From 80a9082166e71585b65611c07bd843ff31b0d269 Mon Sep 17 00:00:00 2001
From: Concedo <39025047+LostRuins@users.noreply.github.com>
Date: Sun, 3 May 2026 13:40:24 +0800
Subject: [PATCH] q5_1 kv in cuda

---
 ggml/src/ggml-cuda/fattn.cu | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ggml/src/ggml-cuda/fattn.cu b/ggml/src/ggml-cuda/fattn.cu
index 4cfe21d10..1a2cda7f6 100644
--- a/ggml/src/ggml-cuda/fattn.cu
+++ b/ggml/src/ggml-cuda/fattn.cu
@@ -409,11 +409,11 @@ static best_fattn_kernel ggml_cuda_get_best_fattn_kernel(const int device, const
             break;
         case GGML_TYPE_Q4_1:
         case GGML_TYPE_Q5_0:
-        case GGML_TYPE_Q5_1:
 #ifndef GGML_CUDA_FA_ALL_QUANTS
             return BEST_FATTN_KERNEL_NONE;
 #endif // GGML_CUDA_FA_ALL_QUANTS
         case GGML_TYPE_Q4_0:
+        case GGML_TYPE_Q5_1: //kcpp: support q5_1 kv
         case GGML_TYPE_Q8_0:
         case GGML_TYPE_BF16:
             break;