From 23ba4d4a8e7389298ab7da4c32007230bd0480a1 Mon Sep 17 00:00:00 2001
From: Tom Lin <tom91136@gmail.com>
Date: Thu, 29 Jul 2021 13:45:38 +0100
Subject: [PATCH] Undo accidental debug changes

---
 cuda/bude.c           |  2 +-
 cuda/bude_cuda.cu     | 12 ------------
 kokkos/CMakeLists.txt |  2 +-
 kokkos/src/bude.h     |  2 +-
 4 files changed, 3 insertions(+), 15 deletions(-)

diff --git a/cuda/bude.c b/cuda/bude.c
index b61220a..1a4a065 100644
--- a/cuda/bude.c
+++ b/cuda/bude.c
@@ -94,7 +94,7 @@ int main(int argc, char *argv[])
       // printf ("Maxdiff: %.2f (%.3f vs %.3f)\n", maxdiff, resultsRef[i], resultsCUDA[i]);
     }
 
-    if (i < 1024)
+    if (i < 8)
       printf("%7.2f    vs   %7.2f  (%5.2f%%)\n", resultsRef[i], resultsCUDA[i], 100*diff);
   }
   printf("\nLargest difference was %.3f%%\n\n", maxdiff*100);
diff --git a/cuda/bude_cuda.cu b/cuda/bude_cuda.cu
index aff581c..82eeaec 100644
--- a/cuda/bude_cuda.cu
+++ b/cuda/bude_cuda.cu
@@ -84,8 +84,6 @@ void runCUDA(float* results)
   cudaDeviceSynchronize();
 
   double start = getTimestamp();
-  printf("<<%d,%d>> -> %d\n",global, local,params.nposes);
-
 
   for(int ii = 0; ii < params.iterations; ++ii)
   {
@@ -135,12 +133,10 @@ __device__ void compute_transformation_matrix(const float transform_0,
   transform[0].y = sx*sy*cz - cx*sz;
   transform[0].z = cx*sy*cz + sx*sz;
   transform[0].w = transform_3;
-  
   transform[1].x = cy*sz;
   transform[1].y = sx*sy*sz + cx*cz;
   transform[1].z = cx*sy*sz - sx*cz;
   transform[1].w = transform_4;
-
   transform[2].x = -sy;
   transform[2].y = sx*cy;
   transform[2].z = cx*cy;
@@ -195,18 +191,12 @@ __global__ void fasten_main(const int natlig,
         transforms_5[index],
         transform[i]);
     etot[i] = ZERO;
-    // etotals[ix] = transform[i][0].x;
   }
 
 #ifdef USE_SHARED
   __syncthreads();
 #endif
 
-    // etotals[ix] =  ix ;//  transform[3][1].y; // (transforms_0[ix + 1 * lsz]) ;// transform[0][0].x ;// transform[0][0].x;
-    // etot[1] = transforms_1[0] ;// transform[0][0].y;
-    // etot[2] = transforms_2[0] ;// transform[0][0].z;
-    // etot[3] = transforms_3[0] ;// transform[0][0].w;
-
   // Loop over ligand atoms
   int il = 0;
   do
@@ -231,7 +221,6 @@ __global__ void fasten_main(const int natlig,
         linitpos.y*transform[i][2].y + linitpos.z*transform[i][2].z;
     }
 
-
     // Loop over protein atoms
     int ip = 0;
     do
@@ -301,7 +290,6 @@ __global__ void fasten_main(const int natlig,
       etotals[td_base+i*blockDim.x] = etot[i]*HALF;
     }
   }
-  // etotals[ix] = transform[0][0].x;
 } //end of fasten_main
 
 
diff --git a/kokkos/CMakeLists.txt b/kokkos/CMakeLists.txt
index a7ab385..10e0d5b 100644
--- a/kokkos/CMakeLists.txt
+++ b/kokkos/CMakeLists.txt
@@ -1,7 +1,7 @@
 cmake_minimum_required(VERSION 3.12 FATAL_ERROR)
 
 project(bude_kokkos)
-set(CMAKE_VERBOSE_MAKEFILE ON)
+set(CMAKE_VERBOSE_MAKEFILE OFF)
 
 set(CMAKE_CXX_STANDARD 14)
 set(CMAKE_CXX_EXTENSIONS OFF)
diff --git a/kokkos/src/bude.h b/kokkos/src/bude.h
index 8ec533e..b9cd9c9 100644
--- a/kokkos/src/bude.h
+++ b/kokkos/src/bude.h
@@ -11,7 +11,7 @@
 #endif
 
 #ifndef WG_SIZE
-#define WG_SIZE 256
+#define WG_SIZE 128
 #endif
 
 #define DEFAULT_ITERS  8