Acceleration-as-a-Service: Exploiting Virtualised GPUs for a Financial Application

How can GPU acceleration be obtained as a service in a cluster? This question has become increasingly significant due to the inefficiency of installing GPUs on all nodes of a cluster. The research reported in this paper is motivated to address the above question by employing rCUDA (remote CUDA), a framework that facilitates Acceleration-as-a-Service (AaaS), such that the nodes of a cluster can request the acceleration of a set of remote GPUs on demand. The rCUDA framework exploits virtualisation and ensures that multiple nodes can share the same GPU. In this paper we test the feasibility of the rCUDA framework on a real-world application employed in the financial risk industry that can benefit from AaaS in the production setting. The results confirm the feasibility of rCUDA and highlight that rCUDA achieves similar performance compared to CUDA, provides consistent results, and more importantly, allows for a single application to benefit from all the GPUs available in the cluster without loosing efficiency.

The design and implementation of a study to investigate the effectiveness of community vs hospital eye service follow-up for patients with neovascular age-related macular degeneration with quiescent disease

Introduction
Standard treatment for neovascular age-related macular degeneration (nAMD) is intravitreal injections of anti-VEGF drugs. Following multiple injections, nAMD lesions often become quiescent but there is a high risk of reactivation, and regular review by hospital ophthalmologists is the norm. The present trial examines the feasibility of community optometrists making lesion reactivation decisions.

Methods
The Effectiveness of Community vs Hospital Eye Service (ECHoES) trial is a virtual trial; lesion reactivation decisions were made about vignettes that comprised clinical data, colour fundus photographs, and optical coherence tomograms displayed on a web-based platform. Participants were either hospital ophthalmologists or community optometrists. All participants were provided with webinar training on the disease, its management, and assessment of the retinal imaging outputs. In a balanced design, 96 participants each assessed 42 vignettes; a total of 288 vignettes were assessed seven times by each professional group.The primary outcome is a participant's judgement of lesion reactivation compared with a reference standard. Secondary outcomes are the frequency of sight threatening errors; judgements about specific lesion components; participant-rated confidence in their decisions about the primary outcome; cost effectiveness of follow-up by optometrists rather than ophthalmologists.

Discussion
This trial addresses an important question for the NHS, namely whether, with appropriate training, community optometrists can make retreatment decisions for patients with nAMD to the same standard as hospital ophthalmologists. The trial employed a novel approach as participation was entirely through a web-based application; the trial required very few resources compared with those that would have been needed for a conventional randomised controlled clinical trial.