DarthAffe/stable-diffusion.cpp
1
0 Fork 0
mirror of https://github.com/leejet/stable-diffusion.cpp.git synced 2026-11-12 11:56:36 +00:00
Code Issues Packages Projects Releases Wiki Activity

perf: optimize H3 inference and segmented weight prefetch (#2240)

Browse Source
This commit is contained in:
leejet authored and GitHub committed 2026-10-30 00:38:14 +08:00
1 parent e7b7ec6d6f
commit 37bc88b5e9
17 files changed
+367 -210
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

src/model_manager.h
+5
View File
@@ -76,6 +76,7 @@ private:
ggml_backend_t compute_backend = nullptr;
ggml_backend_t transfer_backend = nullptr;
ggml_backend_event_t event = nullptr;
bool submitted = false;
std::vector<std::unique_ptr<ComputeStagingBlock>> staging_blocks;
};
@@ -292,6 +293,10 @@ public:
WeightPrefetchResult prefetch_params(
uintptr_t owner_id,
const std::vector<ggml_tensor*>& tensors) override;
WeightPrefetchResult prepare_prefetch_params(
uintptr_t owner_id,
const std::vector<ggml_tensor*>& tensors) override;
void submit_prefetched_params(uintptr_t owner_id) override;
bool activate_prefetched_params(uintptr_t owner_id,
const std::vector<ggml_tensor*>& tensors) override;
void clear_prefetched_params(uintptr_t owner_id) override;