[v0.22] Small various fixes

build.sh

@@ -18,13 +18,13 @@ fi
 # - set CMAKE_CUDA_COMPILER to your nvcc binary:
 # - set CMAKE_CUDA_COMPILER_TOOLKIT_ROOT to the CUDA root dir
 # for example:
-# cmake -DCMAKE_CUDA_COMPILER=/usr/local/cuda/bin/nvcc -DCMAKE_CUDA_COMPILER_TOOLKIT_ROOT=/usr/local/cuda/ ..
+# cmake -DCMAKE_BUILD_TYPE=$BUILD_TYPE -DCMAKE_CUDA_COMPILER=/usr/local/cuda/bin/nvcc -DCMAKE_CUDA_COMPILER_TOOLKIT_ROOT=/usr/local/cuda/ ..
 
 # In case you want to explicitely disable a backend, you can:
 # Disable CUDA backend:
-# cmake -DENABLE_CUDA_BACKEND=OFF ..
+# cmake -DCMAKE_BUILD_TYPE=$BUILD_TYPE -DENABLE_CUDA_BACKEND=OFF ..
 # Disable Intel backend:
-# cmake -DENABLE_INTEL_BACKEND=OFF ..
+# cmake -DCMAKE_BUILD_TYPE=$BUILD_TYPE -DENABLE_INTEL_BACKEND=OFF ..
 
 cmake -DCMAKE_BUILD_TYPE=$BUILD_TYPE ..
 make -j$(nproc)

src/common/gpu.hpp

@@ -60,18 +60,20 @@ struct memory {
 };
 
 struct gpu_info {
+  int32_t idx;
   VENDOR vendor;
   struct uarch* arch;
   char* name;
   int64_t freq;
   struct pci* pci;
-  struct topology_c* topo_c;
-  struct topology_i* topo_i;
+  int64_t peak_performance;
+  // CUDA specific
+  int64_t peak_performance_tcu;
   struct memory* mem;
   struct cache* cach;
-  int64_t peak_performance;
-  int64_t peak_performance_tcu;
-  int32_t idx;
+  struct topology_c* topo_c;
+  // Intel specific
+  struct topology_i* topo_i;
 };
 
 VENDOR get_gpu_vendor(struct gpu_info* gpu);

src/common/main.cpp

@@ -8,7 +8,7 @@
 #include "../cuda/cuda.hpp"
 #include "../cuda/uarch.hpp"
 
-static const char* VERSION = "0.21";
+static const char* VERSION = "0.22";
 
 void print_help(char *argv[]) {
   const char **t = args_str;

src/intel/intel.cpp

@@ -43,7 +43,7 @@ struct gpu_info* get_gpu_info_intel() {
 bool print_gpu_intel(struct gpu_info* gpu) {
   if(gpu->vendor != GPU_VENDOR_INTEL) return false;
 
-  printf("Intel %s\n", gpu->name);
+  printf("%s\n", gpu->name);
 
   return true;
 }

src/intel/uarch.cpp

@@ -166,7 +166,7 @@ struct uarch* get_uarch_from_pci(struct pci* pci) {
 
 char* get_name_from_uarch(struct uarch* arch) {
   char* name = (char *) emalloc(sizeof(char) * (strlen(arch->chip_str) + 6 + 1));
-  sprintf(name, "Intel %s", arch->chip_str);
+  sprintf(name, "%s", arch->chip_str);
   return name;
 }