Explanation of INT8 ConvRot (FP8 is no longer needed)

ComfyUI v0.27.0 introduced INT8 ConvRot, a quantization method that outperforms the previously standard FP8 formats on both GeForce RTX 20/30 and 40/50 series GPUs, making it poised to become the new standard for 8-bit quantized AI models. The article explains that INT8 ConvRot combines encoding methods, scaling techniques, and quantization algorithms to reduce model file sizes while maintaining precision, and clarifies the distinctions between file formats, encoding schemes, scaling methods, and quantization approaches in AI model storage.