Merge pull request #31 from Zhaoli2042/main

changes nersc compilation (works with latest nvhpc/cuda)

Cluster-Setup/NERSC/NVC_COMPILE_NERSC

@@ -1,15 +1,16 @@
 #!/bin/bash
 
+# The following line uses latest nvhpc/cuda version
 module load PrgEnv-nvhpc
 
 # The following lines are added since March 2024
 # NERSC changed their default nvhpc version
-
-module load nvhpc/22.7
-module unload nvhpc/22.7
+# Commented since latest gRASPA code now works
+#module load nvhpc/22.7
+#module unload nvhpc/22.7
 #module load cudatoolkit/11.7
-module load cudatoolkit/11.5
-module load nvhpc/22.7
+#module load cudatoolkit/11.5
+#module load nvhpc/22.7
 
 rm *.o nvc_main.x
 

Cluster-Setup/NERSC/NVC_COMPILE_NERSC_VANILLA

@@ -1,15 +1,16 @@
 #!/bin/bash
 
+# The following line uses latest nvhpc/cuda version
 module load PrgEnv-nvhpc
 
 # The following lines are added since March 2024
 # NERSC changed their default nvhpc version
-
-module load nvhpc/22.7
-module unload nvhpc/22.7
+# Commented since latest gRASPA code now works
+#module load nvhpc/22.7
+#module unload nvhpc/22.7
 #module load cudatoolkit/11.7
-module load cudatoolkit/11.5
-module load nvhpc/22.7
+#module load cudatoolkit/11.5
+#module load nvhpc/22.7
 
 rm *.o nvc_main.x
 

Cluster-Setup/NERSC/readme.md

@@ -4,8 +4,8 @@
   * If you don't need the ML potential, download the code, `cd src_clean/`, and start from [step 6](#Step-6).
   * Installation on other clusters are similar to on Perlmutter of NERSC. Follow the instructions here, and if you encounter issues, please consult with your institution's IT. 
     * [Northwestern Quest IT](https://www.it.northwestern.edu/departments/it-services-support/research/computing/quest/)
-  * Check the [nvhpc](https://developer.nvidia.com/hpc-sdk) version. Currently the code works for **22.5 and 22.7**
-    * on NERSC, you can do ```module load nvhpc/22.7``` to use the 22.7 version of nvhpc.
+  * Check the [nvhpc](https://developer.nvidia.com/hpc-sdk) version. 
+    * on NERSC, you can do ```module load PrgEnv-nvhpc``` to use the latest version of nvhpc/cuda.
     * Also check out the [issue on this topic](https://github.com/snurr-group/gRASPA/issues/9)
 # Step 1
 We download TensorFlow2 C++ API to a local directory: (assuming in the HOME directory)