RSN 38 - Moving libcumlprims, libcugraphops, pylibcugraphops to the rapidsai channel
| Author | RAPIDS Ops |
| Status | In Progress |
| Topic | Platform Support Change |
| RAPIDS Version | v24.06 |
| Created | 03 June 2024 |
| Updated | N/A |
Overview
Currently, most RAPIDS conda packages are published to the rapidsai channel except for libcumlprims, libcugraphops, and pylibcugraphops, which are published to the nvidia channel.
These packages will be moved to the rapidsai channel, to simplify the user experience and ensure that all RAPIDS packages are available from the same channel.
For the 24.06 release, libcumlprims, libcugraphops, and pylibcugraphops will be added to the rapidsai channel.
These packages will be available on both the rapidsai and nvidia channels for 24.06 and 24.08.
Starting in 24.10, the packages will only be available on the rapidsai channel.
RAPIDS users should continue to use the conda channel priority -c rapidsai -c conda-forge -c nvidia to ensure that packages are installed from the correct channel (this is unchanged).
Impact
New installations of RAPIDS will be unaffected, as the rapidsai channel is already required for most RAPIDS packages.
Users updating libcumlprims, libcugraphops, and pylibcugraphops should be aware that the channel will change.