diff --git a/.github/workflows/cron.yml b/.github/workflows/cron.yml
index 77d43d35bd..c864b4603e 100644
--- a/.github/workflows/cron.yml
+++ b/.github/workflows/cron.yml
@@ -62,7 +62,7 @@ jobs:
     if: github.repository == 'Project-MONAI/MONAI'
     strategy:
       matrix:
-        container: ["pytorch:21.02", "pytorch:21.08"]  # 21.02 for backward comp.
+        container: ["pytorch:21.02", "pytorch:21.09"]  # 21.02 for backward comp.
     container:
       image: nvcr.io/nvidia/${{ matrix.container }}-py3  # testing with the latest pytorch base image
       options: "--gpus all"
@@ -106,7 +106,7 @@ jobs:
     if: github.repository == 'Project-MONAI/MONAI'
     strategy:
       matrix:
-        container: ["pytorch:21.02", "pytorch:21.08"]  # 21.02 for backward comp.
+        container: ["pytorch:21.02", "pytorch:21.09"]  # 21.02 for backward comp.
     container:
       image: nvcr.io/nvidia/${{ matrix.container }}-py3  # testing with the latest pytorch base image
       options: "--gpus all"
@@ -204,7 +204,7 @@ jobs:
     if: github.repository == 'Project-MONAI/MONAI'
     needs: cron-gpu  # so that monai itself is verified first
     container:
-      image: nvcr.io/nvidia/pytorch:21.08-py3  # testing with the latest pytorch base image
+      image: nvcr.io/nvidia/pytorch:21.09-py3  # testing with the latest pytorch base image
       options: "--gpus all --ipc=host"
     runs-on: [self-hosted, linux, x64, common]
     steps:
diff --git a/.github/workflows/pythonapp-gpu.yml b/.github/workflows/pythonapp-gpu.yml
index edaa2487ce..8f285af859 100644
--- a/.github/workflows/pythonapp-gpu.yml
+++ b/.github/workflows/pythonapp-gpu.yml
@@ -43,9 +43,9 @@ jobs:
           - environment: PT19+CUDA114
             # we explicitly set pytorch to -h to avoid pip install error
             # https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes
-            # 21.08: 1.10.0a0+3fd9dcf
+            # 21.09: 1.10.0a0+3fd9dcf
             pytorch: "-h"
-            base: "nvcr.io/nvidia/pytorch:21.08-py3"
+            base: "nvcr.io/nvidia/pytorch:21.09-py3"
           - environment: PT19+CUDA102
             pytorch: "torch==1.9.1 torchvision==0.10.1"
             base: "nvcr.io/nvidia/cuda:10.2-devel-ubuntu18.04"
diff --git a/Dockerfile b/Dockerfile
index 77fe1f828f..ce4306c639 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -11,7 +11,7 @@
 
 # To build with a different base image
 # please run `docker build` using the `--build-arg PYTORCH_IMAGE=...` flag.
-ARG PYTORCH_IMAGE=nvcr.io/nvidia/pytorch:21.08-py3
+ARG PYTORCH_IMAGE=nvcr.io/nvidia/pytorch:21.09-py3
 FROM ${PYTORCH_IMAGE}
 
 LABEL maintainer="monai.contact@gmail.com"
diff --git a/tests/test_integration_fast_train.py b/tests/test_integration_fast_train.py
index 2cf7eee479..9fd37a0897 100644
--- a/tests/test_integration_fast_train.py
+++ b/tests/test_integration_fast_train.py
@@ -76,7 +76,7 @@ def tearDown(self):
         shutil.rmtree(self.data_dir)
 
     # test the fast training speed is as expected
-    @TimedCall(seconds=30, daemon=False)
+    @TimedCall(seconds=100, daemon=False, force_quit=False)
     def test_train_timing(self):
         images = sorted(glob(os.path.join(self.data_dir, "img*.nii.gz")))
         segs = sorted(glob(os.path.join(self.data_dir, "seg*.nii.gz")))