[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
This commit is contained in:
@@ -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_t = get_peak_performance_t(gpu);
|
||||
gpu->peak_performance = get_peak_performance_cuda(gpu);
|
||||
gpu->peak_performance_tcu = get_peak_performance_tcu(gpu);
|
||||
|
||||
return gpu;
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user