upstream-archive/stable-diffusion.cpp
1
0
Fork 0
mirror of https://github.com/leejet/stable-diffusion.cpp.git synced 2026-10-02 02:48:04 -05:00
Code Issues Packages Projects Releases Wiki Activity

feat: add native CUDA SageAttention support (#2005)

Browse Source
This commit is contained in:
leejet
2026-09-21 01:16:59 +08:00
committed by GitHub
parent 15f335daa5
commit 187b2561ea
13 changed files with 176 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/core/ggml_runner.cpp
+2 -1
View File
@@ -25,7 +25,7 @@ ggml_tensor* ggml_ext_attention_ext(GGMLRunnerContext* ctx,
if (ctx->attn_scale > 0.f) {
kv_scale = ctx->attn_scale;
}
return ggml_ext_attention_ext(ctx->ggml_ctx, ctx->backend, q, k, v, n_head, mask, skip_reshape, flash_attn, kv_scale);
return ggml_ext_attention_ext(ctx->ggml_ctx, ctx->backend, q, k, v, n_head, mask, skip_reshape, flash_attn, kv_scale, ctx->sage_attn_enabled);
}
void GGMLRunner::alloc_params_ctx() {
@@ -520,6 +520,7 @@ GGMLRunnerContext GGMLRunner::get_context() {
runner_ctx.ggml_ctx = compute_ctx;
runner_ctx.backend = runtime_backend;
runner_ctx.flash_attn_enabled = flash_attn_enabled;
runner_ctx.sage_attn_enabled = sage_attn_enabled;
runner_ctx.linear_scale = linear_scale;
runner_ctx.attn_scale = attn_scale;
runner_ctx.conv2d_direct_enabled = conv2d_direct_enabled;