Related papers: A Method for Hiding the Increased Non-Volatile Cac…
Non-volatile memory (NVM), also known as persistent memory, is an emerging paradigm for memory that preserves its contents even after power loss. NVM is widely expected to become ubiquitous, and hardware architectures are already providing…
The proliferation of 100B+ parameter Large Language Models (LLMs) with 100k+ context length support have resulted in increasing demands for on-chip memory to support large KV caches. Techniques such as StreamingLLM and SnapKV demonstrate…
Non-volatile memory (NVM) technologies suffer from limited write endurance. To address this challenge, we propose Predict and Write (PNW), a K/V-store that uses a clustering-based machine learning approach to extend the lifetime of NVMs.…
Large Language Models (LLMs) are increasingly deployed in large-scale online services, enabling sophisticated applications. However, the computational overhead of generating key-value (KV) caches in the prefill stage presents a major…
Address translation is a performance bottleneck in data-intensive workloads due to large datasets and irregular access patterns that lead to frequent high-latency page table walks (PTWs). PTWs can be reduced by using (i) large hardware TLBs…
DRAM-based memory is a critical factor that creates a bottleneck on the system performance since the processor speed largely outperforms the DRAM latency. In this thesis, we develop a low-cost mechanism, called ChargeCache, which enables…
This work studies how to adaptively recompute key-value (KV) caches for diffusion large language models (DLMs) to maximize prediction accuracy while minimizing decoding latency. Prior methods' decoders recompute QKV for all tokens at every…
Autoregressive Models (ARMs) have long dominated the landscape of Large Language Models. Recently, a new paradigm has emerged in the form of diffusion-based Large Language Models (dLLMs), which generate text by iteratively denoising masked…
Deploying Large Language Models (LLMs) in streaming applications such as multi-round dialogue, where long interactions are expected, is urgently needed but poses two major challenges. Firstly, during the decoding stage, caching previous…
Large language model (LLM) pruning with fixed N:M structured sparsity significantly limits the expressivity of the sparse model, yielding sub-optimal performance. In contrast, supporting multiple N:M patterns to provide sparse…
Performance and reliability are two prominent factors in the design of data storage systems. To achieve higher performance, recently storage system designers use DRAM-based buffers. The volatility of DRAM brings up the possibility of data…
Data obfuscation is a promising technique for mitigating attribute inference attacks by semi-trusted parties with access to time-series data emitted by sensors. Recent advances leverage conditional generative models together with…
Carbon nanotube field-effect transistors (CNFET) emerge as a promising alternative to CMOS transistors for the much higher speed and energy efficiency, which makes the technology particularly suitable for building the energy-hungry last…
Serverless Large Language Models (LLMs) have emerged as a cost-effective solution for deploying AI services by enabling a 'pay-as-you-go' pricing model through GPU resource sharing. However, cold-start latency, especially the model loading…
Multi-LLM systems harness the complementary strengths of diverse Large Language Models, achieving performance and efficiency gains that are not attainable by a single model. In existing designs, LLMs communicate through text, forcing…
Large language model (LLM) applications often reuse previously processed context, such as chat history and documents, which introduces significant redundant computation. Existing LLM serving systems address such redundant computation by…
Large Language Models (LLMs) have achieved remarkable progress across reasoning, generation, and decision-making tasks, yet deploying them on mobile, embedded, and edge devices remains particularly challenging. On-device LLM inference is…
Modern DRAM modules are often equipped with hardware error correction capabilities, especially for DRAM deployed in large-scale data centers, as process technology scaling has increased the susceptibility of these devices to errors. To…
Large Language Models(LLMs) have had a profound impact on AI applications, particularly in the domains of long-text comprehension and generation. KV Cache technology is one of the most widely used techniques in the industry. It ensures…
Digital computing-in-memory (DCIM) has emerged as a promising solution for large language model (LLM) acceleration by minimizing data transfers between external DRAM and on-chip accelerators while maintaining high precision for superior…