Related papers: Pushing the Limits of Online Auto-tuning: Machine …

We improve complexity bounds for energy-efficient speed scheduling problems for both the single processor and multi-processor cases. Energy conservation has become a major concern, so revisiting traditional scheduling problems to take into…

GPUs have been favored for training deep learning models due to their highly parallelized architecture. As a result, most studies on training optimization focus on GPUs. There is often a trade-off, however, between cost and efficiency when…

Large Language Models (LLMs) have seen great advance in both academia and industry, and their popularity results in numerous open-source frameworks and techniques in accelerating LLM pre-training, fine-tuning, and inference. Training and…

Resource allocation in distributed and networked systems such as the Cloud is becoming increasingly flexible, allowing these systems to dynamically adjust toward the workloads they serve, in a demand-aware manner. Online balanced…

We investigate deterministic non-preemptive online scheduling with delayed commitment for total completion time minimization on parallel identical machines. In this problem, jobs arrive one-by-one and their processing times are revealed…

Kernel development in deep learning requires optimizing computational units across hardware while balancing memory management, parallelism, and hardware-specific optimizations through extensive empirical tuning. Although domain-specific…

Multi-socket multi-core servers are used for solving some of the important problems in computing. Remote DRAM accesses can impact performance of certain applications running on such servers. This paper presents a new near linear operating…

In high-performance computing, hotspot GPU kernels are primary bottlenecks, and expert manual tuning is costly and hard to port. Large language model methods often assume kernels can be compiled and executed cheaply, which fails in large…

CPU scheduling is the reason behind the performance of multiprocessing and in time-shared operating systems. Different scheduling criteria are used to evaluate Central Processing Unit Scheduling algorithms which are based on different…

We study the \emph{bounded-delay model} for Qualify-of-Service buffer management. Time is discrete. There is a buffer. Unit-length jobs (also called \emph{packets}) arrive at the buffer over time. Each packet has an integer release time, an…

Designing reliable integrated energy systems for industrial processes requires optimization and verification models across multiple fidelities, from architecture-level sizing to high-fidelity dynamic operation. However, model mismatch…

Computing systems rarely deliver best possible performance due to ever increasing hardware and software complexity and limitations of the current optimization technology. Additional code and architecture optimizations are often required to…

Kernelization is a general theoretical framework for preprocessing instances of NP-hard problems into (generally smaller) instances with bounded size, via the repeated application of data reduction rules. For the fundamental Max Cut…

Customized processors are attractive solutions for vast domain-specific applications due to their high energy efficiency. However, designing a processor in traditional flows is time-consuming and expensive. To address this, researchers have…

Many hardware vendors have introduced specialized deep neural networks (DNN) accelerators owing to their superior performance and efficiency. As such, how to generate and optimize the code for the hardware accelerator becomes an important…

Many optimizers have been proposed for training deep neural networks, and they often have multiple hyperparameters, which make it tricky to benchmark their performance. In this work, we propose a new benchmarking protocol to evaluate both…

Accelerating Machine Learning (ML) workloads requires efficient methods due to their large optimization space. Autotuning has emerged as an effective approach for systematically evaluating variations of implementations. Traditionally,…

We consider the first, and most well studied, speed scaling problem in the algorithmic literature: where the scheduling quality of service measure is a deadline feasibility constraint, and where the power objective is to minimize the total…

With the ever-growing need of data in HPC applications, the congestion at the I/O level becomes critical in super-computers. Architectural enhancement such as burst-buffers and pre-fetching are added to machines, but are not sufficient to…

We extend and combine several tools of the literature to design fast, adaptive, anytime and scale-free online learning algorithms. Scale-free regret bounds must scale linearly with the maximum loss, both toward large losses and toward very…