From ea2e5ed1e924b39093b0618a531942b5ba838f5a Mon Sep 17 00:00:00 2001
From: Concedo <39025047+LostRuins@users.noreply.github.com>
Date: Fri, 9 May 2025 18:30:11 +0800
Subject: [PATCH] mmq debug log

---
 gpttype_adapter.cpp | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/gpttype_adapter.cpp b/gpttype_adapter.cpp
index 594439ea5..e56c1802b 100644
--- a/gpttype_adapter.cpp
+++ b/gpttype_adapter.cpp
@@ -2168,8 +2168,9 @@ ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in
         #if defined(GGML_USE_CUDA)
         if(cu_parseinfo_maindevice>0)
         {
-            printf("CUBLAS: Set main device to %d\n",cu_parseinfo_maindevice);
+            printf("CUDA: Set main device to %d\n",cu_parseinfo_maindevice);
         }
+        printf("CUDA MMQ: %s\n",(inputs.use_mmq?"True":"False"));
         ggml_cuda_set_mul_mat_q(inputs.use_mmq);
         #endif
         if((file_format_meta.model_architecture == GGUFArch::ARCH_QWEN2 || file_format_meta.model_architecture == GGUFArch::ARCH_QWEN2VL) && !kcpp_data->flash_attn)