求助amdgpu-pro显卡驱动怎么安装

楼主是否指linux平台，可以访问amd最新的下载页面，里面有指导安装方法。

http://support.amd.com/en-us/kb-articles/Pages/AMDGPU-PRO-Driver-for-Linux-Release-Notes.aspx

比如ubuntu16.04系统下的步骤原文：

Article Number: GPU-635

ContentMain1

There are four simple steps involved in the installation of the AMDGPU-Pro Driver: Download, Extract, Install and Configure. The instructions to perform the installation are intended for an Ubuntu installation of 16.04.2, and should take less than 10 minutes to complete. Before installing the driver, a quick note on how to check if your system already has AMDGPU-PRO installed. In addition, the recommended best practice is to bring the system up-to-date before starting the driver installation, with:

apt update

apt dist-upgrade

System Check

The easiest way to find out if you have AMDGPU-Pro already installed on your Ubuntu System is to query the Debian package manager.

Using the following command at a terminal will provide you with the version of the AMDGPU-Pro stack on your system, or inform you that there are no packages found:

dpkg -l amdgpu-pro

Download

A direct link to download the Linux AMDGPU-PRO driver is given below and it is also available on the Driver Download Page. This file has a "tar.xz" extension which reflects a more-effective c ompression algorithm that (in most cases) creates a smaller archive than the more common gzip format.

AMD Radeon™ Software AMDGPU-PRO Beta Driver for Linux® Version 17.10 for Ubuntu 16.04.2

NOTE: This file can also be located via the Driver Download Page by locating your card and selecting the Linux Driver link.

Extract

After the archive is downloaded, extract the contents to a temporary location from which you can install it. The example below assumes you have downloaded the archive to /tmp and will extract to the same location. If your file was downloaded into the ~/Downloads/ folder by default, you can also extract and install from there, and afterwards you can remove the install files.

cd /tmp

tar -Jxvf amdgpu-pro-17.10-401251.tar.xz

Install

Once the archive is expanded on the local machine, run the included script (amdgpu-pro-install) to install the graphics stack. During the installation you will be required to provide sudo access, and also to provide two confirmations to:

Install the packages, and

Allow installation of "unverified" packages from the AMDGPU-PRO archive.

The script will use the package manager to install the components of the graphics stack, with a short delay during the DKMS (Dynamic Kernel Module Support) installation. From the directory where you extracted the archive, issue the following command:

cd amdgpu-pro-17.10-401251

./amdgpu-pro-install -y

After this you can restart your machine to launch using the new graphics stack.

Configure

Ensure that your user account is a member of the "video" group prior to using the vulkan driver. You can find which groups you are a member of with the following command:

groups

To add yourself to the video group you will need the sudo password and can use the following command:

sudo usermod -a -G video $LOGNAME

You will need to log out and in again to activate this change.

Uninstalling the AMD GPU-PRO Driver

If for any reason you wish to remove the AMDGPU-PRO graphics stack you can do this using the uninstallation script which was part of the installation and is present in your path. From the command prompt enter the following command:

amdgpu-pro-uninstall

上期给大家做了一期Vega64对2080的 游戏 性能对比，这期我将使用这款显卡在专业领域进行测试。Vega64双精度高达822 FLOPS，超过2080 230%，也超过2080Ti多达170%，规格极高的Vega64，在专业领域真的能大展身手吗？

3dsmax是一款三维动画渲染与设计工具，被广泛应用于 游戏 、动漫、影视特效制作，下面是Vega64、RTX2080、RTX2080Ti、P6000四款显卡在3dsmax中的性能对比（P6000参考数据来自官方参考数据）。（下面各图最后一个柱状图都是Vega64）

Catia这款软件你可能不知道，但是开发公司你肯定听说过：法国达索。这款软件主要应用于数控加工领域，支持设计对象的混合建模、变量和参数化的混合建模、几何和智能工程的混合建模等多种建模技术。在这款软件中，Vega64的性能超过了2080和2080Ti。

EnergyPlus是一款专业的建筑-能源模拟引擎平台，支持支持模拟建筑物采暖、制冷、照明、通风以及其他能源消耗。

maya（玛雅）是目前使用人数最多、最流行的一款三维动画软件，主要面向影视广告、 游戏 动画和电影特技，渲染出的画面十分逼真，支持英伟达毛发技术、布料物理特性、粒子特性等先进物理技术。

Autodesk Showcase是一款三维可视化软件，应用于建筑工程领域，最新版本支持硬件加速的光线追踪渲染，多名设计师可以同时在线进行开发。

Adobe Premiere Pro CS6可以开启水银加速，提高出片速度。水银加速需要配合支持CUDA的硬件，显然Vega64不支持CUDA，也就没法使用水银加速，只能使用默认的CPU渲染。我以前文章测试过6核12线程CPU和2080渲染速度对比，差距为13倍。

Adobe After Effects CC 2018支持NVIDIA® OptiX光线追踪引擎支持下的光线追踪渲染、C4D渲染和经典渲染，由于A卡无法使用OptiX光追引擎和C4D引擎进行GPU渲染，我依然无法使用A卡进行GPU渲染加速。

深度学习需要搭建tensorflow环境，大黄我现在Win10没动力换Linux（VM又不支持独显，魔改后支持性能会缩水），测试计算性能使用GPUPI/OPENCL API下进行圆周率计算和在Maltab下进行矩阵计算进行。

首先进行圆周率计算，批处理大小为100M，Reduction模型大小64（Vega没法跑512大小的模型，RTX可以），该计算主要测试显卡OpenCL环境下双精度计算性能。结果如下（时间为秒，耗时越少越好），2080Ti双精度浮点性能是Vega64的410%。

然而当我进行Matlab进行矩阵计算时，RTX2080Ti顺利完成了两个5000*5000二维矩阵的计算，而因为Matlab不支持AMD GPU计算，Vega64计算失败。

小工作室常用的OCtane渲染器基准测试，Vega64报错，无法支持。

室内设计中最主流的v-ray渲染器基准测试，Vega64无法支持。

Vega64公版这张显卡目前某东全新入手4499，在 游戏 上表现接近2070，温度和功耗稍高；而专业方面由于AMD不支持图形行业广泛采用的CUDA标准，所以不少专业设计软件、渲染器以及计算应用A卡都没法使用。可以说AMD显卡图形性能不算太弱，在全球GPU厂商中位列第二，然而在专业市场上占比太低的主因还是不符合市场规范的设计导致专业应用处处受限。

最后我提一个问题：为什么RTX公版都是双风扇了，厂商还要做这种单风扇涡轮显卡？知道的在评论区见。

---