Although the Transformer has been the dominant architecture for time series forecasting tasks in recent years, a fundamental challenge remains: the permutationinvariant self-attention mechanism within Transformers leads to a loss of temporal information. To tackle these challenges, we propose PatchMixer, a novel CNN-based model. It introduces a permutation-variant convolutional structure to preserve temporal information. Diverging from conventional CNNs in this field, which often employ multiple scales or numerous branches, our method relies exclusively on depthwise separable convolutions. This allows us to extract both local features and global correlations using a single-scale architecture. Furthermore, we employ dual forecasting heads that encompass both linear and nonlinear components to better model future curve trends and details. Our experimental results on seven time-series forecasting benchmarks indicate that compared with the state-of-the-art method and the best-performing CNN, PatchMixer yields 3.9% and 21.2% relative improvements, respectively, while being 2-3x faster than the most advanced method. We will release our code and model. Long-term time series forecasting (LTSF) is a crucial task aimed at predicting future trends over an extended period by leveraging substantial historical time-series data. LTSF applications span a wide range of domains, including traffic flow estimation, energy management, and financial investment. Transformer (Vaswani et al., 2017) has been the dominant architecture in time series forecasting tasks in the last few years.