diff --git a/extra/sync-ggml.last b/extra/sync-ggml.last
index fe7f320..0e11d4a 100644
--- a/extra/sync-ggml.last
+++ b/extra/sync-ggml.last
@@ -1 +1 @@
-f96711108d55bdbbd277e6be07204dce6a94fb93
+3eace58911ea8d2cf35defdc59848d99b91a57f5
diff --git a/ggml-cuda.cu b/ggml-cuda.cu
index 10c2161..7578d21 100644
--- a/ggml-cuda.cu
+++ b/ggml-cuda.cu
@@ -9689,8 +9689,8 @@ static void ggml_backend_cuda_buffer_set_tensor(ggml_backend_buffer_t buffer, gg
 
     ggml_cuda_set_device(ctx->device);
     CUDA_CHECK(cudaDeviceSynchronize());
-
     CUDA_CHECK(cudaMemcpy((char *)tensor->data + offset, data, size, cudaMemcpyHostToDevice));
+    CUDA_CHECK(cudaDeviceSynchronize());
 }
 
 static void ggml_backend_cuda_buffer_get_tensor(ggml_backend_buffer_t buffer, const ggml_tensor * tensor, void * data, size_t offset, size_t size) {