtime-cu12==12.1.105 (from torch)Downloading nvidia_cuda_runtime_cu12-12.1.105-py3-none-manylinux1_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cuda-cupti-cu12==12.1.105 (from torch)Downloading nvidia_cuda_cupti_cu12-12.1.105-py3-none-manylinux1_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-cudnn-cu12==9.1.0.70 (from torch)Downloading nvidia_cudnn_cu12-9.1.0.70-py3-none-manylinux2014_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-cublas-cu12==12.1.3.1 (from torch)Downloading nvidia_cublas_cu12-12.1.3.1-py3-none-manylinux1_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cufft-cu12==11.0.2.54 (from torch)Downloading nvidia_cufft_cu12-11.0.2.54-py3-none-manylinux1_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-curand-cu12==10.3.2.106 (from torch)Downloading nvidia_curand_cu12-10.3.2.106-py3-none-manylinux1_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cusolver-cu12==11.4.5.107 (from torch)Downloading nvidia_cusolver_cu12-11.4.5.107-py3-none-manylinux1_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-cusparse-cu12==12.1.0.106 (from torch)Downloading nvidia_cusparse_cu12-12.1.0.106-py3-none-manylinux1_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-nccl-cu12==2.20.5 (from torch)Downloading nvidia_nccl_cu12-2.20.5-py3-none-manylinux2014_x86_64.whl.metadata (1.8 kB)
Collecting nvidia-nvtx-cu12==12.1.105 (from torch)Downloading nvidia_nvtx_cu12-12.1.105-py3-none-manylinux1_x86_64.whl.metadata (1.7 kB)
Collecting triton==3.0.0 (from torch)Downloading triton-3.0.0-1-cp38-cp38-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (1.3 kB)
Collecting nvidia-nvjitlink-cu12 (from nvidia-cusolver-cu12==11.4.5.107->torch)Downloading nvidia_nvjitlink_cu12-12.8.61-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl.metadata (1.7 kB)
Collecting MarkupSafe>=2.0 (from jinja2->torch)Downloading MarkupSafe-2.1.5-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (3.0 kB)
Collecting mpmath<1.4,>=1.1.0 (from sympy->torch)Downloading mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
Downloading torch-2.4.1-cp38-cp38-manylinux1_x86_64.whl (797.1 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 797.1/797.1 MB 566.0 kB/s eta 0:00:00
Downloading nvidia_cublas_cu12-12.1.3.1-py3-none-manylinux1_x86_64.whl (410.6 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 410.6/410.6 MB 958.2 kB/s eta 0:00:00
Downloading nvidia_cuda_cupti_cu12-12.1.105-py3-none-manylinux1_x86_64.whl (14.1 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 14.1/14.1 MB 1.1 MB/s eta 0:00:00
Downloading nvidia_cuda_nvrtc_cu12-12.1.105-py3-none-manylinux1_x86_64.whl (23.7 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 23.7/23.7 MB 1.5 MB/s eta 0:00:00
Downloading nvidia_cuda_runtime_cu12-12.1.105-py3-none-manylinux1_x86_64.whl (823 kB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 823.6/823.6 kB 1.8 MB/s eta 0:00:00
Downloading nvidia_cudnn_cu12-9.1.0.70-py3-none-manylinux2014_x86_64.whl (664.8 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╸ 656.9/664.8 MB 867.7 kB/s eta 0:00:10Downloading nvidia_cudnn_cu12-9.1.0.70-py3-none-manylinux2014_x86_64.whl (664.8 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 664.8/664.8 MB 2.4 MB/s eta 0:00:00
Downloading nvidia_cufft_cu12-11.0.2.54-py3-none-manylinux1_x86_64.whl (121.6 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 121.6/121.6 MB 2.8 MB/s eta 0:00:00
Downloading nvidia_curand_cu12-10.3.2.106-py3-none-manylinux1_x86_64.whl (56.5 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 56.5/56.5 MB 2.2 MB/s eta 0:00:00
Downloading nvidia_cusolver_cu12-11.4.5.107-py3-none-manylinux1_x86_64.whl (124.2 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 124.2/124.2 MB 2.6 MB/s eta 0:00:00
Downloading nvidia_cusparse_cu12-12.1.0.106-py3-none-manylinux1_x86_64.whl (196.0 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 196.0/196.0 MB 3.0 MB/s eta 0:00:00
Downloading nvidia_nccl_cu12-2.20.5-py3-none-manylinux2014_x86_64.whl (176.2 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 176.2/176.2 MB 3.0 MB/s eta 0:00:00
Downloading nvidia_nvtx_cu12-12.1.105-py3-none-manylinux1_x86_64.whl (99 kB)
Downloading triton-3.0.0-1-cp38-cp38-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (209.4 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 209.4/209.4 MB 3.1 MB/s eta 0:00:00
Downloading typing_extensions-4.12.2-py3-none-any.whl (37 kB)
Downloading fsspec-2025.2.0-py3-none-any.whl (184 kB)
Downloading jinja2-3.1.5-py3-none-any.whl (134 kB)
Downloading networkx-3.1-py3-none-any.whl (2.1 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 2.1/2.1 MB 3.5 MB/s eta 0:00:00
Downloading sympy-1.13.3-py3-none-any.whl (6.2 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 3.2 MB/s eta 0:00:00
Downloading MarkupSafe-2.1.5-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (26 kB)
Downloading mpmath-1.3.0-py3-none-any.whl (536 kB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 536.2/536.2 kB 4.8 MB/s eta 0:00:00
Downloading nvidia_nvjitlink_cu12-12.8.61-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl (39.2 MB)━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 39.2/39.2 MB 3.0 MB/s eta 0:00:00
Installing collected packages: mpmath, typing-extensions, triton, sympy, nvidia-nvtx-cu12, nvidia-nvjitlink-cu12, nvidia-nccl-cu12, nvidia-curand-cu12, nvidia-cufft-cu12, nvidia-cuda-runtime-cu12, nvidia-cuda-nvrtc-cu12, nvidia-cuda-cupti-cu12, nvidia-cublas-cu12, networkx, MarkupSafe, fsspec, nvidia-cusparse-cu12, nvidia-cudnn-cu12, jinja2, nvidia-cusolver-cu12, torchAttempting uninstall: tritonFound existing installation: triton 3.1.0Uninstalling triton-3.1.0:Successfully uninstalled triton-3.1.0
Successfully installed MarkupSafe-2.1.5 fsspec-2025.2.0 jinja2-3.1.5 mpmath-1.3.0 networkx-3.1 nvidia-cublas-cu12-12.1.3.1 nvidia-cuda-cupti-cu12-12.1.105 nvidia-cuda-nvrtc-cu12-12.1.105 nvidia-cuda-runtime-cu12-12.1.105 nvidia-cudnn-cu12-9.1.0.70 nvidia-cufft-cu12-11.0.2.54 nvidia-curand-cu12-10.3.2.106 nvidia-cusolver-cu12-11.4.5.107 nvidia-cusparse-cu12-12.1.0.106 nvidia-nccl-cu12-2.20.5 nvidia-nvjitlink-cu12-12.8.61 nvidia-nvtx-cu12-12.1.105 sympy-1.13.3 torch-2.4.1 triton-3.0.0 typing-extensions-4.12.2运行test示例
vector-add.py
# https://github.com/triton-lang/triton/blob/main/python/tutorials/01-vector-add.py
import torch
import triton
import triton.language as tl@triton.jit
def add_kernel(x_ptr, # *Pointer* to first input vector. 指向第一个输入向量的指针。y_ptr, # *Pointer* to second input vector. 指向第二个输入向量的指针。output_ptr, # *Pointer* to output vector. 指向输出向量的指针。n_elements, # Size of the vector. 向量的大小。BLOCK_SIZE: tl.constexpr, # Number of elements each program should process. 每个程序应处理的元素数量。# NOTE: `constexpr` so it can be used as a shape value. 注意：`constexpr` 因此它可以用作形状值。):# There are multiple 'programs' processing different data. We identify which program# 有多个“程序”处理不同的数据。需要确定是哪一个程序：pid = tl.program_id(axis=0) # We use a 1D launch grid so axis is 0. 使用 1D 启动网格，因此轴为 0。# This program will process inputs that are offset from the initial data.# 该程序将处理相对初始数据偏移的输入。# For instance, if you had a vector of length 256 and block_size of 64, the programs would each access the elements [0:64, 64:128, 128:192, 192:256].# 例如，如果有一个长度为 256, 块大小为 64 的向量，程序将各自访问 [0:64, 64:128, 128:192, 192:256] 的元素。# Note that offsets is a list of pointers:# 注意 offsets 是指针列表：block_start = pid * BLOCK_SIZEoffsets = block_start + tl.arange(0, BLOCK_SIZE)# Create a mask to guard memory operations against out-of-bounds accesses.# 创建掩码以防止内存操作超出边界访问。mask = offsets < n_elements# Load x and y from DRAM, masking out any extra elements in case the input is not a multiple of the block size.# 从 DRAM 加载 x 和 y，如果输入不是块大小的整数倍，则屏蔽掉任何多余的元素。x = tl.load(x_ptr + offsets, mask=mask)y = tl.load(y_ptr + offsets, mask=mask)output = x + y# Write x + y back to DRAM.# 将 x + y 写回 DRAM。tl.store(output_ptr + offsets, output, mask=mask)def add(x: torch.Tensor, y: torch.Tensor):# We need to preallocate the output.# 需要预分配输出。output = torch.empty_like(x)assert x.is_cuda and y.is_cuda and output.is_cudan_elements = output.numel()# The SPMD launch grid denotes the number of kernel instances that run in parallel.# SPMD 启动网格表示并行运行的内核实例的数量。# It is analogous to CUDA launch grids. It can be either Tuple[int], or Callable(metaparameters) -> Tuple[int].# 它类似于 CUDA 启动网格。它可以是 Tuple[int]，也可以是 Callable(metaparameters) -> Tuple[int]。# In this case, we use a 1D grid where the size is the number of blocks:# 在这种情况下，使用 1D 网格，其中大小是块的数量：grid = lambda meta: (triton.cdiv(n_elements, meta['BLOCK_SIZE']), )# NOTE:# 注意：# - Each torch.tensor object is implicitly converted into a pointer to its first element.# - 每个 torch.tensor 对象都会隐式转换为其第一个元素的指针。# - `triton.jit`'ed functions can be indexed with a launch grid to obtain a callable GPU kernel.# - `triton.jit` 函数可以通过启动网格索引来获得可调用的 GPU 内核。# - Don't forget to pass meta-parameters as keywords arguments.# - 不要忘记以关键字参数传递元参数。add_kernel[grid](x, y, output, n_elements, BLOCK_SIZE=1024)# We return a handle to z but, since `torch.cuda.synchronize()` hasn't been called, the kernel is still running asynchronously at this point.# 返回 z 的句柄，但由于 `torch.cuda.synchronize()` 尚未被调用，此时内核仍在异步运行。return outputtorch.manual_seed(0)
size = 98432
x = torch.rand(size, device='cuda')
y = torch.rand(size, device='cuda')
output_torch = x + y
output_triton = add(x, y)
print("Torch计算结果:")
print(output_torch)
print("Triton计算结果:")
print(output_triton)
print(f'Torch 和 Triton 之间的最大差异是'f'{torch.max(torch.abs(output_torch - output_triton))}')
launch.json
{// Use IntelliSense to learn about possible attributes.// Hover to view descriptions of existing attributes.// For more information, visit: https://go.microsoft.com/fwlink/?linkid=830387"version": "0.2.0","configurations": [{"name":"Python Debugger: Current File","type":"debugpy","request":"launch","program":"${file}","console":"integratedTerminal"},{"name": "Python Debugger: Current File","type": "debugpy","request": "launch","program": "${file}","python": "/home/cifi/anaconda3/envs/tritonenv/bin/python3","console": "integratedTerminal"}]
}
相关文章：
高性能 ：OpenAI Triton Open-source GPU programming Language LINUX 环境配置
目录 配置triton环境cudabuild-essential带有pip的python环境直接安装pipanaconda 安装 triton 环境pip install tritonpip install torch 运行test示例vector-add.pylaunch.json 配置triton环境 cuda wget http://developer.download.nvidia.com/compute/cuda/11.0.2/local_…...
TCP 端口号为何位于首部前四个字节？协议设计的智慧与启示
知乎的一个问题很有意思：“为什么在TCP首部中要把TCP的端口号放入最开始的四个字节？” 这种问题很适合我这种搞历史的人，大年初一我给出了一个简短的解释，但仔细探究这个问题，我们将会获得 TCP/IP 被定义的过程。 文…...
HTML之JavaScript函数声明
HTML之JavaScript函数声明 1. function 函数名(){}2. var 函数名 function(){}<!DOCTYPE html> <html lang"en"><head><meta charset"UTF-8"><meta name"viewport" content"widthdevice-width, initial-scale1…...
R 数组：高效数据处理的基础
R 数组：高效数据处理的基础 引言 在数据科学和统计分析领域，R 语言以其强大的数据处理和分析能力而备受推崇。R 数组是 R 语言中用于存储和操作数据的基本数据结构。本文将详细介绍 R 数组的创建、操作和优化，帮助读者掌握 R 数组的使用技巧…...
git服务器搭建，gitea服务搭建，使用systemclt管理服务
文章目录 页面展示使用二进制文件安装git服务下载选择架构使用wget下载安装 验证 GPG 签名服务器设置准备环境创建systemctl文件 备份与恢复备份命令 (dump)恢复命令 (restore) 页面展示 使用二进制文件安装git服务 所有打包的二进制程序均包含 SQLite，MySQL 和 Po…...
Pdf手册阅读（1）--数字签名篇
原文阅读摘要 PDF支持的数字签名， 不仅仅是公私钥签名，还可以是指纹、手写、虹膜等生物识别签名。PDF签名的计算方式，可以基于字节范围进行计算，也可以基于Pdf 对象（pdf object）进行计算。 PDF文件可能包…...
嵌入式WebRTC压缩至670K，目标将so动态库压缩至500K，.a静态库还可以更小
最近把EasyRTC的效果发布出去给各大IPC厂商体验了一下，直接就用EasyRTC与各个厂商的负责人进行的通话，在通话中，用户就反馈效果确实不错！ 这两天有用户要在海思hi3516cv610上使用EasyRTC，工具链是：gcc-2024…...
百度高德地图坐标转换
百度地图和高德地图的侧重点不太一样。同样一个地名，在百度地图网站上搜索到的地点可能是商业网点，在高德地图网站上搜索到的地点可能是自然行政地点。 高德地图api 在高德地图中，搜索地名，如“乱石头川”，该地名会出…...
ES 索引结构
ES 既不像 MySQL 这样有严格的 Schema，也不像 MongoDB 那样完全无 Schema，而是介于两者之间。 1️⃣ ES 的 Schema 模式 ES 默认是 Schema-less（无模式） 的，允许动态添加字段。 但 ES 也支持 Schema（映射 …...
HPM_SDK应用本地化——基于6750evkmini
文章目录 前言一、准备工作1、下载官方的SDK2、解压SDK 二、实操1、新建目标工程文件夹2、回到SDK中将相关文件复制1、Borad文件夹2、hello_world文件夹 三、实验现象总结 前言 为什么要对sdk进行应用本地化？在嵌入式开发中我们一般将官方提供的SDK作为参考&#x…...
【deepseek-r1本地部署】
首先需要安装ollama,之前已经安装过了，这里不展示细节 在cmd中输入官网安装命令：ollama run deepseek-r1:32b，开始下载 出现success后，下载完成 接下来就可以使用了，不过是用cmd来运行使用 可以安装UI可视化界面&a…...
查询语句来提取 detail 字段中包含 xxx 的 URL 里的 commodity/ 后面的数字串
您可以使用以下 SQL 查询语句来提取 detail 字段中包含 oss.kxlist.com 的 URL 里的 commodity/ 后面的数字串： <p><img style"max-width:100%;" src"https://oss.kxlist.com//8a989a0c55e4a7900155e7fd7971000b/commodity/20170925/20170…...
堆排序
目录 堆排序（不稳定）： 代码实现： 思路分析： 总结： 堆排序（不稳定）： 如果想要一段数据从小到大进行排序，则要先建立大根堆，因为这样每次堆顶上都能…...
【MySQL】我在广州学Mysql 系列—— 数据备份与还原
ℹ️大家好，我是练小杰，今天周一，过两天就是元宵节了，今年元宵节各位又要怎么过呢！！ 本文主要对Mysql数据库中的数据备份与还原内容进行讨论！！ 回顾：👉【MySQ…...
【LeetCode Hot100 双指针】移动零、盛最多水的容器、三数之和、接雨水
双指针 1. 移动零题目描述解题思路关键思路：步骤：时间复杂度：空间复杂度： 代码实现 2. 盛最多水的容器题目解析解题思路代码实现 3. 三数之和问题描述：解题思路：算法步骤：代码实现： …...
HTML应用指南：利用POST请求获取接入比亚迪业态的充电桩位置信息
在新能源汽车快速发展的今天，充电桩的分布和可用性成为了影响用户体验的关键因素之一。比亚迪作为全球领先的新能源汽车制造商，不仅在车辆制造方面取得了卓越成就，也在充电基础设施建设上投入了大量资源。为了帮助用户更方便地找到比亚迪充电桩的位置，本篇文章，我们将探究…...
 AOSP12下载编译)
Android车机DIY开发之软件篇(十二) AOSP12下载编译
Android车机DIY开发之软件篇(十二) AOSP12下载编译 sudo apt-get update sudo apt-get install git-core gnupg flex bison gperf build-essential zip curl zlib1g-dev gcc-multilib gmultilib libc6-dev-i386 lib32ncurses5-dev libx11-dev lib32z-dev ccache libgl1-mesa-…...
Jenkins+gitee 搭建自动化部署
Jenkinsgitee 搭建自动化部署 环境说明： 软件版本备注CentOS8.5.2111JDK1.8.0_211Maven3.8.8git2.27.0Jenkins2.319最好选稳定版本，不然安装插件有点麻烦 一、安装Jenkins程序 1、到官网下载相应的版本war或者直接使用yum安装 Jenkins官网下载 直接…...
【文本处理】如何在批量WORD和txt文本提取手机号码，固话号码，提取邮箱，删除中文，删除英文，提取车牌号等等一些文本提取固定格式的操作，基于WPF的解决方案
企业的应用场景 数据清洗：在进行数据导入或分析之前，往往需要对大量文本数据进行预处理，比如去除文本中的无关字符（中文、英文），只保留需要的联系信息（手机号码、固话号码、邮箱）。…...
Linux系统引导与服务管理
目录 一、Linux引导过程 1、引导过程概述 1.1、BIOS开机自检 1.2、MBR读取 1.3、加载引导加载程序（GRUB） 1.4、内核加载 1.5、初始化进程（init） 二、服务 2.1、服务类型 2.2、服务管理工具 三、运行级别 四、systemd …...
rknn优化教程（二）
文章目录 1. 前述2. 三方库的封装2.1 xrepo中的库2.2 xrepo之外的库2.2.1 opencv2.2.2 rknnrt2.2.3 spdlog 3. rknn_engine库 1. 前述 OK，开始写第二篇的内容了。这篇博客主要能写一下： 如何给一些三方库按照xmake方式进行封装，供调用如何按…...
边缘计算医疗风险自查APP开发方案
核心目标：在便携设备（智能手表/家用检测仪）部署轻量化疾病预测模型，实现低延迟、隐私安全的实时健康风险评估。 一、技术架构设计 #mermaid-svg-iuNaeeLK2YoFKfao {font-family:"trebuchet ms",verdana,arial,sans-serif;font-size:16px;fill:#333;}#mermaid-svg…...
Mybatis逆向工程，动态创建实体类、条件扩展类、Mapper接口、Mapper.xml映射文件
今天呢，博主的学习进度也是步入了Java Mybatis 框架，目前正在逐步杨帆旗航。 那么接下来就给大家出一期有关 Mybatis 逆向工程的教学，希望能对大家有所帮助，也特别欢迎大家指点不足之处，小生很乐意接受正确的建议&…...
Opencv中的addweighted函数
一.addweighted函数作用 addweighted（）是OpenCV库中用于图像处理的函数，主要功能是将两个输入图像（尺寸和类型相同）按照指定的权重进行加权叠加（图像融合），并添加一个标量值&#x…...
苍穹外卖--缓存菜品
1.问题说明 用户端小程序展示的菜品数据都是通过查询数据库获得，如果用户端访问量比较大，数据库访问压力随之增大 2.实现思路 通过Redis来缓存菜品数据，减少数据库查询操作。 缓存逻辑分析： ①每个分类下的菜品保持一份缓存数据…...
聊一聊接口测试的意义有哪些？
目录 一、隔离性 & 早期测试 二、保障系统集成质量 三、验证业务逻辑的核心层 四、提升测试效率与覆盖度 五、系统稳定性的守护者 六、驱动团队协作与契约管理 七、性能与扩展性的前置评估 八、持续交付的核心支撑 接口测试的意义可以从四个维度展开，首…...
Cilium动手实验室: 精通之旅---13.Cilium LoadBalancer IPAM and L2 Service Announcement
Cilium动手实验室: 精通之旅---13.Cilium LoadBalancer IPAM and L2 Service Announcement 1. LAB环境2. L2公告策略2.1 部署Death Star2.2 访问服务2.3 部署L2公告策略2.4 服务宣告 3. 可视化 ARP 流量3.1 部署新服务3.2 准备可视化3.3 再次请求 4. 自动IPAM4.1 IPAM Pool4.2 …...
自然语言处理——文本分类
文本分类 传统机器学习方法文本表示向量空间模型 特征选择文档频率互信息信息增益（IG） 分类器设计贝叶斯理论：线性判别函数 文本分类性能评估P-R曲线ROC曲线 将文本文档或句子分类为预定义的类或类别， 有单标签多类别文本分类和多…...
【1】跨越技术栈鸿沟：字节跳动开源TRAE AI编程IDE的实战体验
2024年初，人工智能编程工具领域发生了一次静默的变革。当字节跳动宣布退出其TRAE项目（一款融合大型语言模型能力的云端AI编程IDE）时，技术社区曾短暂叹息。然而这一退场并非终点——通过开源社区的接力，TRAE在WayToAGI等…...
VSCode 使用CMake 构建 Qt 5 窗口程序
首先，目录结构如下图： 运行效果： cmake -B build cmake --build build 运行： windeployqt.exe F:\testQt5\build\Debug\app.exe main.cpp #include "mainwindow.h"#include <QAppli...