DarthAffe/stable-diffusion.cpp
1
0 Fork 0
mirror of https://github.com/leejet/stable-diffusion.cpp.git synced 2026-11-11 11:26:34 +00:00
Code Issues Packages Projects Releases Wiki Activity

perf: make ggml_conv_2d faster

Browse Source
This commit is contained in:
leejet committed 2023-08-26 17:08:59 +08:00
1 parent 008d80a0b1
commit d765b95ed1
2 files changed
+13 -1
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

ggml
+1 -1
Submodule ggml updated: d73671e5f1...3a0b87bde9.
stable-diffusion.cpp
+12
View File
@@ -3256,6 +3256,10 @@ class StableDiffusionGGML {
} }
} }
} }
#ifdef GGML_PERF
ggml_graph_print(&diffusion_graph);
#endif
int64_t t1 = ggml_time_ms(); int64_t t1 = ggml_time_ms();
LOG_INFO("step %d sampling completed, taking %.2fs", i + 1, (t1 - t0) * 1.0f / 1000); LOG_INFO("step %d sampling completed, taking %.2fs", i + 1, (t1 - t0) * 1.0f / 1000);
LOG_DEBUG("diffusion graph use %.2fMB runtime memory: static %.2fMB, dynamic %.2fMB", LOG_DEBUG("diffusion graph use %.2fMB runtime memory: static %.2fMB, dynamic %.2fMB",
@@ -3345,6 +3349,10 @@ class StableDiffusionGGML {
int64_t t0 = ggml_time_ms(); int64_t t0 = ggml_time_ms();
ggml_graph_compute_with_ctx(ctx, &vae_graph, n_threads); ggml_graph_compute_with_ctx(ctx, &vae_graph, n_threads);
int64_t t1 = ggml_time_ms(); int64_t t1 = ggml_time_ms();
#ifdef GGML_PERF
ggml_graph_print(&vae_graph);
#endif
LOG_DEBUG("computing vae graph completed, taking %.2fs", (t1 - t0) * 1.0f / 1000); LOG_DEBUG("computing vae graph completed, taking %.2fs", (t1 - t0) * 1.0f / 1000);
result = ggml_dup_tensor(res_ctx, moments); result = ggml_dup_tensor(res_ctx, moments);
@@ -3470,6 +3478,10 @@ class StableDiffusionGGML {
int64_t t0 = ggml_time_ms(); int64_t t0 = ggml_time_ms();
ggml_graph_compute_with_ctx(ctx, &vae_graph, n_threads); ggml_graph_compute_with_ctx(ctx, &vae_graph, n_threads);
int64_t t1 = ggml_time_ms(); int64_t t1 = ggml_time_ms();
#ifdef GGML_PERF
ggml_graph_print(&vae_graph);
#endif
LOG_DEBUG("computing vae graph completed, taking %.2fs", (t1 - t0) * 1.0f / 1000); LOG_DEBUG("computing vae graph completed, taking %.2fs", (t1 - t0) * 1.0f / 1000);
result_img = ggml_dup_tensor(res_ctx, img); result_img = ggml_dup_tensor(res_ctx, img);