Keita Iwabuchi
Keita Iwabuchi
Hello @vid-koci, I'm also interested in memory efficient node2vec. Veles looks very interesting! Do you have more detailed documents/papers about it? I read your code, but I'm still not sure...
@vid-koci Thank you so much for the publication. I'll take a look at it.
Thanks for the feedback! Unfortunately, that would be very hard to implement because Metall does not know which memory regions are used to allocate the object associated with a given...
The error is an expected one due to the limitation of the current implementation. A chunk holds multiple slots (allocation space) of the same allocation size. Metall uses a multi-layer...
Were you able to get a notable performance improvement by using a large chunk size? Unfortunately, there is no simple relationship between the chunk size and Metall's performance. I always...
Is supporting up to 2^27 chunk size still okay for you?
That's a good idea. I'll implement a public function`stateful_allocatorator_available()` . Oh, allocatorator is a typo😂. I'll also fix that.
BTW, would you like to have a web meeting to discuss Metall sometime this Month? I'm happy to take questions and/or new feature requests.
Thanks for the report, Liss! Yeah, I know it's not a public API, and it is better to avoid using it generally. But, unfortunately, the object construction logic is complicated...
I was able to fix the bug you found: https://github.com/LLNL/metall/pull/333 I'll merge the PR once I update some documentation.