CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-04 09:32:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

Vulkan: RTE rounding for cpy to quant (#12480)

Browse Source 
* Vulkan: RTE rounding for cpy to quant

Co-Authored-By: Jeff Bolz <jbolz@nvidia.com>

* remove trailing whitespace

* avoid duplicating pipeline_cpy_f32_quant

* fix copypasting issue

* remove duplicated code

---------

Co-authored-by: Jeff Bolz <jbolz@nvidia.com>
This commit is contained in:
stduhpf
2025-03-21 20:34:50 +01:00
committed by GitHub
parent 30c42ef5cb
commit 4375415b4a
3 changed files with 21 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
ggml/src/ggml-vulkan/vulkan-shaders/copy_to_quant.comp
View File

@@ -1,5 +1,10 @@
#version 450
#if RTE16
#extension GL_EXT_spirv_intrinsics : enable
spirv_execution_mode(capabilities = [4467], 4462, 16); // RoundingModeRTE, 16 bits
#endif // RTE16
#include "types.comp"
#include "generic_unary_head.comp"

1
ggml/src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp
View File

@@ -445,6 +445,7 @@ void process_shaders() {
for (std::string t : {"q4_0", "q4_1", "q5_0", "q5_1", "q8_0", "iq4_nl"}) {
string_to_spv("cpy_f32_" + t, "copy_to_quant.comp", {{"DATA_A_" + to_uppercase(t), "1"}, {"D_TYPE", "float"}, {"FLOAT_TYPE", "float"}});
string_to_spv("cpy_f32_" + t + "_rte", "copy_to_quant.comp", {{"DATA_A_" + to_uppercase(t), "1"}, {"D_TYPE", "float"}, {"FLOAT_TYPE", "float"}, {"RTE16", "1"}});
string_to_spv("cpy_" + t + "_f32", "copy_from_quant.comp", {{"DATA_A_" + to_uppercase(t), "1"}, {"D_TYPE", "float"}, {"FLOAT_TYPE", "float"}});
}