Add resources section to README. #1085
Conversation
| - [Getting Started](https://rapids.ai/start.html): Instructions for installing RMM. | ||
| - [RAPIDS Community](https://rapids.ai/community.html): Get help, contribute, and collaborate. | ||
| - [GitHub repository](https://github.com/rapidsai/rmm): Download the cuDF source code. | ||
| - [Issue tracker](https://github.com/rapidsai/rmm/issues): Report issues or request features. |
There was a problem hiding this comment.
| - [Issue tracker](https://github.com/rapidsai/rmm/issues): Report issues or request features. | |
| - [Issue tracker](https://github.com/rapidsai/rmm/issues): Report issues or request features. | |
| - [Stream-ordered Allocation](https://developer.nvidia.com/blog/using-cuda-stream-ordered-memory-allocator-part-1/): More information about the semantics of stream-ordered allocation. |
There was a problem hiding this comment.
@jrhemstad This blog post isn't about RMM as a library, it's a topic of background info that developers should know. It can definitely be linked in the README, but it doesn't feel appropriate for the Resources top-level section. I'd link it in the section titled "Stream-ordered Memory Allocation" below. If you're okay with that, let me know and I'll make the change. edit: went ahead and did it.
There was a problem hiding this comment.
I tackled this in 4d675f2. Feel free to suggest changes / approve if you're happy with it.
There was a problem hiding this comment.
If that's what you have in mind for "Resources" then sure.
I had thought this section was more of a "Further Reading" as just a collection of links to things anyone interested in RMM would also be interested in.
There was a problem hiding this comment.
I try to keep the "Resources" section brief since it's a top level thing -- just a set of links to crucial project information, like how to acquire, use, report issues with, and contribute to the code. "Further Reading" about theory/concepts underlying the software design definitely fits in the README, just maybe not in the core project info.
|
That reminds me there is also a blog post about RMM to link. https://developer.nvidia.com/blog/fast-flexible-allocation-for-cuda-with-rapids-memory-manager/ |
|
If there are any videos of RMM talks, those might be nice to link as well |
Co-authored-by: Jake Hemstad <[email protected]>
Resolved in 21d4644. |
|
|
||
| Achieving optimal performance in GPU-centric workflows frequently requires customizing how host and | ||
| device memory are allocated. For example, using "pinned" host memory for asynchronous | ||
| host <-> device memory transfers, or using a device memory pool sub-allocator to reduce the cost of | ||
| dynamic device memory allocation. | ||
| dynamic device memory allocation. |
There was a problem hiding this comment.
My editor trimmed trailing whitespace. Sorry for the noisy diff -- can we merge this with the whitespace trimmed, or do I need to revert?
|
@gpucibot merge |
Description
This PR adds a "Resources" section similar to the RMM README, similar to what I did in cuDF some time ago. My primary reason for adding this is to have a quick link to the RMM documentation.
Checklist