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。

Memory Profiling: Introduction

94 点作者 SerCe超过 1 年前

4 条评论

koverstreet超过 1 年前
Coming to the kernel: <a href="https:&#x2F;&#x2F;lore.kernel.org&#x2F;linux-mm&#x2F;f4abca3f1d89d11f31b965e58a397aa6074be9d1.camel@linux.intel.com&#x2F;T&#x2F;" rel="nofollow">https:&#x2F;&#x2F;lore.kernel.org&#x2F;linux-mm&#x2F;f4abca3f1d89d11f31b965e58a3...</a>
评论 #39382414 未加载
malkia超过 1 年前
Some of our tools do so much allocation that capturing all the information (using MTuner) to disk and later loading require by itself hundreths of GB.<p>Instead I&#x27;ve added random sampling - e.g. if ptr % modulo &gt; level - output it or not.<p>Another factor that slows down is doing a callstack capture. It&#x27;s not for free at all, like on Windows it has to go through the exception handlers, etc. I think perfetto simply captures the whole stack (need to check again), and then offline decodes it - or something like this.<p>Windows ETW Tracing can also capture the stack, but I guess it&#x27;ll incur also some penalty - it can&#x27;t come for free.<p>I also wish there was some kind of standard binary format for emitting alloc&#x2F;free sequences with callstack&#x2F;etc.
评论 #39381034 未加载
评论 #39379746 未加载
kohlerm超过 1 年前
<a href="http:&#x2F;&#x2F;kohlerm.blogspot.com&#x2F;2009&#x2F;02&#x2F;how-to-really-measure-memory-usage-of.html" rel="nofollow">http:&#x2F;&#x2F;kohlerm.blogspot.com&#x2F;2009&#x2F;02&#x2F;how-to-really-measure-me...</a> do we really still not have a decent memory usage analysis tool for c&#x2F; C++ ?
评论 #39377360 未加载
评论 #39379713 未加载
评论 #39381007 未加载
irogers超过 1 年前
Somewhat related, &quot;Data-type profiling for perf&quot;: <a href="https:&#x2F;&#x2F;lwn.net&#x2F;Articles&#x2F;955709&#x2F;" rel="nofollow">https:&#x2F;&#x2F;lwn.net&#x2F;Articles&#x2F;955709&#x2F;</a>