Update GPU_ARCHS default to RAPIDS (#1064)

Signed-off-by: Jason Lowe <[email protected]>
NVIDIA · Apr 12, 2023 · c61f613 · c61f613
1 parent e6564f1
commit c61f613
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 13 deletions.
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -67,18 +67,18 @@ settings. If an explicit reconfigure of libcudf is needed (e.g.: when changing c
 The following build properties can be set on the Maven command-line (e.g.: `-DCPP_PARALLEL_LEVEL=4`)
 to control aspects of the build:
 
-|Property Name                       |Description                            |Default|
-|------------------------------------|---------------------------------------|-------|
-|`CPP_PARALLEL_LEVEL`                |Parallelism of the C++ builds          |10     |
-|`GPU_ARCHS`                         |CUDA architectures to target           |ALL    |
-|`CUDF_USE_PER_THREAD_DEFAULT_STREAM`|CUDA per-thread default stream         |ON     |
-|`RMM_LOGGING_LEVEL`                 |RMM logging control                    |OFF    |
-|`USE_GDS`                           |Compile with GPU Direct Storage support|OFF    |
-|`BUILD_TESTS`                       |Compile tests                          |OFF    |
-|`BUILD_BENCHMARKS`                  |Compile benchmarks                     |OFF    |
-|`libcudf.build.configure`           |Force libcudf build to configure       |false  |
-|`libcudf.clean.skip`                |Whether to skip cleaning libcudf build |true   |
-|`submodule.check.skip`              |Whether to skip checking git submodules|false  |
+|Property Name                       |Description                            | Default |
+|------------------------------------|---------------------------------------|---------|
+|`CPP_PARALLEL_LEVEL`                |Parallelism of the C++ builds          | 10      |
+|`GPU_ARCHS`                         |CUDA architectures to target           | RAPIDS  |
+|`CUDF_USE_PER_THREAD_DEFAULT_STREAM`|CUDA per-thread default stream         | ON      |
+|`RMM_LOGGING_LEVEL`                 |RMM logging control                    | OFF     |
+|`USE_GDS`                           |Compile with GPU Direct Storage support| OFF     |
+|`BUILD_TESTS`                       |Compile tests                          | OFF     |
+|`BUILD_BENCHMARKS`                  |Compile benchmarks                     | OFF     |
+|`libcudf.build.configure`           |Force libcudf build to configure       | false   |
+|`libcudf.clean.skip`                |Whether to skip cleaning libcudf build | true    |
+|`submodule.check.skip`              |Whether to skip checking git submodules| false   |
 
 
 ### Local testing of cross-repo contributions cudf, spark-rapids-jni, and spark-rapids

diff --git a/pom.xml b/pom.xml
@@ -76,7 +76,7 @@
   <properties>
     <arrow.version>0.15.1</arrow.version>
     <CPP_PARALLEL_LEVEL>10</CPP_PARALLEL_LEVEL>
-    <GPU_ARCHS>ALL</GPU_ARCHS>
+    <GPU_ARCHS>RAPIDS</GPU_ARCHS>
     <CUDF_USE_PER_THREAD_DEFAULT_STREAM>ON</CUDF_USE_PER_THREAD_DEFAULT_STREAM>
     <RMM_LOGGING_LEVEL>OFF</RMM_LOGGING_LEVEL>
     <SPARK_RAPIDS_JNI_CXX_FLAGS/>