Bradley Dice
Bradley Dice
@amkrajewski I checked most of the boxes above. I had one remaining question from the first review that I didn't see addressed (please forgive me if I missed something!). **References:**...
Sounds good! Thanks!
Sorry, I missed your last comment here. I think your proposal sounds good. Something like: 1. Either prime (or not) the async allocator 2. Do some allocations -- I like...
Yes, we're currently only exploring a change of behavior to skip priming when `initial_pool_size` is left as the default of `std::nullopt`.
In https://github.com/rapidsai/rmm/pull/2051, I ran PDS-H benchmarks and also microbenchmarks. Priming the async allocator seems to have no beneficial impact for workflow performance and increases startup costs, so I recommend we...
Thanks @vyasr -- I have a revised plan for CCCL here: https://github.com/NVIDIA/cccl/issues/1939#issuecomment-2830993942
@vyasr You were interested in increasing automation. What do you think of this proposal?
/ok to test 09d5cd5
/ok to test
@tpn Would you be able to avoid force-pushing here? It has a tendency to mess up review comments and we generally avoid it. We always squash on merge, so it...