8 papers
FlashAttention for Scalable Vector Architectures
Sonia Rani Gupta, Nikela Papadopoulou, Miquel Pericàs
Inference with transformer models on CPUs is increasingly important, especially for Small Language Models (SLMs), where vector architectures are emerging as a promising execution s…
Sparse Computations in Deep Learning Inference
Ioanna Tasou, Panagiotis Mpakos, Angelos Vlachos +25
The computational demands of modern Deep Neural Networks (DNNs) are immense and constantly growing. While training costs usually capture public attention, inference demands are als…
Challenges and Opportunities in the Co-design of Convolutions and RISC-V Vector Processors
Sonia Rani Gupta, Nikela Papadopoulou, Miquel Pericàs
The RISC-V "V" extension introduces vector processing to the RISC-V architecture. Unlike most SIMD extensions, it supports long vectors which can result in significant improvement…
Analysis and Characterization of Performance Variability for OpenMP Runtime
Minyu Cui, Nikela Papadopoulou, Miquel Pericàs
In the high performance computing (HPC) domain, performance variability is a major scalability issue for parallel computing applications with heavy synchronization and communicatio…
ODIN: Overcoming Dynamic Interference in iNference pipelines
Pirah Noor Soomro, Nikela Papadopoulou, Miquel Pericàs
As an increasing number of businesses becomes powered by machine-learning, inference becomes a core operation, with a growing trend to be offered as a service. In this context, the…
Feature-based SpMV Performance Analysis on Contemporary Devices
Panagiotis Mpakos, Dimitrios Galanopoulos, Petros Anastasiadis +3
The SpMV kernel is characterized by high performance variation per input matrix and computing platform. While GPUs were considered State-of-the-Art for SpMV, with the emergence of…