Printer support

src/common/gpu.cpp

@@ -101,6 +101,16 @@ char* get_str_bus_width(struct gpu_info* gpu) {
   return string;
 }
 
+char* get_str_lds_size(struct gpu_info* gpu) {
+  uint32_t size = 3+1+3+1;
+  assert(strlen(STRING_UNKNOWN)+1 <= size);
+  char* string = (char *) ecalloc(size, sizeof(char));
+
+  sprintf(string, "%d KB", gpu->mem->lds_size);
+
+  return string;
+}
+
 char* get_str_memory_clock(struct gpu_info* gpu) {
   return get_freq_as_str_mhz(gpu->mem->freq);
 }

src/common/gpu.hpp

@@ -61,6 +61,7 @@ struct memory {
   int32_t bus_width;
   int32_t freq;
   int32_t clk_mul; // clock multiplier
+  int32_t lds_size; // HSA specific for now
 };
 
 struct gpu_info {
@@ -88,6 +89,7 @@ char* get_str_freq(struct gpu_info* gpu);
 char* get_str_memory_size(struct gpu_info* gpu);
 char* get_str_memory_type(struct gpu_info* gpu);
 char* get_str_bus_width(struct gpu_info* gpu);
+char* get_str_lds_size(struct gpu_info* gpu);
 char* get_str_memory_clock(struct gpu_info* gpu);
 char* get_str_l2(struct gpu_info* gpu);
 char* get_str_peak_performance(struct gpu_info* gpu);

src/common/printer.cpp

@@ -48,6 +48,7 @@ enum {
   ATTRIBUTE_FREQUENCY,     // ALL
   ATTRIBUTE_PEAK,          // ALL
   ATTRIBUTE_COMPUTE_UNITS, // HSA
+  ATTRIBUTE_LDS_SIZE,      // HSA
   ATTRIBUTE_STREAMINGMP,   // CUDA
   ATTRIBUTE_CORESPERMP,    // CUDA
   ATTRIBUTE_CUDA_CORES,    // CUDA
@@ -55,7 +56,7 @@ enum {
   ATTRIBUTE_L2,            // CUDA
   ATTRIBUTE_MEMORY,        // CUDA
   ATTRIBUTE_MEMORY_FREQ,   // CUDA
-  ATTRIBUTE_BUS_WIDTH,     // CUDA
+  ATTRIBUTE_BUS_WIDTH,     // CUDA,HSA
   ATTRIBUTE_PEAK_TENSOR,   // CUDA
   ATTRIBUTE_EUS,           // Intel
   ATTRIBUTE_GT,            // Intel
@@ -69,6 +70,7 @@ static const AttributeField ATTRIBUTE_INFO[] = {
   { ATTRIBUTE_FREQUENCY,     "Max Frequency:",          "Max Freq.:" },
   { ATTRIBUTE_PEAK,          "Peak Performance:",       "Peak Perf.:" },
   { ATTRIBUTE_COMPUTE_UNITS, "Compute Units (CUs):",    "CUs" },
+  { ATTRIBUTE_LDS_SIZE,      "Local Data Share (LDS):"  "LDS:" },
   { ATTRIBUTE_STREAMINGMP,   "SMs:",                    "SMs:" },
   { ATTRIBUTE_CORESPERMP,    "Cores/SM:",               "Cores/SM:" },
   { ATTRIBUTE_CUDA_CORES,    "CUDA Cores:",             "CUDA Cores:" },
@@ -487,6 +489,8 @@ bool print_gpufetch_amd(struct gpu_info* gpu, STYLE s, struct color** cs, struct
   char* manufacturing_process = get_str_process(gpu->arch);
   char* cus = get_str_cu(gpu);
   char* max_frequency = get_str_freq(gpu);
+  char* bus_width = get_str_bus_width(gpu);
+  char* lds_size = get_str_lds_size(gpu);
 
   setAttribute(art, ATTRIBUTE_NAME, gpu_name);
   if (gpu_chip != NULL) {
@@ -496,6 +500,8 @@ bool print_gpufetch_amd(struct gpu_info* gpu, STYLE s, struct color** cs, struct
   setAttribute(art, ATTRIBUTE_TECHNOLOGY, manufacturing_process);
   setAttribute(art, ATTRIBUTE_FREQUENCY, max_frequency);
   setAttribute(art, ATTRIBUTE_COMPUTE_UNITS, cus);
+  setAttribute(art, ATTRIBUTE_BUS_WIDTH, bus_width);
+  setAttribute(art, ATTRIBUTE_LDS_SIZE, lds_size);
 
   bool use_short = false;
   uint32_t longest_attribute = longest_attribute_length(art, use_short);

src/hsa/hsa.cpp

@@ -52,7 +52,7 @@ hsa_status_t get_lds_size_callback(hsa_region_t region, void* data) {
 
     err = hsa_region_get_info(region, HSA_REGION_INFO_SIZE, &size);
     RET_IF_HSA_ERR(err);
-    
+
     *(size_t*)data = size;    
   }
   return HSA_STATUS_SUCCESS;
@@ -100,6 +100,15 @@ struct topology_h* get_topology_info(struct agent_info info) {
   return topo;
 }
 
+struct memory* get_memory_info(struct gpu_info* gpu, struct agent_info info) {
+  struct memory* mem = (struct memory*) emalloc(sizeof(struct memory));
+  
+  mem->bus_width = info.bus_width;
+  mem->lds_size = info.lds_size;
+
+  return mem;
+}
+
 struct gpu_info* get_gpu_info_hsa(int gpu_idx) {
   struct gpu_info* gpu = (struct gpu_info*) emalloc(sizeof(struct gpu_info));
   gpu->pci = NULL;
@@ -143,6 +152,7 @@ struct gpu_info* get_gpu_info_hsa(int gpu_idx) {
   gpu->name = (char *) emalloc(sizeof(char) * (strlen(info.device_mkt_name) + 1));
   strcpy(gpu->name, info.device_mkt_name);
   gpu->arch = get_uarch_from_hsa(gpu, info.gpu_name);
+  gpu->mem = get_memory_info(gpu, deviceProp);
 
   if (gpu->arch == NULL) {
     return NULL;