Topic: GPUs

Katarzyna Swirydowicz uses an asynchronous solver on an AMD system to examine the practical implementation of computational linear algebra across CPUs and GPUs. The talk introduces the relevant computational ideas, programming models, and software tools, then considers how algorithmic structure interacts with hardware capabilities. This case study illustrates both the opportunities and implementation challenges of asynchronous methods for large-scale scientific computing on GPU-accelerated systems.

Zain Asgar examines running agent inference across heterogeneous hardware. The discussion covers splitting workloads between GPUs and CPUs, compilation for different targets, automated kernel optimisation and hardware-aware scheduling. It also explores networking constraints and trade-offs among numerical precision, application accuracy, performance and cost.

The Brian simulator, a simulator for spiking neural networks, started its life in 2007 and has been continuously developed ever since. A complete rewrite in 2014 led to the "brian2" package, which is now built as a code generation framework. This means that users write their code as a Python script, with textual representations of the model equations, and the simulator transparently translates this code into a target language such as C++. In this talk, I will give an overview over recent developments in the Brian ecosystem, showcasing a number of packages that enhance the simulator's capabili

Since the popularization of GPUs for machine learning (ML) workloads, several dedicated accelerator chips have emerged, offering architectures optimized for common ML operations and requirements. The types of tasks targeted by such platforms however - often sparse and highly parallel - are not confined to the realm of traditional layer-based learning. Established simulators for large-scale spiking networks with biologically plausible connectivity and synaptic density have historically targeted CPUs, with GPU support and ports being a relatively recent development[1][2]. Here, we present a work