summaryrefslogtreecommitdiff
path: root/examples/benchmark/benchmark-matmult.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'examples/benchmark/benchmark-matmult.cpp')
-rw-r--r--examples/benchmark/benchmark-matmult.cpp6
1 files changed, 2 insertions, 4 deletions
diff --git a/examples/benchmark/benchmark-matmult.cpp b/examples/benchmark/benchmark-matmult.cpp
index e89f3de2..47cb16c6 100644
--- a/examples/benchmark/benchmark-matmult.cpp
+++ b/examples/benchmark/benchmark-matmult.cpp
@@ -189,12 +189,10 @@ int main(int argc, char ** argv) {
int32_t nelements = sizex*sizey;
- std::vector<int64_t> hist_cur(1 << 4, 0);
-
// Set up a the benchmark matrices
// printf("Creating new tensor q11 & Running quantize\n");
struct ggml_tensor * q11 = ggml_new_tensor_2d(ctx, qtype, sizex, sizey);
- ggml_quantize_chunk(qtype, (const float *) m11->data, q11->data, 0, nelements/m11->ne[0], m11->ne[0], hist_cur.data(), nullptr);
+ ggml_quantize_chunk(qtype, (const float *) m11->data, q11->data, 0, nelements/m11->ne[0], m11->ne[0], nullptr);
// Set up a the compute graph
// printf("Creating new tensor q31\n");
@@ -207,7 +205,7 @@ int main(int argc, char ** argv) {
// Set up a second graph computation to make sure we override the CPU cache lines
// printf("Creating new tensor q12 & Running quantize\n");
struct ggml_tensor * q12 = ggml_new_tensor_2d(ctx, qtype, sizex, sizey);
- ggml_quantize_chunk(qtype, (const float *) m12->data, q12->data, 0, nelements/m12->ne[0], m12->ne[0], hist_cur.data(), nullptr);
+ ggml_quantize_chunk(qtype, (const float *) m12->data, q12->data, 0, nelements/m12->ne[0], m12->ne[0], nullptr);
// printf("Creating new tensor q32\n");
struct ggml_tensor * q32 = ggml_mul_mat(ctx, q12, m2);