相关论文: Freeway to Memory Level Parallelism in Slice-Out-o…

Training LLMs larger than the aggregated memory of multiple GPUs is increasingly necessary due to the faster growth of LLM sizes compared to GPU memory. To this end, multi-tier host memory or disk offloading techniques are proposed by state…

A modern GPU aims to simultaneously execute more warps for higher Thread-Level Parallelism (TLP) and performance. When generating many memory requests, however, warps contend for limited cache space and thrash cache, which in turn severely…

Major chip manufacturers have all introduced Multithreaded processors. These processors are used for running a variety of workloads. Efficient resource utilization is an important design aspect in such processors. Particularly, it is…

In recent years, Large Language Models (LLMs) have exhibited remarkable capabilities, driving advancements in real-world applications. However, training LLMs on increasingly long input sequences imposes significant challenges due to high…

We introduce the Coarse-Grain Out-of-Order (CG- OoO) general purpose processor designed to achieve close to In-Order processor energy while maintaining Out-of-Order (OoO) performance. CG-OoO is an energy-performance proportional general…

We present the Cuckoo Trie, a fast, memory-efficient ordered index structure. The Cuckoo Trie is designed to have memory-level parallelism -- which a modern out-of-order processor can exploit to execute DRAM accesses in parallel -- without…

Parallel thinking has emerged as a new paradigm for large reasoning models (LRMs) in tackling complex problems. Recent methods leverage Reinforcement Learning (RL) to enhance parallel thinking, aiming to address the limitations in…

Pipeline parallelism (PP) is widely used for training large language models (LLMs), yet its scalability is often constrained by high activation memory consumption as the number of in-flight microbatches grows with the degree of PP. In this…

We consider energy-efficient scheduling on multiprocessors, where the speed of each processor can be individually scaled, and a processor consumes power $s^{\alpha}$ when running at speed $s$, for $\alpha>1$. A scheduling algorithm needs to…

One of the main advantages of Logic Programming (LP) is that it provides an excellent framework for the parallel execution of programs. In this work we investigate novel techniques to efficiently exploit parallelism from real-world…

An effective way to improve energy efficiency is to throttle hardware resources to meet a certain performance target, specified as a QoS constraint, associated with all applications running on a multicore system. Prior art has proposed…

Large language models (LLMs) iteratively generate text token by token, with memory usage increasing with the length of generated token sequences. Since the request generation length is generally unpredictable, it is difficult to estimate…

Fine-tuning large language models (LLMs) remains resource-intensive due to their sheer scale. While zeroth-order (ZO) optimization provides a memory-efficient alternative by eliminating backward passes, its application to…

The dedicated memory of hardware accelerators can be insufficient to store all weights and/or intermediate states of large deep learning models. Although model parallelism is a viable approach to reduce the memory pressure issue,…

The growing memory demands of modern applications have driven the adoption of far memory technologies in data centers to provide cost-effective, high-capacity memory solutions. However, far memory presents new performance challenges because…

This paper explores the potential of communicating information gained by static analysis from compilers to Out-of-Order (OoO) machines, focusing on the memory dependence predictor (MDP). The MDP enables loads to issue without all in-flight…

Large language models (LLMs) have achieved remarkable success in natural language tasks, but their inference incurs substantial computational and memory overhead. To improve efficiency, parallel decoding methods like Skeleton-of-Thought…

Despite major advancements in nonlinear programming (NLP) and convex relaxations, most system operators around the world still predominantly use some form of linear programming (LP) approximation of the AC power flow equations. This is…

Lazy search algorithms have been developed to efficiently solve planning problems in domains where the computational effort is dominated by the cost of edge evaluation. The existing algorithms operate by intelligently balancing…

Improving the multi-step reasoning ability of large language models (LLMs) with offline reinforcement learning (RL) is essential for quickly adapting them to complex tasks. While Direct Preference Optimization (DPO) has shown promise in…