# MixDiT: Accelerating Image Diffusion Transformer Inference with Mixed-Precision MX Quantization

arXiv: https://arxiv.org/abs/2504.08398 · HTML: https://mlsanity.com/paper/2504.08398/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models.
- Builds on 9 of its 15 references that are in the map.
- First posted on arXiv 11 Apr 2025.

## Placed in

- Diffusion models / [Diffusion foundations and sampling](https://mlsanity.com/diffusion/diffusion-foundations-and-sampling/)

## Builds on (most cited first, 9 of 9)

- [Scalable Diffusion Models with Transformers](https://mlsanity.com/paper/2212.09748/) (2022)
- [Scaling Rectified Flow Transformers for High-Resolution Image Synthesis](https://mlsanity.com/paper/2403.03206/) (2024)
- [PixArt-Sigma: Weak-to-Strong Training of Diffusion Transformer for 4K Text-to-Image Generation](https://mlsanity.com/paper/2403.04692/) (2024)
- [ViDiT-Q: Efficient and Accurate Quantization of Diffusion Transformers for Image and Video Generation](https://mlsanity.com/paper/2406.02540/) (2024)
- [Q-DiT: Accurate Post-Training Quantization for Diffusion Transformers](https://mlsanity.com/paper/2406.17343/) (2024)
- [PTQ4DiT: Post-training Quantization for Diffusion Transformers](https://mlsanity.com/paper/2405.16005/) (2024)
- [With Shared Microexponents, A Little Shifting Goes a Long Way](https://mlsanity.com/paper/2302.08007/) (2023)
- [MixDQ: Memory-Efficient Few-Step Text-to-Image Diffusion Models with Metric-Decoupled Mixed Precision Quantization](https://mlsanity.com/paper/2405.17873/) (2024)
- [DaCapo: Accelerating Continuous Learning in Autonomous Systems for Video Analytics](https://mlsanity.com/paper/2403.14353/) (2024)

## Built on by (most cited first, 8 of 8)

- [Neo: Real-Time On-Device 3D Gaussian Splatting with Reuse-and-Update Sorting Acceleration](https://mlsanity.com/paper/2511.12930/) (2025)
- [TreeQ: Pushing the Quantization Boundary of Diffusion Transformer via Tree-Structured Mixed-Precision Search](https://mlsanity.com/paper/2512.06353/) (2025)
- [Rethinking FID Through the Geometry of the Reference Dataset](https://mlsanity.com/paper/2605.29335/) (2026)
- [TMPDiff: Temporal Mixed-Precision for Diffusion Models](https://mlsanity.com/paper/2603.14062/) (2026)
- [Realizing Native INT8 Compute for Diffusion Transformers on Consumer GPUs: A Fused INT8 GEMM Kernel for Ideogram 4.0](https://mlsanity.com/paper/2606.14598/) (2026)
- [Understanding and Exploiting Diagonal Attention Sparsity in Autoregressive Image Generation](https://mlsanity.com/paper/2609.19702/) (2026)
- [Precision As You Need: Stochastic Computing Is a Dense Adaptive Quantizer](https://mlsanity.com/paper/2609.32922/) (2026)
- [What Does Fr'echet Distance Measure? A Directional Decomposition](https://mlsanity.com/paper/2610.05518/) (2026)

## Abstract (arXiv)

Diffusion Transformer (DiT) has driven significant progress in image generation tasks. However, DiT inferencing is notoriously compute-intensive and incurs long latency even on datacenter-scale GPUs, primarily due to its iterative nature and heavy reliance on GEMM operations inherent to its encoder-based structure. To address the challenge, prior work has explored quantization, but achieving low-precision quantization for DiT inferencing with both high accuracy and substantial speedup remains an open problem. To this end, this paper proposes MixDiT, an algorithm-hardware co-designed acceleration solution that exploits mixed Microscaling (MX) formats to quantize DiT activation values. MixDiT quantizes the DiT activation tensors by selectively applying higher precision to magnitude-based outliers, which produce mixed-precision GEMM operations. To achieve tangible speedup from the mixed-precision arithmetic, we design a MixDiT accelerator that enables precision-flexible multiplications and efficient MX precision conversions. Our experimental results show that MixDiT delivers a speedup of 2.10-5.32 times over RTX 3090, with no loss in FID.