CUDA: add unused vars to mmvf and mmvq (#16807)

am17an · web-flow · commit 463bbf20bfbe · 2025-10-28T10:31:21.000+08:00
diff --git a/ggml/src/ggml-cuda/mmvf.cu b/ggml/src/ggml-cuda/mmvf.cu
@@ -343,6 +343,10 @@ static __global__ void mul_mat_vec_f(
     }
 
     dst[tid*stride_col_dst + row] = value;
+
+    if constexpr (!has_fusion) {
+        GGML_UNUSED_VARS(use_gate, use_bias, use_gate_bias, glu_op, gate_x, x_bias, gate_bias, sumf_gate);
+    }
 }
 
 template<typename T, typename type_acc, int ncols_dst, int block_size>
diff --git a/ggml/src/ggml-cuda/mmvq.cu b/ggml/src/ggml-cuda/mmvq.cu
@@ -310,6 +310,10 @@ static __global__ void mul_mat_vec_q(
             dst[j*stride_col_dst + threadIdx.x] = result;
         }
     }
+
+    if constexpr (!has_fusion) {
+        GGML_UNUSED_VARS(use_gate, use_bias, use_gate_bias, active_glu, gate_bias, x_bias, tmp_gate);
+    }
 }
 
 static std::pair<dim3, dim3> calc_launch_params(

Original file line number	Diff line number	Diff line change
`@@ -343,6 +343,10 @@ static __global__ void mul_mat_vec_f(`
`343`	`343`	`}`
`344`	`344`
`345`	`345`	`dst[tid*stride_col_dst + row] = value;`
	`346`	`+`
	`347`	`+ if constexpr (!has_fusion) {`
	`348`	`+ GGML_UNUSED_VARS(use_gate, use_bias, use_gate_bias, glu_op, gate_x, x_bias, gate_bias, sumf_gate);`
	`349`	`+ }`
`346`	`350`	`}`
`347`	`351`
`348`	`352`	`template<typename T, typename type_acc, int ncols_dst, int block_size>`
Original file line number	Diff line number	Diff line change
`@@ -310,6 +310,10 @@ static __global__ void mul_mat_vec_q(`
`310`	`310`	`dst[j*stride_col_dst + threadIdx.x] = result;`
`311`	`311`	`}`
`312`	`312`	`}`
	`313`	`+`
	`314`	`+ if constexpr (!has_fusion) {`
	`315`	`+ GGML_UNUSED_VARS(use_gate, use_bias, use_gate_bias, active_glu, gate_bias, x_bias, tmp_gate);`
	`316`	`+ }`
`313`	`317`	`}`
`314`	`318`
`315`	`319`	`static std::pair<dim3, dim3> calc_launch_params(`