llama.cpp/ggml/src/vulkan-shaders/rope_neox.comp
Georgi Gerganov f3f65429c4
llama : reorganize source code + improve CMake (#8006)
* scripts : update sync [no ci]

* files : relocate [no ci]

* ci : disable kompute build [no ci]

* cmake : fixes [no ci]

* server : fix mingw build

ggml-ci

* cmake : minor [no ci]

* cmake : link math library [no ci]

* cmake : build normal ggml library (not object library) [no ci]

* cmake : fix kompute build

ggml-ci

* make,cmake : fix LLAMA_CUDA + replace GGML_CDEF_PRIVATE

ggml-ci

* move public backend headers to the public include directory (#8122)

* move public backend headers to the public include directory

* nix test

* spm : fix metal header

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

* scripts : fix sync paths [no ci]

* scripts : sync ggml-blas.h [no ci]

---------

Co-authored-by: slaren <slarengh@gmail.com>
2024-06-26 18:33:02 +03:00

38 lines
929 B
Plaintext

#version 450
#include "rope_head.comp"
void main() {
const uint col = gl_GlobalInvocationID.y * 2;
const uint row = gl_GlobalInvocationID.x;
if (col >= p.ncols) {
return;
}
if (col >= p.n_dims) {
const uint i = row*p.ncols + col;
data_d[i + 0] = data_a[i + 0];
data_d[i + 1] = data_a[i + 1];
return;
}
const uint i = row*p.ncols + col/2;
const uint i2 = row/p.p_delta_rows;
const float theta_base = data_pos[i2] * pow(p.theta_scale, col/2.0f);
const float freq_factor = p.has_ff != 0 ? data_ff[col/2] : 1.0f;
float cos_theta, sin_theta;
rope_yarn(theta_base / freq_factor, col, cos_theta, sin_theta);
const float x0 = float(data_a[i + 0]);
const float x1 = float(data_a[i + p.n_dims/2]);
data_d[i + 0] = D_TYPE(x0*cos_theta - x1*sin_theta);
data_d[i + p.n_dims/2] = D_TYPE(x0*sin_theta + x1*cos_theta);
}