llama.cpp
7c7836d9
- Vulkan Shader Refactor, Memory Debugging Option (#7947)
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Hide Minimap (CTRL+M)
Commit
357 days ago
Vulkan Shader Refactor, Memory Debugging Option (#7947) * Refactor shaders, extract GLSL code from ggml_vk_generate_shaders.py into vulkan-shaders directory * Improve debug log code * Add memory debug output option * Fix flake8 * Fix unnecessary high llama-3 VRAM use
References
#7947 - Vulkan Shader Refactor, Memory Debugging Option
Author
0cc4m
Parents
0c7b3595
Files
54
CMakeLists.txt
Makefile
ggml-vulkan-shaders.hpp
ggml-vulkan.cpp
ggml_vk_generate_shaders.py
vulkan-shaders
add.comp
argsort.comp
clamp.comp
copy.comp
dequant_f32.comp
dequant_funcs.comp
dequant_head.comp
dequant_q2_k.comp
dequant_q3_k.comp
dequant_q4_0.comp
dequant_q4_1.comp
dequant_q4_k.comp
dequant_q5_0.comp
dequant_q5_1.comp
dequant_q5_k.comp
dequant_q6_k.comp
dequant_q8_0.comp
diag_mask_inf.comp
div.comp
gelu.comp
generic_binary_head.comp
generic_head.comp
generic_unary_head.comp
get_rows.comp
get_rows_quant.comp
mul.comp
mul_mat_split_k_reduce.comp
mul_mat_vec.comp
mul_mat_vec_base.comp
mul_mat_vec_nc.comp
mul_mat_vec_p021.comp
mul_mat_vec_q2_k.comp
mul_mat_vec_q3_k.comp
mul_mat_vec_q4_k.comp
mul_mat_vec_q5_k.comp
mul_mat_vec_q6_k.comp
mul_mm.comp
norm.comp
relu.comp
rms_norm.comp
rope_head.comp
rope_neox.comp
rope_norm.comp
scale.comp
silu.comp
soft_max.comp
square.comp
sum_rows.comp
types.comp
Loading