DarthAffe/stable-diffusion.cpp
1
0 Fork 0
mirror of https://github.com/leejet/stable-diffusion.cpp.git synced 2026-11-11 19:36:33 +00:00
Code Issues Packages Projects Releases Wiki Activity

refactor: split ggml extensions and move implementations to cpp files (#1945)

Browse Source
This commit is contained in:
leejet authored and GitHub committed 2026-09-07 23:34:45 +08:00
1 parent d8fb10c029
commit 31ab2b2e08
74 files changed
+4269 -3698
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

src/model/te/clip.hpp
+5 -2
View File
@@ -1,8 +1,11 @@
#ifndef __SD_MODEL_TE_CLIP_HPP__
#define __SD_MODEL_TE_CLIP_HPP__
#include "core/ggml_extend.hpp"
#include "core/ggml_extend.h"
#include "core/ggml_runner.h"
#include "core/util.h"
#include "model.h"
#include "model/common/ggml_block.hpp"
#include "tokenizers/clip_tokenizer.h"
/*================================================ FrozenCLIPEmbedder ================================================*/
@@ -572,7 +575,7 @@ struct CLIPTextModelRunner : public GGMLRunner {
auto get_graph = [&]() -> ggml_cgraph* {
return build_graph(input_ids, num_custom_embeddings, custom_embeddings_data, max_token_idx, return_pooled, clip_skip);
};
auto result = GGMLRunner::compute<float>(get_graph, n_threads, auto_runner_end);
auto result = GGMLRunner::compute(get_graph, n_threads, auto_runner_end);
if (return_pooled) {
return take_or_empty(std::move(result));
}
src/model/te/llm.hpp
+11 -5
View File
@@ -3,6 +3,7 @@
#include <algorithm>
#include <array>
#include <cinttypes>
#include <cmath>
#include <fstream>
#include <functional>
@@ -18,8 +19,13 @@
#include <utility>
#include <vector>
#include "core/ggml_extend.hpp"
#include "core/ggml_extend.h"
#include "core/ggml_extend_backend.h"
#include "core/ggml_runner.h"
#include "core/ggml_tensor_utils.h"
#include "core/util.h"
#include "json.hpp"
#include "model/common/ggml_block.hpp"
#include "model/common/rope.hpp"
#include "model_loader.h"
#include "model_manager.h"
@@ -2091,7 +2097,7 @@ namespace LLM {
out_layers,
return_all_hidden_states);
};
return restore_trailing_singleton_dims(GGMLRunner::compute<float>(get_graph, n_threads, auto_runner_end),
return restore_trailing_singleton_dims(GGMLRunner::compute(get_graph, n_threads, auto_runner_end),
input_ids.dim() + 1);
}
@@ -2175,7 +2181,7 @@ namespace LLM {
auto get_graph = [&]() -> ggml_cgraph* {
return build_encode_image_graph(image);
};
return take_or_empty(GGMLRunner::compute<float>(get_graph, n_threads, auto_runner_end));
return take_or_empty(GGMLRunner::compute(get_graph, n_threads, auto_runner_end));
}
ggml_cgraph* build_encode_image_outputs_graph(const sd::Tensor<float>& image_tensor) {
@@ -2287,7 +2293,7 @@ namespace LLM {
auto get_graph = [&]() -> ggml_cgraph* {
return build_encode_image_outputs_graph(image);
};
auto combined = take_or_empty(GGMLRunner::compute<float>(get_graph, n_threads, auto_runner_end));
auto combined = take_or_empty(GGMLRunner::compute(get_graph, n_threads, auto_runner_end));
if (combined.empty()) {
return {};
}
@@ -2313,7 +2319,7 @@ namespace LLM {
auto get_graph = [&]() -> ggml_cgraph* {
return build_encode_video_block_outputs_graph(pixel_values, grid_h, grid_w);
};
auto combined = take_or_empty(GGMLRunner::compute<float>(get_graph, n_threads, auto_runner_end));
auto combined = take_or_empty(GGMLRunner::compute(get_graph, n_threads, auto_runner_end));
if (combined.empty()) {
return {};
}
src/model/te/t5.hpp
+7 -2
View File
@@ -10,7 +10,12 @@
#include <string>
#include <unordered_map>
#include "core/ggml_extend.hpp"
#include "core/ggml_extend.h"
#include "core/ggml_extend_backend.h"
#include "core/ggml_runner.h"
#include "core/ggml_tensor_utils.h"
#include "core/util.h"
#include "model/common/ggml_block.hpp"
#include "model_loader.h"
#include "model_manager.h"
#include "tokenizers/t5_unigram_tokenizer.h"
@@ -455,7 +460,7 @@ struct T5Runner : public GGMLRunner {
auto get_graph = [&]() -> ggml_cgraph* {
return build_graph(input_ids, attention_mask);
};
return restore_trailing_singleton_dims(GGMLRunner::compute<float>(get_graph, n_threads, auto_runner_end), 3);
return restore_trailing_singleton_dims(GGMLRunner::compute(get_graph, n_threads, auto_runner_end), 3);
}
static std::vector<int> _relative_position_bucket(const std::vector<int>& relative_position,