MILLION: MasterIng Long-Context LLM Inference Via Outlier-Immunized KV Product QuaNtization

Large language models (LLMs) are increasingly utilized for complex tasks requiring longer context lengths, with some models supporting up to 128 K or 1 M tokens. This trend, however, presents significant challenges in inference speed and memory management. The primary bottleneck in long-context LLM...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:2025 62nd ACM/IEEE Design Automation Conference (DAC) s. 1 - 7
Hlavní autori: Wang, Zongwu, Xu, Peng, Liu, Fangxin, Hu, Yiwei, Sun, Qingxiao, Li, Gezi, Li, Cheng, Wang, Xuan, Jiang, Li, Guan, Haibing
Médium: Konferenčný príspevok..
Jazyk:English
Vydavateľské údaje: IEEE 22.06.2025
Predmet:
Accuracy
Inference algorithms
Kernel
Large language models
Low latency communication
Memory management
Performance gain
Pipelines
Quantization (signal)
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:Large language models (LLMs) are increasingly utilized for complex tasks requiring longer context lengths, with some models supporting up to 128 K or 1 M tokens. This trend, however, presents significant challenges in inference speed and memory management. The primary bottleneck in long-context LLM inference is the quadratic computational complexity of attention mechanisms, causing substantial slowdowns as sequence length increases. KV cache mechanism alleviates this issue by storing pre-computed data, but introduces memory requirements that scale linearly with context length, hindering efficient LLM deployment. Quantization emerges as a promising approach to address the widening gap between LLM size and memory capacity. However, traditional quantization schemes often yield suboptimal compression results for KV caches due to two key factors: i) On-the-fly quantization and de-quantization, causing significant performance overhead; ii) Prevalence of outliers in KV values, challenging low-bitwidth uniform quantization. To this end, we propose MILLION, a novel quantization framework achieving low-bitwidth KV cache through product quantization. First, we conduct a thorough analysis of KV cache distribution, revealing the limitations of existing quantization schemes. Second, we introduce a non-uniform quantization algorithm based on product quantization, which efficiently compresses data while preserving accuracy. Third, we develop a high-performance GPU inference framework with efficient attention kernel and pipeline design for MILLION that leverages sparse computation and asynchronous quantization, significantly enhancing inference speed. Comprehensive evaluation results demonstrate that MILLION can achieve 4 bits quantization with trivial perplexity and accuracy loss, and achieve 2.09 x end-to-end performance gains at 32 K context length. Code is released at https://github.com/ZongwuWang/MILLION.
DOI:10.1109/DAC63849.2025.11132862