Build and Package cuvs_bench in CI
#367
Conversation
divyegala
added
feature request
|
I see some c++-level cusolver include build errors in the Python package. |
|
/merge |
| @@ -0,0 +1,65 @@ | |||
| # ============================================================================= | |||
| # Copyright (c) 2023-2024, NVIDIA CORPORATION. | |||
There was a problem hiding this comment.
Please use only current year for all new files.
There was a problem hiding this comment.
Sorry, Divye. I noticed in your other PR that there are many places RAFT still needs to be change dto cuVS. I can't approve this yet. However, these should be minor and simple changes to make. Shall we use your other PR going forward and close this one to make things more manageable?
| @@ -37,10 +37,12 @@ HELP="$0 [<target> ...] [<flag> ...] [--cmake-args=\"<args>\"] [--cache-tool=<to | |||
| -n - no install step | |||
| --uninstall - uninstall files for specified targets which were built and installed prior | |||
| --compile-static-lib - compile static library for all components | |||
| --cpu-only - build CPU only components without CUDA. Currently only applies to bench-ann. | |||
| --limit-tests - semicolon-separated list of test executables to compile (e.g. NEIGHBORS_TEST;CLUSTER_TEST) | |||
| --limit-bench-ann - semicolon-separated list of ann benchmark executables to compute (e.g. HNSWLIB_ANN_BENCH;RAFT_IVF_PQ_ANN_BENCH) | |||
There was a problem hiding this comment.
RAFT needs to be changed to cuVS everywhere.
|
Approving, should be fine to merge once the RAFT/cuVS name changes are figured out. |
|
Closing for now because changes are in #368. Please open back up if this is absolutely neeed. |
Builds upon #367 Authors: - Divye Gala (https://github.com/divyegala) - Dante Gama Dessavre (https://github.com/dantegd) - Corey J. Nolet (https://github.com/cjnolet) Approvers: - Corey J. Nolet (https://github.com/cjnolet) - Bradley Dice (https://github.com/bdice) URL: #368
This PR sets up building and packaging of
cuvs_benchin CI, which will automatically setup publishing ofcuvs_benchpackages to rapids conda channels