TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Show HN: TensorFlow for AWS on a Real GPU

77 点作者 alexkern超过 9 年前

2 条评论

Smerity超过 9 年前
Honestly, there are enough issues with TensorFlow right now due to CUDA 3.0 that using it with AWS is highly problematic. I appreciate the author&#x27;s attempt, but there&#x27;s no way the five lines of code he changed to allow CUDA 3.0 has fixed any of the issues found in [1], such as NaNs during training, equally slow training on a g2.2xlarge as a g2.8xlarge, etc ...<p>If you&#x27;re just interested in playing around, then your laptop will do fine - TensorFlow is happy with just about any hardware you throw at it. Hell, your modern Android phone will run it =]<p>If you&#x27;re interested in a more involved experiment, develop and debug your task locally on your laptop. By the time you&#x27;re ready for large scale training, there might be a stable and battle tested AMI such that people are no longer reporting issues in [1] about it.<p>Again, if you&#x27;re interested, follow the CUDA 3.0 issue on GitHub[1] - this is nowhere near a solved problem and will only cause headaches if you&#x27;re using it for education.<p>[1]: <a href="https:&#x2F;&#x2F;github.com&#x2F;tensorflow&#x2F;tensorflow&#x2F;issues&#x2F;25" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;tensorflow&#x2F;tensorflow&#x2F;issues&#x2F;25</a>
评论 #10556488 未加载
评论 #10556893 未加载
tacos超过 9 年前
Top link points to a non-existent AMI so this &quot;recipe&quot; -- like almost every &quot;here&#x27;s how to ...&quot; recipe on GitHub that gets posted on HN -- doesn&#x27;t work. I swear, it&#x27;s like 100% fail on these types of posts.<p>Also this &quot;real GPU&quot; is explicitly called out in the Google docs as unsupported.
评论 #10557245 未加载
评论 #10556270 未加载