Logo image
返回
Tiamat: A 98-to-134ms/Step Transformer-Based Diffusion Model Processor Supporting Classifier-Free Guidance for Image Generation
會議論文集

Tiamat: A 98-to-134ms/Step Transformer-Based Diffusion Model Processor Supporting Classifier-Free Guidance for Image Generation

Po-Yen Lu, Po-Wei Chen, Shau-Hua Yeh, Zhi-Jun Zhang, Yong-Tai Chen, Ting-Yu Wang 和 Chao-Tsung Huang
Digest of technical papers - IEEE International Solid-State Circuits Conference, 卷.69, 頁碼.54-56
15/02/2026
DOI: https://doi.org/10.1109/ISSCC49663.2026.11409204
Web of Science ID: WOS:001760995900011

摘要

Diffusion models FinFETs Image synthesis Quantization (signal) Solid state circuits Text to image Transformers
A 16 n m FinFET transformer-based diffusion model processor chip is fabricated for supporting class-conditional DiT-XL/2 and text-to-image PixArt-a with 98 ms and 134 ms generation time per step with 7.37TOPS and 1477 mW at 400 MHz. Classifier-free guidance batching reduces external memory access (EMA) for weights by 50%, and weight-reordered quantization brings another 37% reduction. Hybrid microscaling blocking reduces featuremap buffer size by 37%, while approaching floating-point quality.

相關連結

指標

1 檢視次數

詳細資料

Logo image