blocks|key|32596|text|请在脚本的最开始的中添加以下：|type|unstyled|depth|inlineStyleRanges|offset|length|style|BOLD|entityRanges|data|32597|os.environ['TF_XLA_FLAGS']+=+'--tf_xla_enable_xla_devices'|code-block|syntax|javascript|32598|entityMap^0|9|5|0|0^^$0|@$1|2|3|4|5|6|7|M|8|@$9|N|A|O|B|C]]|D|@]|E|$]]|$1|F|3|G|5|H|7|P|8|@]|D|@]|E|$I|J]]|$1|K|3|-4|5|6|7|Q|8|@]|D|@]|E|$]]]|L|$]]

Please add the following at the very beginning of your script:
<pre><code>os.environ['TF_XLA_FLAGS'] = '--tf_xla_enable_xla_devices'
</code></pre>

blocks|key|32603|text|或者，可以在Windows设置中设置环境变量：TF_XLA_FLAGS+=+--tf_xla_enable_xla_devices。|type|unstyled|depth|inlineStyleRanges|offset|length|style|CODE|entityRanges|data|32604|根据这一讨论的说法，这有助于更有效地使用GPU，而不是简单地忽略信息。|32605|entityMap|0|LINK|mutability|MUTABLE|url|https://github.com/tensorflow/tensorflow/issues/44683^0|N|16|0|2|4|0|0^^$0|@$1|2|3|4|5|6|7|P|8|@$9|Q|A|R|B|C]]|D|@]|E|$]]|$1|F|3|G|5|6|7|S|8|@]|D|@$9|T|A|U|1|V]]|E|$]]|$1|H|3|-4|5|6|7|W|8|@]|D|@]|E|$]]]|I|$J|$5|K|L|M|E|$N|O]]]]

Alternatively, one can set the environment variable in Windows system settings: <code>TF_XLA_FLAGS = --tf_xla_enable_xla_devices</code>.
According to <a href="https://github.com/tensorflow/tensorflow/issues/44683" rel="nofollow noreferrer">this discussion</a>, this helps use the GPU much more efficiently than simply ignoring the info.

blocks|key|1905061|text|通常情况下，这个信息不应该被干扰。|type|unstyled|depth|inlineStyleRanges|entityRanges|data|1905062|请尝试|1905063|import+tensorflow+as+tf
print("GPUs:+",+len(tf.config.experimental.list_physical_devices('GPU')))|code-block|syntax|javascript|1905064|检查它们是否被检测到。|1905065|如果您需要，请查看github的发布说明：|1905066|默认情况下，XLA:CPU和XLA:GPU设备不再注册。如果您真的需要使用tf_xla_+this+=-tf_xla_enable_xla_设备，但是这个标志最终会在以后的版本中被删除。|blockquote|1905067|entityMap^0|0|0|0|0|0|0^^$0|@$1|2|3|4|5|6|7|R|8|@]|9|@]|A|$]]|$1|B|3|C|5|6|7|S|8|@]|9|@]|A|$]]|$1|D|3|E|5|F|7|T|8|@]|9|@]|A|$G|H]]|$1|I|3|J|5|6|7|U|8|@]|9|@]|A|$]]|$1|K|3|L|5|6|7|V|8|@]|9|@]|A|$]]|$1|M|3|N|5|O|7|W|8|@]|9|@]|A|$]]|$1|P|3|-4|5|6|7|X|8|@]|9|@]|A|$]]]|Q|$]]

Usually this message should not interfere.
Please try
<pre><code>import tensorflow as tf
print(&quot;GPUs: &quot;, len(tf.config.experimental.list_physical_devices('GPU')))
</code></pre>
to check that they are detected.
If you need it, please check the release notes at github:
<blockquote>
XLA:CPU and XLA:GPU devices are no longer registered by default. Use TF_XLA_FLAGS=--tf_xla_enable_xla_devices if you really need them, but this flag will eventually be removed in subsequent releases.
</blockquote>

blocks|key|1905140|text|我不确定Tensorflow的XLA:CPU和XLA:GPU是否会“在后续版本中被删除”。XLA:CPU被描述为在GPU执行用于建模Tensorflow的代码时提供对CPU的访问。当设置其环境变量时，XLA功能允许CPU在特定的关键性能阶段“帮助”，从而加快人们通常认为需要GPU的进程。这在https://www.tensorflow.org/xla中得到了解释。|type|unstyled|depth|inlineStyleRanges|entityRanges|offset|length|data|1905141|对于如何在R中设置所需的环境变量，或者更具体地说，在Rstudio中，我还没有找到真正明确的说明。在另一个博客中，一个人建议：(a)添加一个环境路径变量来打开XLA，然后是另一个专门供Tensorflow使用的变量。|1905142|设置正确的环境变量可能因操作系统、脚本程序()以及安装了哪些CUDA和CUDNN版本而有所不同。|1905143|对我来说，最重要的是查看特定环境变量的输出，该变量没有设置，因此导致XLA没有激活。但是，XLA可能需要在整个R会话或脚本中被激活。|1905144|在我的示例中，使用Ubuntu20.04使用NVIDIA+GPU和驱动程序470.57.02、CUDA-11.4和CUDNN8_8.2.4.15+(随CUDA+11.4一起运行Rstudio+2021.09.0+)，以下两个R+Setenv()命令导致我自己的"XLA错误“在我的Rstudio脚本的输出中消失：|1905145|Sys.setenv("XLA_FLAGS=--xla_gpu_cuda_data_dir"="/usr/local/cuda-11.4")

Sys.setenv("TF_XLA_FLAGS"="--tf_xla_enable_xla_devices")|code-block|syntax|javascript|1905146|请注意，两个环境变量都已设置，并且xla_gpu_cuda_data的目录在/usr/local/cuda-11.X中正确命名。|style|CODE|1905147|此外，在重新启动(或终端中的LD_LIBRARY_PATH命令)时，用户配置文件脚本(我的是.profile)中设置的环境路径和source变量也需要指示正确的cuda目录。|1905148|entityMap|0|LINK|mutability|MUTABLE|url|https://www.tensorflow.org/xla^0|42|U|0|0|0|0|0|0|0|H|H|12|K|0|1A|8|1S|6|0^^$0|@$1|2|3|4|5|6|7|14|8|@]|9|@$A|15|B|16|1|17]]|C|$]]|$1|D|3|E|5|6|7|18|8|@]|9|@]|C|$]]|$1|F|3|G|5|6|7|19|8|@]|9|@]|C|$]]|$1|H|3|I|5|6|7|1A|8|@]|9|@]|C|$]]|$1|J|3|K|5|6|7|1B|8|@]|9|@]|C|$]]|$1|L|3|M|5|N|7|1C|8|@]|9|@]|C|$O|P]]|$1|Q|3|R|5|6|7|1D|8|@$A|1E|B|1F|S|T]|$A|1G|B|1H|S|T]]|9|@]|C|$]]|$1|U|3|V|5|6|7|1I|8|@$A|1J|B|1K|S|T]|$A|1L|B|1M|S|T]]|9|@]|C|$]]|$1|W|3|-4|5|6|7|1N|8|@]|9|@]|C|$]]]|X|$Y|$5|Z|10|11|C|$12|13]]]]

I'm not sure that Tensorflow's XLA:CPU and XLA:GPU would be &quot;removed in subsequent releases.&quot; XLA:CPU is described as providing access to the CPU while the GPU is executing the code for modeling Tensorflow. The XLA functionality, when its environmental variable is set, allows the CPU to &quot;help out&quot; at specific, critical performance stages and thus to accelerate the processes one might usually think the GPU is needed for. This is explained in <a href="https://www.tensorflow.org/xla" rel="nofollow noreferrer">https://www.tensorflow.org/xla</a>.
I had not found really clear instructions on how to set the required environmental variables in R or, more specifically, Rstudio. In another blog, an individual suggested (a) adding an environmental path variable to open up XLA in general, but then another one specifically for use by Tensorflow.
Setting up the right environmental variable(s) may vary by operating system, scripting program (Python vs R vs Windows Ubuntu), and which CUDA and CUDNN versions are installed.
The most important thing for me was to look at the output for the specific environmental variable which was not set and thus caused XLA to not activate. But also, XLA may need to be activated in the entire R session or script.
In my case, using Ubuntu 20.04 to run Rstudio 2021.09.0 with an NVIDIA GPU and driver version 470.57.02, CUDA-11.4 and CUDNN8_8.2.4.15 (which goes with CUDA 11.4), the following two R Setenv() commands caused my own &quot;XLA error&quot; to go away within the output from my Rstudio script:
<pre><code>Sys.setenv(&quot;XLA_FLAGS=--xla_gpu_cuda_data_dir&quot;=&quot;/usr/local/cuda-11.4&quot;)

Sys.setenv(&quot;TF_XLA_FLAGS&quot;=&quot;--tf_xla_enable_xla_devices&quot;)
</code></pre>
Notice that both environment variables are set and the directory for <code>xla_gpu_cuda_data</code> is correctly named in <code>/usr/local/cuda-11.X</code>.
In addition, the environmental PATH and LD_LIBRARY_PATH variables as set in the user's profile script(s) (mine is <code>.profile</code>) upon reboot (or <code>source</code> command in the terminal) need to also indicate the correct cuda directories.

I run drive.py program from <a href="https://www.codeproject.com/Articles/1273179/A-Complete-guide-to-self-driving-Car?fbclid=IwAR1L7mOaPTZ7-vwWicXcAS7FLhVLcU9BUgaivhK0P9tbXAGe0zOThAuktMs" rel="noreferrer">Code Project | A Complete guide to self driving car</a>
but when i start program i have error:
Not creating XLA devices, tf_xla_enable_xla_devices not set
Does anyone know how I can fix this problem? What should I download or reinstall?
I use:
Python 3.8.7
CUDA 11.0
tensorflow 2.4.1
On <a href="http://0.0.0.0:4567/" rel="noreferrer">http://0.0.0.0:4567/</a> of course I see nothing
<a href="https://i.stack.imgur.com/3pXqE.jpg" rel="noreferrer"><img src="https://i.stack.imgur.com/3pXqE.jpg" alt="enter image description here" /></a>

tensorflow: Not creating XLA devices, tf_xla_enable_xla_devices not set

翻译质量差，导致语言生硬或混乱。

没有提供实际的解决方法或示例。

解答不清晰，无法理解或解决问题。

页面排版不美观，阅读体验差。

文章

问答

视频

学习中心

腾讯云实验室

直播

竞赛

腾讯云代码分析专区

腾讯iOA零信任安全管理系统专区

腾讯云架构师技术同盟交流圈

腾讯云数据库专区

腾讯云顾问专区

腾讯云原生专区

腾讯混元专区

腾讯云TCE专区

腾讯云Lighthouse专区

腾讯云HAI专区

腾讯云Edgeone专区

腾讯云存储专区

腾讯云智能专区

腾讯轻联专区 

腾讯云开发专区

TAPD专区

腾讯轻量云游戏服专区

腾讯云最具价值专家

腾讯云架构师技术同盟

腾讯云创作之星

腾讯云开发者先锋 

腾讯云代码助手

CODING DevOps

Cloud Studio

SDK中心

API中心

命令行工具

我从drive.py运行程序但是，当我启动程序时，我有错误：不创建XLA设备，tf_xla_enable_xla_devices未设置有人知道我怎样才能解决这个问题吗？我应该下载或重新安装什么？我用：Python 3.8.7库达11.0上的tensorflow 2.4.1当然什么也没看到​​

问tensorflow:不创建XLA设备，tf_xla_enable_xla_devices未设置
EN

回答 4

Stack Overflow用户

Stack Overflow用户

Stack Overflow用户

社区

活动

圈层

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问tensorflow:不创建XLA设备，tf_xla_enable_xla_devices未设置EN

回答 4

Stack Overflow用户

Stack Overflow用户

Stack Overflow用户

社区

活动

圈层

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问tensorflow:不创建XLA设备，tf_xla_enable_xla_devices未设置
EN