schihei/llama2.c
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #128 from richinseattle/patch-1

Browse Source 
update openmp pragmas for MSVC compatibility
This commit is contained in:
Andrej
2023-07-26 22:20:49 -07:00
committed by GitHub
parent 34cce6a6b5 539dc73196
commit 5c55d59325
1 changed files with 9 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
run.c
+9 -5
View File
@@ -13,9 +13,11 @@ $ ./run
#include <time.h> #include <time.h>
#include <math.h> #include <math.h>
#include <string.h> #include <string.h>
#include <unistd.h>
#include <fcntl.h> #include <fcntl.h>
#ifndef _WIN32
#include <unistd.h>
#include <sys/mman.h> #include <sys/mman.h>
#endif
// ---------------------------------------------------------------------------- // ----------------------------------------------------------------------------
// Transformer and RunState structs, and related memory management // Transformer and RunState structs, and related memory management
@@ -190,8 +192,9 @@ void softmax(float* x, int size) {
void matmul(float* xout, float* x, float* w, int n, int d) { void matmul(float* xout, float* x, float* w, int n, int d) {
// W (d,n) @ x (n,) -> xout (d,) // W (d,n) @ x (n,) -> xout (d,)
#pragma omp parallel for int i;
for (int i = 0; i < d; i++) { #pragma omp parallel for private(i)
for (i = 0; i < d; i++) {
float val = 0.0f; float val = 0.0f;
for (int j = 0; j < n; j++) { for (int j = 0; j < n; j++) {
val += w[i * n + j] * x[j]; val += w[i * n + j] * x[j];
@@ -255,8 +258,9 @@ void transformer(int token, int pos, Config* p, RunState* s, TransformerWeights*
memcpy(value_cache_row, s->v, dim*sizeof(*value_cache_row)); memcpy(value_cache_row, s->v, dim*sizeof(*value_cache_row));
// multihead attention. iterate over all heads // multihead attention. iterate over all heads
#pragma omp parallel for int h;
for (int h = 0; h < p->n_heads; h++) { #pragma omp parallel for private(h)
for (h = 0; h < p->n_heads; h++) {
// get the query vector for this head // get the query vector for this head
float* q = s->q + h * head_size; float* q = s->q + h * head_size;
// attention scores for this head // attention scores for this head