# KVNAND: Efficient On-Device Large Language Model Inference Using DRAM-Free In-Flash Computing

arXiv: https://arxiv.org/abs/2512.03608 · HTML: https://mlsanity.com/paper/2512.03608/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Serving systems and request scheduling, one of 12 directions in Efficient LLM inference and serving.
- Builds on 51 of its 89 references that are in the map.
- First posted on arXiv 3 Dec 2025.

## Placed in

- Efficient LLM inference and serving / [Serving systems and request scheduling](https://mlsanity.com/inference/serving-systems-and-request-scheduling/)

## Builds on (most cited first, 30 of 51)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge](https://mlsanity.com/paper/1803.05457/) (2018)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [OPT: Open Pre-trained Transformer Language Models](https://mlsanity.com/paper/2205.01068/) (2022)
- [QLoRA: Efficient Finetuning of Quantized LLMs](https://mlsanity.com/paper/2305.14314/) (2023)
- [Pointer Sentinel Mixture Models](https://mlsanity.com/paper/1609.07843/) (2016)
- [Efficient Streaming Language Models with Attention Sinks](https://mlsanity.com/paper/2309.17453/) (2023)
- [GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers](https://mlsanity.com/paper/2210.17323/) (2022)
- [Mixtral of Experts](https://mlsanity.com/paper/2401.04088/) (2024)
- [SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models](https://mlsanity.com/paper/2211.10438/) (2022)
- [GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints](https://mlsanity.com/paper/2305.13245/) (2023)
- [AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration](https://mlsanity.com/paper/2306.00978/) (2023)
- [H2O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models](https://mlsanity.com/paper/2306.14048/) (2023)
- [Fast Transformer Decoding: One Write-Head is All You Need](https://mlsanity.com/paper/1911.02150/) (2019)
- [FlexGen: High-Throughput Generative Inference of Large Language Models with a Single GPU](https://mlsanity.com/paper/2303.06865/) (2023)
- [Scissorhands: Exploiting the Persistence of Importance Hypothesis for LLM KV Cache Compression at Test Time](https://mlsanity.com/paper/2305.17118/) (2023)
- [Towards Reasoning Era: A Survey of Long Chain-of-Thought for Reasoning Large Language Models](https://mlsanity.com/paper/2503.09567/) (2025)
- [Deja Vu: Contextual Sparsity for Efficient LLMs at Inference Time](https://mlsanity.com/paper/2310.17157/) (2023)
- [InfiniGen: Efficient Generative Inference of Large Language Models with Dynamic KV Cache Management](https://mlsanity.com/paper/2406.19707/) (2024)
- [A Survey on Efficient Inference for Large Language Models](https://mlsanity.com/paper/2404.14294/) (2024)
- [BitNet: Scaling 1-bit Transformers for Large Language Models](https://mlsanity.com/paper/2310.11453/) (2023)
- [LLM in a flash: Efficient Large Language Model Inference with Limited Memory](https://mlsanity.com/paper/2312.11514/) (2023)
- [LLM Maybe LongLM: Self-Extend LLM Context Window Without Tuning](https://mlsanity.com/paper/2401.01325/) (2024)
- [ShadowKV: KV Cache in Shadows for High-Throughput Long-Context LLM Inference](https://mlsanity.com/paper/2410.21465/) (2024)

## Built on by (most cited first, 1 of 1)

- [LLM Inference in a Flash!](https://mlsanity.com/paper/2609.16161/) (2026)

## Abstract (arXiv)

Deploying large language models (LLMs) on edge devices enables personalized agents with strong privacy and low cost. However, with tens to hundreds of billions of parameters, single-batch autoregressive inference suffers from extremely low arithmetic intensity, creating severe weight-loading and bandwidth pressures on resource-constrained platforms. Recent in-flash computing (IFC) solutions alleviate this bottleneck by co-locating weight-related linear computations in the decode phase with flash, yet still rely on DRAM for the key-value (KV) cache. As context length grows, the KV cache can exceed model weights in size, imposing prohibitive DRAM cost and capacity requirements. Attempts to offload KV cache to flash suffer from severe performance penalties. We propose KVNAND, the first DRAM-free, IFC-based architecture that stores both model weights and KV cache entirely in compute-enabled 3D NAND flash. KVNAND addresses the fundamental performance challenges of flash under intensive KV cache access by leveraging IFC for all memory-bound operations to reduce data transfer overhead, introducing head-group parallelism to boost throughput, and employing page-level KV cache mapping to align token access patterns with flash organization. In addition, we propose a design space exploration framework that evaluates discrete and compact KVNAND variants to balance weight and KV placement, automatically identifying the optimal design trade-off. These techniques mitigate latency, energy, and reliability concerns, turning flash into a practical medium for long-context KV storage. Evaluations on MHA 7B and GQA 70B LLMs show that KVNAND achieves 1.98(×)/1.94(×)/2.05(×) geomean speedup at 128/1K/10K-token contexts compared to DRAM-equipped IFC designs and addresses out-of-memory failures at 100K context length.