tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-28 04:15:21 +00:00
Code Issues Packages Projects Releases Wiki Activity

ggml: Add POOL2D OP for GPU acceleration to the Vulkan backend in the MobileVLM model. (#9763)

Browse Source 
* ggml: Add POOL2D OP for GPU ACC to the Vulkan.

- The MobileVLM model now supports inference acceleration through GPU by utilizing the Vulkan backend.
- A GGML_OP_POOL_2D shader has been added. (Pooling)
- The encoding performance of the CLIP model improved from 2.8s on the CPU to 0.7s on the GPU.

Signed-off-by: Changyeon Kim <cyzero.kim@samsung.com>

* [fix] Correct the incorrect order of the parameters.

fix casting to int.

Signed-off-by: Changyeon Kim <cyzero.kim@samsung.com>

---------

Signed-off-by: Changyeon Kim <cyzero.kim@samsung.com>
This commit is contained in:
Changyeon Kim
2024-10-29 17:52:56 +09:00
committed by GitHub
parent 8d8ff71536
commit 8f275a7c45
3 changed files with 150 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
ggml/src/vulkan-shaders/vulkan-shaders-gen.cpp
View File

@ -493,6 +493,10 @@ void process_shaders(std::vector<std::future<void>>& tasks) {
tasks.push_back(std::async(std::launch::async, [=] {
string_to_spv("timestep_embedding_f32", "timestep_embedding.comp", merge_maps(base_dict, {{"A_TYPE", "float"}, {"D_TYPE", "float"}}));
}));
tasks.push_back(std::async(std::launch::async, [=] {
string_to_spv("pool2d_f32", "pool2d.comp", merge_maps(base_dict, {{"A_TYPE", "float"}, {"D_TYPE", "float"}}));
}));
}
void write_output_files() {