diff --git a/llama.cpp b/llama.cpp
index 582e82260ea85..494b7a4e33a1d 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -8001,11 +8001,32 @@ void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * can
         auto comp = [](const llama_token_data & a, const llama_token_data & b) {
             return a.logit > b.logit;
         };
-        if (k == (int) candidates->size) {
-            std::sort(candidates->data, candidates->data + candidates->size, comp);
-        } else {
-            std::partial_sort(candidates->data, candidates->data + k, candidates->data + candidates->size, comp);
+        constexpr int   nbuckets    = 100;
+        constexpr float bucket_low  = -10.0f;
+        constexpr float bucket_high =  10.0f;
+
+        std::vector<llama_token_data> buckets[nbuckets];
+
+        for (size_t i = 0; i < candidates->size; ++i) {
+            const float val = candidates->data[i].logit;
+            int ib = nbuckets * (val - bucket_low) / (bucket_high - bucket_low);
+            ib = std::max(ib, 0);
+            ib = std::min(ib, nbuckets-1);
+            buckets[ib].push_back(candidates->data[i]);
+        }
+
+        int nsorted = 0;
+        for (int ib = nbuckets-1; ib >= 0; --ib) {
+            std::sort(buckets[ib].begin(), buckets[ib].end(), comp);
+            memcpy(candidates->data + nsorted, buckets[ib].data(), buckets[ib].size()*sizeof(llama_token_data));
+
+            nsorted += buckets[ib].size();
+
+            if (nsorted >= k) {
+                break;
+            }
         }
+
         candidates->sorted = true;
     }
     candidates->size = k;