vrr/prima.cpp
2
0
Fork 0
mirror of https://github.com/Lizonghang/prima.cpp.git synced 2025-09-06 13:49:02 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

add gpu support in device_memory_access_delay

Browse source
This commit is contained in:
Zonghang Li 2024-11-29 21:56:01 +04:00
parent 6f54a12c7d
commit 81fd77033e
1 changed files with 25 additions and 13 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

38
common/profiler.cpp
View file

@ -517,7 +517,7 @@ static size_t get_default_readahead_size() {
file.close(); file.close();
return read_ahead_kb * 1024; // convert to bytes return read_ahead_kb * 1024; // convert to bytes
} else { } else {
std::cerr << "Unable to open: " << sysfs_path << "\n"; LOG_INF("Unable to open: %s\n", sysfs_path.c_str());
return 0; return 0;
} }
#elif __APPLE__ #elif __APPLE__
@ -880,6 +880,8 @@ static float device_compute_delay(struct device_info & dev_info, int n_layers, c
total_latency += gpu_latency_per_layer * n_gpu_layers; total_latency += gpu_latency_per_layer * n_gpu_layers;
total_latency += cpu_latency_per_layer * (n_layers - n_gpu_layers); total_latency += cpu_latency_per_layer * (n_layers - n_gpu_layers);
#else #else
(void)n_gpu_layers;
(void)gpu_latency_per_layer;
total_latency += cpu_latency_per_layer * n_layers; total_latency += cpu_latency_per_layer * n_layers;
#endif #endif
@ -897,30 +899,40 @@ static float device_compute_delay(struct device_info & dev_info, int n_layers, c
} }
// estimate the memory access delay, except for the input embedding because it has been considered in n_flops.inp_embd_ms // estimate the memory access delay, except for the input embedding because it has been considered in n_flops.inp_embd_ms
static float device_memory_access_delay(struct device_info & dev_info, int n_layers) { static float device_memory_access_delay(struct device_info & dev_info, const struct llama_context_params cparams, int n_layers) {
struct model_params n_params = dev_info.model_params; struct model_params n_params = dev_info.model_params;
int64_t total_bytes = int64_t layer_bytes =
n_params.layer_f32 * 4 + n_params.layer_f32 * 4 +
n_params.layer_f16 * 2 + n_params.layer_f16 * 2 +
n_params.layer_q4k / 2 + n_params.layer_q4k / 2 +
n_params.layer_q6k * 3 / 8 + n_params.layer_q6k * 3 / 8 +
n_params.layer_q80; n_params.layer_q80;
total_bytes *= n_layers; int64_t output_bytes =
n_params.output_f32 * 4 +
total_bytes += n_params.output_f32 * 4 +
n_params.output_f16 * 2 + n_params.output_f16 * 2 +
n_params.output_q4k / 2 + n_params.output_q4k / 2 +
n_params.output_q6k * 3 / 8 + n_params.output_q6k * 3 / 8 +
n_params.output_q80; n_params.output_q80;
#if defined(GGML_USE_CUDA) || defined(GGML_USE_METAL)
int64_t vram_bytes = layer_bytes * cparams.n_gpu_layers;
int64_t ram_bytes = layer_bytes * (n_layers - cparams.n_gpu_layers) + output_bytes;
#ifdef GGML_USE_CUDA #ifdef GGML_USE_CUDA
return (double)total_bytes / 1e6 / dev_info.gpu_props.cuda_read_vram_bw; // ms double vram_access_delay = (double)(vram_bytes) / 1e6 / dev_info.gpu_props.cuda_read_vram_bw;
#elif GGML_USE_METAL #elif GGML_USE_METAL
return (double)total_bytes / 1e6 / dev_info.gpu_props.metal_read_vram_bw; // ms double vram_access_delay = (double)(vram_bytes) / 1e6 / dev_info.gpu_props.metal_read_vram_bw;
#endif
double ram_access_delay = (double)(ram_bytes) / 1e6 / dev_info.memory.cpu_read_ram_bw;
return static_cast<float>(vram_access_delay + ram_access_delay); // ms
#else #else
return (double)total_bytes / 1e6 / dev_info.memory.cpu_read_ram_bw; // ms int64_t ram_bytes = layer_bytes * n_layers;
double ram_access_delay = (double)(ram_bytes) / 1e6 / dev_info.memory.cpu_read_ram_bw;
return static_cast<float>(ram_access_delay); // ms
#endif #endif
} }
@ -1336,10 +1348,10 @@ void device_print_props(struct device_info * dev_info_set, int n, struct llama_m
// todo: calculate for each device, not only master // todo: calculate for each device, not only master
float latency = 0.0f; float latency = 0.0f;
int n_layers = llama_model_n_layers(model); int n_layers = llama_model_n_layers (model);
latency += device_compute_delay(dev_info_set[0], n_layers, cparams); latency += device_compute_delay (dev_info_set[0], n_layers, cparams);
latency += device_memory_access_delay(dev_info_set[0], n_layers); latency += device_memory_access_delay(dev_info_set[0], cparams, n_layers);
latency += device_disk_access_delay(dev_info_set[0], model, cparams); // if physical memory is not enough, some tensor weights will be released from memory and reloaded by mmap later latency += device_disk_access_delay (dev_info_set[0], model, cparams); // if physical memory is not enough, some tensor weights will be released from memory and reloaded by mmap later
LOG_INF("| Token latency (ms) "); LOG_INF("| Token latency (ms) ");
LOG_INF("| %-10.2f ", latency); LOG_INF("| %-10.2f ", latency);