diff --git a/README.md b/README.md
index a0e39a3..914f27f 100644
--- a/README.md
+++ b/README.md
@@ -75,6 +75,8 @@ The optimizer uses PyTorch and CUDA extensions in a Python environment to produc
 - 24 GB VRAM (to train to paper evaluation quality)
 - Please see FAQ for smaller VRAM configurations
 
+PS: 24 GB VRAM is not the minimum configuration. When using `--data_devie cpu` to lazy load image, 8GB VRAM is already sufficient to handle most scenarios.
+
 ### Software Requirements
 - Conda (recommended for easy setup)
 - C++ Compiler for PyTorch extensions (we used Visual Studio 2019 for Windows)