diff --git a/llama.cpp b/llama.cpp
index f48149209..85c0bd9f0 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -1649,7 +1649,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s
             new_data = work.addr;
             std::vector<int64_t> hist_cur(1 << 4, 0);
 
-            const int chunk_size = 32 * 512;
+            constexpr int chunk_size = 32 * 512;
             const int nchunk = (nelements + chunk_size - 1)/chunk_size;
             const int nthread_use = nthread > 1 ? std::max(1, std::min(nthread, nchunk)) : 1;
             if (nthread_use < 2) {