From 80dfe2ba498e6672a6242b168d9b033a12fd05a5 Mon Sep 17 00:00:00 2001
From: Concedo <39025047+LostRuins@users.noreply.github.com>
Date: Sun, 8 Oct 2023 15:51:18 +0800
Subject: [PATCH] blasthreads should apply for any thread count below 32

---
 llama.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/llama.cpp b/llama.cpp
index a5bc5e53c..462e985ca 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -4972,7 +4972,7 @@ static int llama_decode_internal(
 
     GGML_ASSERT(n_tokens <= n_batch);
 
-    int n_threads = n_tokens == 1 ? cparams.n_threads : cparams.n_threads_batch;
+    int n_threads = n_tokens < 32 ? cparams.n_threads : cparams.n_threads_batch;
     GGML_ASSERT((!batch.token && batch.embd) || (batch.token && !batch.embd)); // NOLINT
 
     const int64_t t_start_us = ggml_time_us();