upstream-archive/stable-diffusion.cpp
1
0
Fork 0
mirror of https://github.com/leejet/stable-diffusion.cpp.git synced 2026-09-27 16:37:49 -05:00
Code Issues Packages Projects Releases Wiki Activity

perf: parallelize host tensor elementwise and broadcast ops (#1998)

Browse Source
This commit is contained in:
leejet
2026-09-19 21:46:16 +08:00
committed by GitHub
parent 275ab58e01
commit 17860c0e45
10 changed files with 361 additions and 95 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/pipeline/diffusion_engine.cpp
+4 -2
View File
@@ -863,8 +863,10 @@ bool StableDiffusionGGML::init(const sd_ctx_params_t* sd_ctx_params) {
return false;
}
}
auto configuration = std::make_unique<ModelConfig>(*sd_ctx_params);
n_threads = sd_ctx_params->n_threads;
auto configuration = std::make_unique<ModelConfig>(*sd_ctx_params);
n_threads = sd_ctx_params->n_threads;
tensor_executor = std::make_unique<sd::ParallelExecutor>(n_threads > 0 ? n_threads : sd_get_num_physical_cores());
sd::ParallelScope tensor_scope(tensor_executor.get());
enable_mmap = sd_ctx_params->enable_mmap;
disable_prefetch = sd_ctx_params->disable_prefetch;
disable_segmented_compute = sd_ctx_params->disable_segmented_compute;