diff --git a/test/e2e/scheduling/nvidia-gpus.go b/test/e2e/scheduling/nvidia-gpus.go
index 3d5c8643bf5..2e6b962f4c9 100644
--- a/test/e2e/scheduling/nvidia-gpus.go
+++ b/test/e2e/scheduling/nvidia-gpus.go
@@ -191,6 +191,7 @@ func testNvidiaGPUs(f *framework.Framework) {
 	// Wait for all pods to succeed
 	for _, pod := range podList {
 		f.PodClient().WaitForSuccess(pod.Name, 5*time.Minute)
+		logContainers(f, pod)
 	}
 
 	e2elog.Logf("Stopping ResourceUsageGather")
@@ -201,6 +202,14 @@ func testNvidiaGPUs(f *framework.Framework) {
 	framework.ExpectNoError(err, "getting resource usage summary")
 }
 
+func logContainers(f *framework.Framework, pod *v1.Pod) {
+	for _, container := range pod.Spec.Containers {
+		logs, err := e2epod.GetPodLogs(f.ClientSet, f.Namespace.Name, pod.Name, container.Name)
+		framework.ExpectNoError(err, "Should be able to get container logs for container: %s", container.Name)
+		e2elog.Logf("Got container logs for %s:\n%v", container.Name, logs)
+	}
+}
+
 var _ = SIGDescribe("[Feature:GPUDevicePlugin]", func() {
 	f := framework.NewDefaultFramework("device-plugin-gpus")
 	ginkgo.It("run Nvidia GPU Device Plugin tests", func() {
diff --git a/test/images/cuda-vector-add/Dockerfile b/test/images/cuda-vector-add/Dockerfile
index c7b9cfdc31e..968626bf424 100644
--- a/test/images/cuda-vector-add/Dockerfile
+++ b/test/images/cuda-vector-add/Dockerfile
@@ -23,4 +23,4 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
 WORKDIR /usr/local/cuda/samples/0_Simple/vectorAdd
 RUN make
 
-CMD ./vectorAdd
+CMD nvidia-smi && ./vectorAdd