Update README (#2)

* Update README

* fix branch references

---------

Co-authored-by: James Lamb <[email protected]>

Author: bdice

.github/workflows/build.yaml

@@ -34,7 +34,7 @@ concurrency:
 jobs:
   cpp-build:
     secrets: inherit
-    uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-build.yaml@cuda13.0
+    uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-build.yaml@branch-25.10
     with:
       build_type: ${{ inputs.build_type || 'branch' }}
       branch: ${{ inputs.branch }}
@@ -45,7 +45,7 @@ jobs:
     if: ${{ !startsWith(github.ref, 'refs/tags/') }}
     needs: [cpp-build]
     secrets: inherit
-    uses: rapidsai/shared-workflows/.github/workflows/conda-upload-packages.yaml@cuda13.0
+    uses: rapidsai/shared-workflows/.github/workflows/conda-upload-packages.yaml@branch-25.10
     with:
       build_type: ${{ inputs.build_type || 'branch' }}
       branch: ${{ inputs.branch }}
@@ -55,7 +55,7 @@ jobs:
     if: ${{ startsWith(github.ref, 'refs/tags/') }}
     needs: [cpp-build]
     secrets: inherit
-    uses: rapidsai/shared-workflows/.github/workflows/conda-upload-packages.yaml@cuda13.0
+    uses: rapidsai/shared-workflows/.github/workflows/conda-upload-packages.yaml@branch-25.10
     with:
       build_type: ${{ inputs.build_type || 'branch' }}
       branch: ${{ inputs.branch }}

.github/workflows/pr.yaml

@@ -14,7 +14,7 @@ jobs:
       - devcontainer
       - telemetry-setup
     secrets: inherit
-    uses: rapidsai/shared-workflows/.github/workflows/pr-builder.yaml@cuda13.0
+    uses: rapidsai/shared-workflows/.github/workflows/pr-builder.yaml@branch-25.10
   telemetry-setup:
     runs-on: ubuntu-latest
     continue-on-error: true
@@ -29,20 +29,20 @@ jobs:
   checks:
     secrets: inherit
     needs: telemetry-setup
-    uses: rapidsai/shared-workflows/.github/workflows/checks.yaml@cuda13.0
+    uses: rapidsai/shared-workflows/.github/workflows/checks.yaml@branch-25.10
     with:
       ignored_pr_jobs: telemetry-summarize
   conda-cpp-build:
     needs: checks
     secrets: inherit
-    uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-build.yaml@cuda13.0
+    uses: rapidsai/shared-workflows/.github/workflows/conda-cpp-build.yaml@branch-25.10
     with:
       build_type: pull-request
       script: ci/build_cpp.sh
   devcontainer:
     secrets: inherit
     needs: telemetry-setup
-    uses: rapidsai/shared-workflows/.github/workflows/build-in-devcontainer.yaml@cuda13.0
+    uses: rapidsai/shared-workflows/.github/workflows/build-in-devcontainer.yaml@branch-25.10
     with:
       arch: '["amd64", "arm64"]'
       cuda: '["13.0"]'

.github/workflows/trigger-breaking-change-alert.yaml

@@ -12,7 +12,7 @@ jobs:
   trigger-notifier:
     if: contains(github.event.pull_request.labels.*.name, 'breaking')
     secrets: inherit
-    uses: rapidsai/shared-workflows/.github/workflows/breaking-change-alert.yaml@cuda13.0
+    uses: rapidsai/shared-workflows/.github/workflows/breaking-change-alert.yaml@branch-25.10
     with:
       sender_login: ${{ github.event.sender.login }}
       sender_avatar: ${{ github.event.sender.avatar_url }}

README.md

@@ -1,8 +1,8 @@
 # cuMLPrims
 
-This repository contains  C++ and CUDA code of muti-node multi-GPU (MNMG) ML mathematical primitives and some algorithms, that are used by [the main cuML project](https://github.com/rapidsai/cuml). The build system uses CMake for build configuration, and an out-of-source build is recommended.
+This repository contains C++ and CUDA code of muti-node multi-GPU (MNMG) ML mathematical primitives and some algorithms, that are used by [the main cuML project](https://github.com/rapidsai/cuml). The build system uses CMake for build configuration, and an out-of-source build is recommended.
 
-As of version 0.13, the MNMG code included in cuMLPrims follows the model one-process-per-GPU (OPG), where the code uses a communication library (based on cuML's comms) and each process has one GPU assigned to it. This is in contrast to single-process-multi-GPU (SPMG) approaches, which are no longer part of the code base.
+The MNMG code included in cuMLPrims follows the model one-process-per-GPU (OPG), where the code uses a communication library (based on cuML's comms) and each process has one GPU assigned to it. This is in contrast to single-process-multi-GPU (SPMG) approaches, which are no longer part of the code base.
 
 ## Folder Structure
 
@@ -14,30 +14,28 @@ The folder structure mirrors closely the structure of github cuML. The folders a
     - `cpp/cmake`: CMake related scripts.
     - `cpp/include`: The include folder for headers that are necessary to be installed/distributed to use the libcumlprims.so artifact by users of the library.
     - `cpp/src_prims_opg`: Contains source code for MNMG ML primitives. It also contains source code for algorithms that use the primitives that are still included in cuMLPrims as opposed to cuML.
-    - `cpp/test`: Googletest based unit tests (work in progress).
+    - `cpp/test`: Googletest based unit tests.
 
 ## Building cuMLPrims:
 
 ### Requirements
 
 The main artifact produced by the build system is the shared library libcumlprims. Ensure the following dependencies are satisfied:
 
-1. CMake (>= 3.14.5)
-2. CUDA (>= 10.0)
-3. GCC (>= 5.4.0)
-4. NCCL (>= 2.4.6.1)
-5. UCX with CUDA support (optional)(>=1.7) - enables point-to-point messaging in the cuML communicator.
-6. zlib
+1. CMake (>= 3.30.4)
+2. CUDA (>= 12.0)
+3. GCC (>= 14)
+4. NCCL (>= 2.5)
 
-It is recommended to use conda for environment/package management. If doing so, it is recommended to use the convenience environment .yml file located in [**the cuML repository**, in `conda/environments/cuml_dev_cudax.y.yml` (replace x.y for your CUDA version)](https://github.com/rapidsai/cuml/tree/branch-0.14/conda/environments). This file contains most of the dependencies ment1ioned above (notable exceptions are gcc and zlib). To use it, for example to create an environment named cuml_dev for CUDA 10.0 and Python 3.7, you can use the follow command:
+It is recommended to use conda for environment/package management. See `conda/environments/` for available environment files.
 
 ```bash
-conda env create -n cuml_dev python=3.7 --file=conda/environments/cuml_dev_cuda10.0.yml
+conda env create --name cumlprims_dev --file conda/environments/all_cuda-130_arch-$(arch).yaml
 ```
 
 ### Using build.sh script
 
-As a convenience, a `build.sh` script is provided which can be used to execute the build commands in an automated manner.  Note that the libraries will be installed to the location set in `$INSTALL_PREFIX` if set (i.e. `export INSTALL_PREFIX=/install/path`), otherwise to `$CONDA_PREFIX`.
+As a convenience, a `build.sh` script is provided which can be used to execute the build commands in an automated manner. Note that the libraries will be installed to the location set in `$INSTALL_PREFIX` if set (i.e. `export INSTALL_PREFIX=/install/path`), otherwise to `$CONDA_PREFIX`.
 ```bash
 $ ./build.sh                           # build the libcuml library and tests
                                        # install them to $INSTALL_PREFIX if set, otherwise $CONDA_PREFIX
@@ -46,12 +44,12 @@ $ ./build.sh                           # build the libcuml library and tests
 Other `build.sh` options:
 
 ```bash
-$ ./build.sh clean                     # remove any prior build artifacts and configuration (start over)
+$ ./build.sh clean                          # remove any prior build artifacts and configuration (start over)
 $ ./build.sh libcumlprims -v                # build and install libcumlprims with verbose output
 $ ./build.sh libcumlprims -g                # build and install libcumlprims for debug
 $ PARALLEL_LEVEL=4 ./build.sh libcumlprims  # build and install libcumlprims limiting parallel build jobs to 4 (make -j4)
-$ ./build.sh libcuml -n                # build libcuml but do not install
-$ ./build.sh libcumlprims --allgpuarch        # build the  tests for all supported GPU architectures
+$ ./build.sh libcuml -n                     # build libcuml but do not install
+$ ./build.sh libcumlprims --allgpuarch      # build the tests for all supported GPU architectures
 
 ```
 
@@ -107,11 +105,7 @@ cuMLPrims CMake has the following configurable flags available:
 
 | Flag | Possible Values | Default Value | Behavior |
 | --- | --- | --- | --- |
-| BUILD_OPG_TESTS | [ON, OFF] | OFF | Build MPI cumlcomms based C++ unit tests (in progress, refer to step 4. of the build steps).  |
-| BUILD_CUMLPRIMS_LIBRARY | [ON, OFF]  | ON  | Enable/disable building libcumprims shared library. |
-| CMAKE_CXX11_ABI | [ON, OFF]  | ON  | Enable/disable the GLIBCXX11 ABI  |
-| DISABLE_OPENMP | [ON, OFF]  | OFF  | Set to `ON` to disable OpenMP  |
-| KERNEL_INFO | [ON, OFF]  | OFF  | Enable/disable kernel resource usage info in nvcc. |
-| LINE_INFO | [ON, OFF]  | OFF  | Enable/disable lineinfo in nvcc.  |
-| NVTX | [ON, OFF]  | OFF  | Enable/disable nvtx markers in libcumlprims.  |
-| GPU_ARCHS |  List of GPU architectures, semicolon-separated | 60;70;75  | List of GPU architectures that all artifacts are compiled for.  |
+| BUILD_OPG_TESTS | [ON, OFF] | OFF | Build MPI cumlcomms based C++ unit tests (in progress, refer to step 4. of the build steps). |
+| BUILD_CUMLPRIMS_LIBRARY | [ON, OFF] | ON | Enable/disable building libcumlprims shared library. |
+| DISABLE_OPENMP | [ON, OFF] | OFF | Set to `ON` to disable OpenMP |
+| NVTX | [ON, OFF] | OFF | Enable/disable nvtx markers in libcumlprims.|