[v0.11] Fix compilation error and ambiguity with CUDA and Intel backend when enabled at the same time due to functions with the same name

src/common/gpu.cpp

@@ -143,7 +143,7 @@ char* get_str_peak_performance(struct gpu_info* gpu) {
 }
 
 char* get_str_peak_performance_tensor(struct gpu_info* gpu) {
-  return get_str_peak_performance_generic(gpu->peak_performance_t);
+  return get_str_peak_performance_generic(gpu->peak_performance_tcu);
 }
 
 char* get_str_generic(int32_t data) {

src/common/gpu.hpp

@@ -69,7 +69,7 @@ struct gpu_info {
   struct memory* mem;
   struct cache* cach;
   int64_t peak_performance;
-  int64_t peak_performance_t;
+  int64_t peak_performance_tcu;
   int32_t idx;
 };
 

src/cuda/cuda.cpp

@@ -76,12 +76,12 @@ struct memory* get_memory_info(struct gpu_info* gpu, cudaDeviceProp prop) {
 }
 
 // Compute peak performance when using CUDA cores
-int64_t get_peak_performance(struct gpu_info* gpu) {
+int64_t get_peak_performance_cuda(struct gpu_info* gpu) {
   return gpu->freq * 1000000 * gpu->topo->cuda_cores * 2;
 }
 
 // Compute peak performance when using tensor cores
-int64_t get_peak_performance_t(struct gpu_info* gpu) {
+int64_t get_peak_performance_tcu(struct gpu_info* gpu) {
   return gpu->freq * 1000000 * 4 * 4 * 8 * gpu->topo->tensor_cores;
 }
 
@@ -138,8 +138,8 @@ struct gpu_info* get_gpu_info_cuda(int gpu_idx) {
   gpu->cach = get_cache_info(deviceProp);
   gpu->mem = get_memory_info(gpu, deviceProp);
   gpu->topo = get_topology_info(deviceProp);
-  gpu->peak_performance = get_peak_performance(gpu);
+  gpu->peak_performance = get_peak_performance_cuda(gpu);
-  gpu->peak_performance_t = get_peak_performance_t(gpu);
+  gpu->peak_performance_tcu = get_peak_performance_tcu(gpu);
 
   return gpu;
 }

src/cuda/pci.cpp

@@ -19,7 +19,7 @@
  * or in pci.ids itself)
  */
 
-GPUCHIP get_chip_from_pci(struct pci* pci) {
+GPUCHIP get_chip_from_pci_cuda(struct pci* pci) {
   CHECK_PCI_START
   CHECK_PCI(pci, 0x25e5, CHIP_GA107BM)
   CHECK_PCI(pci, 0x25e2, CHIP_GA107BM)

src/cuda/pci.hpp

@@ -14,7 +14,6 @@
 
 struct pci;
 
-struct pci* get_pci_from_pciutils(struct pci_dev *devices);
+GPUCHIP get_chip_from_pci_cuda(struct pci* pci);
-GPUCHIP get_chip_from_pci(struct pci* pci);
 
 #endif

src/cuda/uarch.cpp

@@ -39,7 +39,7 @@ static const char *uarch_str[] = {
 #define CHECK_UARCH_START if (false) {}
 #define CHECK_UARCH(arch, chip_, str, uarch, process) \
    else if (arch->chip == chip_) fill_uarch(arch, str, uarch, process);
-#define CHECK_UARCH_END else { printBug("map_chip_to_uarch: Unknown chip id: %d", arch->chip); fill_uarch(arch, STRING_UNKNOWN, UARCH_UNKNOWN, 0); }
+#define CHECK_UARCH_END else { printBug("map_chip_to_uarch_cuda: Unknown chip id: %d", arch->chip); fill_uarch(arch, STRING_UNKNOWN, UARCH_UNKNOWN, 0); }
 
 void fill_uarch(struct uarch* arch, char const *str, MICROARCH u, uint32_t process) {
   arch->chip_str = (char *) emalloc(sizeof(char) * (strlen(str)+1));
@@ -54,7 +54,7 @@ void fill_uarch(struct uarch* arch, char const *str, MICROARCH u, uint32_t proce
  * o CHIP_XXXGL: indicates a professional-class (Quadro/Tesla) chip
  * o CHIP_XXXM:  indicates a mobile chip
  */
-void map_chip_to_uarch(struct uarch* arch) {
+void map_chip_to_uarch_cuda(struct uarch* arch) {
   CHECK_UARCH_START
   // TESLA (1.0, 1.1, 1.2, 1.3)                                //
   CHECK_UARCH(arch, CHIP_G80,      "G80",      UARCH_TESLA,   90)
@@ -243,9 +243,8 @@ struct uarch* get_uarch_from_cuda(struct gpu_info* gpu) {
   arch->cc_major = deviceProp.major;
   arch->cc_minor = deviceProp.minor;
   arch->compute_capability = deviceProp.major * 10 + deviceProp.minor;
-  arch->chip = get_chip_from_pci(gpu->pci);
+  arch->chip = get_chip_from_pci_cuda(gpu->pci);
-
+  map_chip_to_uarch_cuda(arch);
-  map_chip_to_uarch(arch);
 
   return arch;
 }

src/intel/intel.cpp

@@ -8,7 +8,7 @@
 #include "../common/pci.hpp"
 #include "../common/global.hpp"
 
-int64_t get_peak_performance(struct gpu_info* gpu) {
+int64_t get_peak_performance_intel(struct gpu_info* gpu) {
   return gpu->freq * 1000000 * gpu->topo_i->eu_subslice * gpu->topo_i->subslices * 8 * 2;
 }
 
@@ -22,7 +22,7 @@ struct gpu_info* get_gpu_info_intel() {
   gpu->name = get_name_from_uarch(gpu->arch);
   gpu->topo_i = get_topology_info(gpu->arch);
   gpu->freq = get_max_freq_from_file(gpu->pci);
-  gpu->peak_performance = get_peak_performance(gpu);
+  gpu->peak_performance = get_peak_performance_intel(gpu);
 
   return gpu;
 }

src/intel/pci.cpp

@@ -13,7 +13,7 @@
 /*
  * https://github.com/mesa3d/mesa/blob/main/include/pci_ids/i965_pci_ids.h
  */
-GPUCHIP get_chip_from_pci(struct pci* pci) {
+GPUCHIP get_chip_from_pci_intel(struct pci* pci) {
   CHECK_PCI_START
   // Gen6
   CHECK_PCI(pci, 0x0102, CHIP_HD_2000)

src/intel/pci.hpp

@@ -14,7 +14,6 @@
 
 struct pci;
 
-struct pci* get_pci_from_pciutils(struct pci_dev *devices);
+GPUCHIP get_chip_from_pci_intel(struct pci* pci);
-GPUCHIP get_chip_from_pci(struct pci* pci);
 
 #endif

src/intel/uarch.cpp

@@ -7,6 +7,7 @@
 #include "../common/global.hpp"
 #include "../common/gpu.hpp"
 #include "chips.hpp"
+#include "pci.hpp"
 
 // Data not available
 #define NA                   -1
@@ -68,7 +69,7 @@ static const char *gt_str[] = {
 #define CHECK_UARCH_START if (false) {}
 #define CHECK_UARCH(arch, chip_, str, uarch, gt, process) \
    else if (arch->chip == chip_) fill_uarch(arch, str, uarch, gt, process);
-#define CHECK_UARCH_END else { printBug("map_chip_to_uarch: Unknown chip id: %d", arch->chip); fill_uarch(arch, STRING_UNKNOWN, UARCH_UNKNOWN, GT_UNKNOWN, 0); }
+#define CHECK_UARCH_END else { printBug("map_chip_to_uarch_intel: Unknown chip id: %d", arch->chip); fill_uarch(arch, STRING_UNKNOWN, UARCH_UNKNOWN, GT_UNKNOWN, 0); }
 
 #define CHECK_TOPO_START if (false) {}
 #define CHECK_TOPO(topo, arch, uarch_, gt_, eu_sub, sub, sli) \
@@ -89,7 +90,7 @@ void fill_uarch(struct uarch* arch, char const *str, MICROARCH u, int32_t gt, ui
   arch->gt = gt;
 }
 
-void map_chip_to_uarch(struct uarch* arch) {
+void map_chip_to_uarch_intel(struct uarch* arch) {
   CHECK_UARCH_START
   // Gen6
   CHECK_UARCH(arch, CHIP_HD_2000,     "HD Graphics 2000",        UARCH_GEN6,   GT1,  32)
@@ -147,8 +148,8 @@ struct uarch* get_uarch_from_pci(struct pci* pci) {
   struct uarch* arch = (struct uarch*) emalloc(sizeof(struct uarch));
 
   arch->chip_str = NULL;
-  arch->chip = get_chip_from_pci(pci);
+  arch->chip = get_chip_from_pci_intel(pci);
-  map_chip_to_uarch(arch);
+  map_chip_to_uarch_intel(arch);
 
   return arch;
 }