From ddd960669e41fda6c2e68edca11ff016cc79c76b Mon Sep 17 00:00:00 2001
From: ravel7524 <58877666+ravel7524@users.noreply.github.com>
Date: Wed, 5 Aug 2026 15:00:23 -0400
Subject: [PATCH] fix: update mmq_use_routed_moe_ncols_picker to include NVIDIA
 + Volta support

---
 ggml/src/ggml-cuda/mmq.cuh | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/ggml/src/ggml-cuda/mmq.cuh b/ggml/src/ggml-cuda/mmq.cuh
index 76b19d1bfb..b235379dfb 100644
--- a/ggml/src/ggml-cuda/mmq.cuh
+++ b/ggml/src/ggml-cuda/mmq.cuh
@@ -1473,7 +1473,8 @@ static void launch_mul_mat_q(ggml_backend_cuda_context & ctx, const mmq_args & a
 }
 
 static bool mmq_use_routed_moe_ncols_picker(const int cc) {
-    return GGML_CUDA_CC_IS_CDNA(cc) ||
+    return (GGML_CUDA_CC_IS_NVIDIA(cc) && cc >= GGML_CUDA_CC_VOLTA) ||
+           GGML_CUDA_CC_IS_CDNA(cc) ||
            GGML_CUDA_CC_IS_RDNA2(cc) ||
            GGML_CUDA_CC_IS_RDNA3(cc) ||
            GGML_CUDA_CC_IS_RDNA4(cc);