# QFlash: Bridging Quantization and Memory Efficiency in Vision Transformer Attention

arXiv: https://arxiv.org/abs/2604.25306 · HTML: https://mlsanity.com/paper/2604.25306/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 16 of its 26 references that are in the map.
- First posted on arXiv 28 Apr 2026.

## Builds on (most cited first, 16 of 16)

- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning](https://mlsanity.com/paper/2307.08691/) (2023)
- [FlashAttention-3: Fast and Accurate Attention with Asynchrony and Low-precision](https://mlsanity.com/paper/2407.08608/) (2024)
- [I-BERT: Integer-only BERT Quantization](https://mlsanity.com/paper/2101.01321/) (2021)
- [PTQ4ViT: Post-training quantization for vision transformers with twin uniform quantization](https://mlsanity.com/paper/2111.12293/) (2021)
- [Online normalizer calculation for softmax](https://mlsanity.com/paper/1805.02867/) (2018)
- [FQ-ViT: Post-Training Quantization for Fully Quantized Vision Transformer](https://mlsanity.com/paper/2111.13824/) (2021)
- [RepQ-ViT: Scale Reparameterization for Post-Training Quantization of Vision Transformers](https://mlsanity.com/paper/2212.08254/) (2022)
- [I-ViT: Integer-only Quantization for Efficient Vision Transformer Inference](https://mlsanity.com/paper/2207.01405/) (2022)
- [ShiftAddLLM: Accelerating Pretrained LLMs via Post-Training Multiplication-Less Reparameterization](https://mlsanity.com/paper/2406.05981/) (2024)
- [I-LLM: Efficient Integer-Only Inference for Fully-Quantized Low-Bit Large Language Models](https://mlsanity.com/paper/2405.17849/) (2024)
- [CLAMP-ViT: Contrastive Data-Free Learning for Adaptive Post-Training Quantization of ViTs](https://mlsanity.com/paper/2407.05266/) (2024)
- [QUARK: Quantization-Enabled Circuit Sharing for Transformer Acceleration by Exploiting Common Patterns in Nonlinear Operations](https://mlsanity.com/paper/2511.06767/) (2025)
- [INT-FlashAttention: Enabling Flash Attention for INT8 Quantization](https://mlsanity.com/paper/2409.16997/) (2024)
- [Mixed Non-linear Quantization for Vision Transformers](https://mlsanity.com/paper/2407.18437/) (2024)
- [Low Overhead Instruction Latency Characterization for NVIDIA GPGPUs](https://mlsanity.com/paper/1905.08778/) (2019)

## Abstract (arXiv)

FlashAttention improves efficiency through tiling, but its online softmax still relies on floating-point arithmetic for numerical stability, making full quantization difficult. We identify three main obstacles to integer-only FlashAttention: (1) scale explosion during tile-wise accumulation, (2) inefficient shift-based exponential operations on GPUs, and (3) quantization granularity constraints requiring uniform scales for integer comparison. To address these challenges, we propose textitQFlash, an end-to-end integer FlashAttention design that performs softmax entirely in the integer domain and runs as a single Triton kernel. On seven attention workloads from ViT, DeiT, and Swin models, QFlash achieves up to 6.73× speedup over I-ViT and up to 8.69× speedup on Swin, while reducing energy consumption by 18.8% compared to FP16 FlashAttention, without sacrificing Top-1 accuracy on ViT/DeiT and remaining competitive on Swin under per-tensor quantization. Our code is publicly available at https://github.com/EfficientCompLab/qflash.