misc update

diff --git a/src/caffeine/layers/inner_product_layer.cu b/src/caffeine/layers/inner_product_layer.cu
index fa40093..8c06dfd 100644 (file)
--- a/src/caffeine/layers/inner_product_layer.cu
+++ b/src/caffeine/layers/inner_product_layer.cu
@@ -75,6 +75,7 @@ void InnerProductLayer<Dtype>::Forward_cpu(const vector<Blob<Dtype>*>& bottom,
                                        (float*)(top_data) + (N_ * i), 1);
                }
        }
+    break;
   case sizeof(double):
     // matrix multiply
        cblas_dgemm(CblasRowMajor, CblasNoTrans, CblasNoTrans, M_, N_, K_,
@@ -87,6 +88,7 @@ void InnerProductLayer<Dtype>::Forward_cpu(const vector<Blob<Dtype>*>& bottom,
                                        (double*)(top_data) + (N_ * i), 1);
                }
        }
+    break;
   default:
        CHECK(false) << "Unknown data type.";
   }
@@ -133,11 +135,13 @@ void InnerProductLayer<Dtype>::Forward_gpu(const vector<Blob<Dtype>*>& bottom,
        CUBLAS_CHECK(cublasSgemm(Caffeine::cublas_handle(), CUBLAS_OP_N,
                        CUBLAS_OP_N, N_, M_, K_, (float*)&alpha, (const float*)weight, N_,
                        (const float*)bottom_data, K_, (float*)&beta, (float*)top_data, N_));
+    break;
   case sizeof(double):
     // matrix multiply
        CUBLAS_CHECK(cublasDgemm(Caffeine::cublas_handle(), CUBLAS_OP_N,
                        CUBLAS_OP_N, N_, M_, K_, (double*)&alpha, (const double*)weight, N_,
                        (const double*)bottom_data, K_, (double*)&beta, (double*)top_data, N_));
+    break;
   default:
        CHECK(false) << "Unknown data type.";
   }

diff --git a/src/caffeine/proto/layer_param.proto b/src/caffeine/proto/layer_param.proto
index 58dbe93..e921665 100644 (file)
--- a/src/caffeine/proto/layer_param.proto
+++ b/src/caffeine/proto/layer_param.proto
@@ -26,7 +26,9 @@ message LayerParameter {
 }
 
 message FillerParameter {
-  required string type = 1 [default = 'constant'];
+  // The filler type. In default we will set it to Gaussian for easy
+  // debugging.
+  required string type = 1 [default = 'gaussian'];
   optional float value = 2 [default = 0]; // the value in constant filler
   optional float min = 3 [default = 0]; // the min value in uniform filler
   optional float max = 4 [default = 1]; // the max value in uniform filler

diff --git a/src/caffeine/test/test_caffeine_main.cpp b/src/caffeine/test/test_caffeine_main.cpp
index ae9d179..a3a3b92 100644 (file)
--- a/src/caffeine/test/test_caffeine_main.cpp
+++ b/src/caffeine/test/test_caffeine_main.cpp
@@ -6,6 +6,8 @@
 
 using namespace std;
 
+cudaDeviceProp CAFFEINE_TEST_CUDA_PROP;
+
 int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);
   ::google::InitGoogleLogging(argv[0]);
@@ -15,27 +17,26 @@ int main(int argc, char** argv) {
   cout << "Cuda number of devices: " << device << endl;
   cudaGetDevice(&device);
   cout << "Current device id: " << device << endl;
-  cudaDeviceProp prop;
-  cudaGetDeviceProperties(&prop, device);
-  printf("Major revision number:         %d\n",  prop.major);
-  printf("Minor revision number:         %d\n",  prop.minor);
-  printf("Name:                          %s\n",  prop.name);
-  printf("Total global memory:           %u\n",  prop.totalGlobalMem);
-  printf("Total shared memory per block: %u\n",  prop.sharedMemPerBlock);
-  printf("Total registers per block:     %d\n",  prop.regsPerBlock);
-  printf("Warp size:                     %d\n",  prop.warpSize);
-  printf("Maximum memory pitch:          %u\n",  prop.memPitch);
-  printf("Maximum threads per block:     %d\n",  prop.maxThreadsPerBlock);
+  cudaGetDeviceProperties(&CAFFEINE_TEST_CUDA_PROP, device);
+  printf("Major revision number:         %d\n",  CAFFEINE_TEST_CUDA_PROP.major);
+  printf("Minor revision number:         %d\n",  CAFFEINE_TEST_CUDA_PROP.minor);
+  printf("Name:                          %s\n",  CAFFEINE_TEST_CUDA_PROP.name);
+  printf("Total global memory:           %u\n",  CAFFEINE_TEST_CUDA_PROP.totalGlobalMem);
+  printf("Total shared memory per block: %u\n",  CAFFEINE_TEST_CUDA_PROP.sharedMemPerBlock);
+  printf("Total registers per block:     %d\n",  CAFFEINE_TEST_CUDA_PROP.regsPerBlock);
+  printf("Warp size:                     %d\n",  CAFFEINE_TEST_CUDA_PROP.warpSize);
+  printf("Maximum memory pitch:          %u\n",  CAFFEINE_TEST_CUDA_PROP.memPitch);
+  printf("Maximum threads per block:     %d\n",  CAFFEINE_TEST_CUDA_PROP.maxThreadsPerBlock);
   for (int i = 0; i < 3; ++i)
-    printf("Maximum dimension %d of block:  %d\n", i, prop.maxThreadsDim[i]);
+    printf("Maximum dimension %d of block:  %d\n", i, CAFFEINE_TEST_CUDA_PROP.maxThreadsDim[i]);
   for (int i = 0; i < 3; ++i)
-    printf("Maximum dimension %d of grid:   %d\n", i, prop.maxGridSize[i]);
-  printf("Clock rate:                    %d\n",  prop.clockRate);
-  printf("Total constant memory:         %u\n",  prop.totalConstMem);
-  printf("Texture alignment:             %u\n",  prop.textureAlignment);
-  printf("Concurrent copy and execution: %s\n",  (prop.deviceOverlap ? "Yes" : "No"));
-  printf("Number of multiprocessors:     %d\n",  prop.multiProcessorCount);
-  printf("Kernel execution timeout:      %s\n",  (prop.kernelExecTimeoutEnabled ? "Yes" : "No"));
+    printf("Maximum dimension %d of grid:   %d\n", i, CAFFEINE_TEST_CUDA_PROP.maxGridSize[i]);
+  printf("Clock rate:                    %d\n",  CAFFEINE_TEST_CUDA_PROP.clockRate);
+  printf("Total constant memory:         %u\n",  CAFFEINE_TEST_CUDA_PROP.totalConstMem);
+  printf("Texture alignment:             %u\n",  CAFFEINE_TEST_CUDA_PROP.textureAlignment);
+  printf("Concurrent copy and execution: %s\n",  (CAFFEINE_TEST_CUDA_PROP.deviceOverlap ? "Yes" : "No"));
+  printf("Number of multiprocessors:     %d\n",  CAFFEINE_TEST_CUDA_PROP.multiProcessorCount);
+  printf("Kernel execution timeout:      %s\n",  (CAFFEINE_TEST_CUDA_PROP.kernelExecTimeoutEnabled ? "Yes" : "No"));
   
   return RUN_ALL_TESTS();
 }