HACK: Homomorphic Acceleration via Compression of the Key-Value Cache for Disaggregated LLM Inference
This addresses the computational and transmission inefficiencies in LLM inference for long prompts, offering a significant improvement over existing methods.
The paper tackles the bottleneck of transmitting Key-Value (KV) data in disaggregated LLM inference by proposing HACK, which eliminates dequantization overhead and directly computes on quantized KV data, reducing Job Completion Time by up to 70.9% compared to baseline and 52.3% compared to state-of-the-art methods.
Disaggregated Large Language Model (LLM) inference has gained popularity as it separates the computation-intensive prefill stage from the memory-intensive decode stage, avoiding the prefill-decode interference and improving resource utilization. However, transmitting Key-Value (KV) data between the two stages can be a bottleneck, especially for long prompts. Additionally, the computation time overhead for prefill and decode is key for optimizing Job Completion Time (JCT), and KV data size can become prohibitive for long prompts and sequences. Existing KV quantization methods can alleviate the transmission bottleneck and reduce memory requirements, but they introduce significant dequantization overhead, exacerbating the computation time. We propose Homomorphic Acceleration via Compression of the KV cache (HACK) for disaggregated LLM inference. HACK eliminates the heavy KV dequantization step, and directly performs computations on quantized KV data to approximate and reduce the cost of the expensive matrix-multiplication step. Extensive trace-driven experiments show that HACK reduces JCT by up to 70.9% compared to disaggregated LLM inference baseline and by up to 52.3% compared to state-of-the-art KV quantization methods.