DarthAffe/stable-diffusion.cpp
1
0
Fork 0
mirror of https://github.com/leejet/stable-diffusion.cpp.git synced 2026-06-23 22:56:42 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: workaround for Ernie with Vulkan and Flash Attention (#1680)

Browse Source
This commit is contained in:
Daniele 2026-06-21 18:21:38 +02:00 committed by GitHub
parent e8e012eef2
commit e9e952462f
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/model/diffusion/ernie_image.hpp
View File

@ -162,6 +162,8 @@ namespace ErnieImage {
int64_t S = x->ne[1]; int64_t S = x->ne[1];
int64_t N = x->ne[2]; int64_t N = x->ne[2];
float scale = (sd_backend_is(ctx->backend, "Vulkan") && ctx->flash_attn_enabled) ? 1.0f / 32.0f : 1.0f;
auto q = to_q->forward(ctx, x); auto q = to_q->forward(ctx, x);
auto k = to_k->forward(ctx, x); auto k = to_k->forward(ctx, x);
auto v = to_v->forward(ctx, x); auto v = to_v->forward(ctx, x);
@ -182,7 +184,7 @@ namespace ErnieImage {
k = ggml_cont(ctx->ggml_ctx, ggml_permute(ctx->ggml_ctx, k, 0, 2, 1, 3)); // [N, heads, S, head_dim] k = ggml_cont(ctx->ggml_ctx, ggml_permute(ctx->ggml_ctx, k, 0, 2, 1, 3)); // [N, heads, S, head_dim]
k = ggml_reshape_3d(ctx->ggml_ctx, k, k->ne[0], k->ne[1], k->ne[2] * k->ne[3]); k = ggml_reshape_3d(ctx->ggml_ctx, k, k->ne[0], k->ne[1], k->ne[2] * k->ne[3]);
x = ggml_ext_attention_ext(ctx->ggml_ctx, ctx->backend, q, k, v, num_heads, attention_mask, true, ctx->flash_attn_enabled); // [N, S, hidden_size] x = ggml_ext_attention_ext(ctx->ggml_ctx, ctx->backend, q, k, v, num_heads, attention_mask, true, ctx->flash_attn_enabled, scale); // [N, S, hidden_size]
x = to_out_0->forward(ctx, x); x = to_out_0->forward(ctx, x);
return x; return x;
} }