TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Libcu++: Nvidia C++ Standard Library

226 pointsby andrew3726over 4 years ago

11 comments

fanf2over 4 years ago
“<i>Whenever a new major CUDA Compute Capability is released, the ABI is broken. A new NVIDIA C++ Standard Library ABI version is introduced and becomes the default and support for all older ABI versions is dropped.</i>”<p><a href="https:&#x2F;&#x2F;github.com&#x2F;NVIDIA&#x2F;libcudacxx&#x2F;blob&#x2F;main&#x2F;docs&#x2F;releases&#x2F;versioning.md" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;NVIDIA&#x2F;libcudacxx&#x2F;blob&#x2F;main&#x2F;docs&#x2F;releases...</a>
评论 #24526985 未加载
评论 #24529128 未加载
lionkorover 4 years ago
&gt; Promising long-term ABI stability would prevent us from fixing mistakes and providing best in class performance. So, we make no such promises.<p>Wait NVidia actually get it? Neat!
评论 #24530213 未加载
larsover 4 years ago
It really is a tiny subset of the C++ standard library, but I&#x27;m happy to see they&#x27;re continuing to expand it: <a href="https:&#x2F;&#x2F;nvidia.github.io&#x2F;libcudacxx&#x2F;api.html" rel="nofollow">https:&#x2F;&#x2F;nvidia.github.io&#x2F;libcudacxx&#x2F;api.html</a>
评论 #24527198 未加载
评论 #24526440 未加载
评论 #24530588 未加载
RcouF1uZ4gsCover 4 years ago
For everyone wondering where are all the data structures and algorithms, vector and several algorithms are implemented by Thrust. <a href="https:&#x2F;&#x2F;docs.nvidia.com&#x2F;cuda&#x2F;thrust&#x2F;index.html" rel="nofollow">https:&#x2F;&#x2F;docs.nvidia.com&#x2F;cuda&#x2F;thrust&#x2F;index.html</a><p>Seems the big addition of the Libcu++ to Thrust would be synchronization.
评论 #24530884 未加载
davvidover 4 years ago
Here&#x27;s a somewhat related talk from CppCon &#x27;19: &quot;The One-Decade Task: Putting std::atomic in CUDA&quot;<p><a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=VogqOscJYvk" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=VogqOscJYvk</a>
jlebarover 4 years ago
This is super-cool.<p>For those of us who can&#x27;t adopt it right away, note that you can compile your cuda code with `--expt-relaxed-constexpr` and call any constexpr function from device code. That includes all the constexpr functions in the standard library!<p>This gets you quite a bit, but not e.g. std::atomic, which is one of the big things in here.
BoppreHover 4 years ago
Unfortunate name, &quot;cu&quot; it&#x27;s the most well known slang for &quot;anus&quot; in Brazil (population: 200+ million). &quot;Libcu++&quot; is sure to cause snickering.
评论 #24526825 未加载
评论 #24530622 未加载
评论 #24527099 未加载
评论 #24531586 未加载
评论 #24529264 未加载
评论 #24526943 未加载
评论 #24526746 未加载
评论 #24530539 未加载
评论 #24528155 未加载
评论 #24526803 未加载
einpoklumover 4 years ago
1. How do we know what parts of the library are usable on CUDA devices, and which are only usable in host-side code?<p>2. How compatible is this with libstdc++ and&#x2F;or libcu++, when used independently?<p>I&#x27;m somewhat suspicious of the presumption of us using NVIDIA&#x27;s version of the standard library for our host-side work.<p>Finally, I&#x27;m not sure that, for device-side work, libc++ is a better base to start off of than, say, EASTL (which I used for my tuple class: <a href="https:&#x2F;&#x2F;github.com&#x2F;eyalroz&#x2F;cuda-kat&#x2F;blob&#x2F;master&#x2F;src&#x2F;kat&#x2F;tuple.hpp" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;eyalroz&#x2F;cuda-kat&#x2F;blob&#x2F;master&#x2F;src&#x2F;kat&#x2F;tupl...</a> ).<p>...<p>partial self-answer to (1.): <a href="https:&#x2F;&#x2F;nvidia.github.io&#x2F;libcudacxx&#x2F;api.html" rel="nofollow">https:&#x2F;&#x2F;nvidia.github.io&#x2F;libcudacxx&#x2F;api.html</a> apparently only a small bit of the library is actually implemented.
评论 #24530762 未加载
Mr_lavosover 4 years ago
Does this mean you can do operations on struct&#x27;s that live on the GPU hardware?
评论 #24527206 未加载
gj_78over 4 years ago
I really do not understand why a (very good) hardware provider is willing to create&#x2F;direct&#x2F;hint custom software for the users.<p>Isn&#x27;t this exactly what a GPU firmware is expected to do ? Why do they need to run software in the same memory space as my mail reader ?
评论 #24530686 未加载
评论 #24529104 未加载
评论 #24527619 未加载
scott31over 4 years ago
A pathetic attempt to lock developers into their hardware.
评论 #24526618 未加载
评论 #24530715 未加载
评论 #24526545 未加载
评论 #24527253 未加载
评论 #24527178 未加载