Merge branch 'master' into multigpu_optimization_2019-07-05

README.md

@@ -152,11 +152,8 @@ globalFunction(void)
 
 Modules used when compiling the code on TIARA cluster. 
 
-  * intel/2016                         
-  * hdf5/1.8.16_openmpi_1.10.2_ic16.0   
   * cmake/3.9.5
-  * openmpi/1.10.2_ic16.0               
-  * gcc/5.3.0
+  * gcc/8.3.0
   * cuda/10.1
 
 

src/core/device.cu

@@ -96,6 +96,7 @@ printDeviceInfo(const Device device)
     printf("    Peak Memory Bandwidth (GiB/s): %f\n",
            2 * (props.memoryClockRate * 1e3) * props.memoryBusWidth / (8. * 1024. * 1024. * 1024.));
     printf("    ECC enabled: %d\n", props.ECCEnabled);
+
     // Memory usage
     size_t free_bytes, total_bytes;
     cudaMemGetInfo(&free_bytes, &total_bytes);
@@ -108,8 +109,10 @@ printDeviceInfo(const Device device)
     printf("    Local L1 cache supported: %d\n", props.localL1CacheSupported);
     printf("    Global L1 cache supported: %d\n", props.globalL1CacheSupported);
     printf("    L2 size: %d KiB\n", props.l2CacheSize / (1024));
-    printf("    Total const mem: %ld KiB\n", props.totalConstMem / (1024));
-    printf("    Shared mem per block: %ld KiB\n", props.sharedMemPerBlock / (1024));
+//MV: props.totalConstMem and props.sharedMemPerBlock cause assembler error
+//MV: while compiling in TIARA gp cluster. Therefore commeted out. 
+//!!    printf("    Total const mem: %ld KiB\n", props.totalConstMem / (1024));
+//!!    printf("    Shared mem per block: %ld KiB\n", props.sharedMemPerBlock / (1024));
     printf("  Other\n");
     printf("    Warp size: %d\n", props.warpSize);
     // printf("    Single to double perf. ratio: %dx\n",

src/standalone/simulation.cc

@@ -246,7 +246,7 @@ run_simulation(void)
 
         // Generate e for k. Needed for the sake of isotrophy.
         AcReal3 e_force;
-        if ((k_force.y == 0.0) && (k_force.z == 0.0)) {
+        if ((k_force.y == AcReal(0.0)) && (k_force.z == AcReal(0.0))) {
             e_force = (AcReal3){0.0, 1.0, 0.0};
         }
         else {