DarthAffe/stable-diffusion.cpp
1
0 Fork 0
mirror of https://github.com/leejet/stable-diffusion.cpp.git synced 2026-11-11 03:24:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat: add --fa option (#1242)

Browse Source
This commit is contained in:
leejet authored and GitHub committed 2026-02-01 21:44:54 +08:00
1 parent c252e03c6b
commit f957fa3d2a
10 files changed
+88 -23
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

examples/cli/README.md
+2 -1
View File
@@ -52,7 +52,8 @@ Context Options:
--control-net-cpu keep controlnet in cpu (for low vram)
--clip-on-cpu keep clip in cpu (for low vram)
--vae-on-cpu keep vae in cpu (for low vram)
--diffusion-fa use flash attention in the diffusion model
--fa use flash attention
--diffusion-fa use flash attention in the diffusion model only
--diffusion-conv-direct use ggml_conv2d_direct in the diffusion model
--vae-conv-direct use ggml_conv2d_direct in the vae model
--circular enable circular padding for convolutions
examples/common/common.hpp
+8 -1
View File
@@ -457,6 +457,7 @@ struct SDContextParams {
bool control_net_cpu = false;
bool clip_on_cpu = false;
bool vae_on_cpu = false;
bool flash_attn = false;
bool diffusion_flash_attn = false;
bool diffusion_conv_direct = false;
bool vae_conv_direct = false;
@@ -615,9 +616,13 @@ struct SDContextParams {
"--vae-on-cpu",
"keep vae in cpu (for low vram)",
true, &vae_on_cpu},
{"",
"--fa",
"use flash attention",
true, &flash_attn},
{"",
"--diffusion-fa",
"use flash attention in the diffusion model",
"use flash attention in the diffusion model only",
true, &diffusion_flash_attn},
{"",
"--diffusion-conv-direct",
@@ -904,6 +909,7 @@ struct SDContextParams {
<< " control_net_cpu: " << (control_net_cpu ? "true" : "false") << ",\n"
<< " clip_on_cpu: " << (clip_on_cpu ? "true" : "false") << ",\n"
<< " vae_on_cpu: " << (vae_on_cpu ? "true" : "false") << ",\n"
<< " flash_attn: " << (flash_attn ? "true" : "false") << ",\n"
<< " diffusion_flash_attn: " << (diffusion_flash_attn ? "true" : "false") << ",\n"
<< " diffusion_conv_direct: " << (diffusion_conv_direct ? "true" : "false") << ",\n"
<< " vae_conv_direct: " << (vae_conv_direct ? "true" : "false") << ",\n"
@@ -968,6 +974,7 @@ struct SDContextParams {
clip_on_cpu,
control_net_cpu,
vae_on_cpu,
flash_attn,
diffusion_flash_attn,
taesd_preview,
diffusion_conv_direct,
examples/server/README.md
+2 -1
View File
@@ -44,7 +44,8 @@ Context Options:
--clip-on-cpu keep clip in cpu (for low vram)
--vae-on-cpu keep vae in cpu (for low vram)
--mmap whether to memory-map model
--diffusion-fa use flash attention in the diffusion model
--fa use flash attention
--diffusion-fa use flash attention in the diffusion model only
--diffusion-conv-direct use ggml_conv2d_direct in the diffusion model
--vae-conv-direct use ggml_conv2d_direct in the vae model
--circular enable circular padding for convolutions