MatKV: Trading Compute for Flash Storage in LLM Inference
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914221702250496 |
|---|---|
| author | Shin, Kun-Woo Park, Jay H. Oh, Moonwook Jo, Yohan Do, Jaeyoung Lee, Sang-Won |
| author_facet | Shin, Kun-Woo Park, Jay H. Oh, Moonwook Jo, Yohan Do, Jaeyoung Lee, Sang-Won |
| contents | We observe two major trends in LLM-based generative AI: (1) inference is becoming the dominant factor in terms of cost and power consumption, surpassing training, and (2) retrieval augmented generation (RAG) is becoming prevalent. When processing long inputs in RAG, the prefill phase of computing the key-value vectors of input text is energy-intensive and time-consuming even with high-end GPUs. Thus, it is crucial to make the prefill phase in RAG inference efficient. To address this issue, we propose MatKV, a scheme that precomputes the key-value vectors (KVs) of RAG objects (e.g., documents), materializes them in inexpensive but fast and power-efficient flash storage, and reuses them at inference time instead of recomputing the KVs using costly and power-inefficient GPU. Experimental results using Hugging Face's Transformers library across state-of-the-art GPUs and flash memory SSDs confirm that, compared to full KV computation on GPUs, MatKV reduces both inference time and power consumption by half for RAG workloads, without severely impacting accuracy in the question-answering task. Furthermore, we demonstrate that MatKV enables additional optimizations in two ways. First, a GPU can decode text while simultaneously loading the materialized KVs for the next instance, reducing load latency. Second, since decoding speed is less sensitive to GPU performance than KV computation, low-end GPUs can be leveraged for decoding without significantly compromising speed once the materialized KVs are loaded into GPU memory. These findings underscore MatKV's potential to make large-scale generative AI applications more cost-effective, power-efficient, and accessible across a wider range of tasks and hardware environments. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_22195 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | MatKV: Trading Compute for Flash Storage in LLM Inference Shin, Kun-Woo Park, Jay H. Oh, Moonwook Jo, Yohan Do, Jaeyoung Lee, Sang-Won Distributed, Parallel, and Cluster Computing Artificial Intelligence Machine Learning We observe two major trends in LLM-based generative AI: (1) inference is becoming the dominant factor in terms of cost and power consumption, surpassing training, and (2) retrieval augmented generation (RAG) is becoming prevalent. When processing long inputs in RAG, the prefill phase of computing the key-value vectors of input text is energy-intensive and time-consuming even with high-end GPUs. Thus, it is crucial to make the prefill phase in RAG inference efficient. To address this issue, we propose MatKV, a scheme that precomputes the key-value vectors (KVs) of RAG objects (e.g., documents), materializes them in inexpensive but fast and power-efficient flash storage, and reuses them at inference time instead of recomputing the KVs using costly and power-inefficient GPU. Experimental results using Hugging Face's Transformers library across state-of-the-art GPUs and flash memory SSDs confirm that, compared to full KV computation on GPUs, MatKV reduces both inference time and power consumption by half for RAG workloads, without severely impacting accuracy in the question-answering task. Furthermore, we demonstrate that MatKV enables additional optimizations in two ways. First, a GPU can decode text while simultaneously loading the materialized KVs for the next instance, reducing load latency. Second, since decoding speed is less sensitive to GPU performance than KV computation, low-end GPUs can be leveraged for decoding without significantly compromising speed once the materialized KVs are loaded into GPU memory. These findings underscore MatKV's potential to make large-scale generative AI applications more cost-effective, power-efficient, and accessible across a wider range of tasks and hardware environments. |
| title | MatKV: Trading Compute for Flash Storage in LLM Inference |
| topic | Distributed, Parallel, and Cluster Computing Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2512.22195 |