The A64FX processor from Fujitsu, being designed for computational simulation and machine learning applications, has the potential for unprecedented performance in HPC systems. In this paper, we evaluate the A64FX by benchmarking against a range of production HPC platforms that cover a number of processor technologies. We investigate the performance of complex scientific applications across multiple nodes, as well as single node and mini-kernel benchmarks. This paper finds that the performance of the A64FX processor across our chosen benchmarks often significantly exceeds other platforms, even without specific application optimisations for the processor instruction set or hardware. However, this is not true for all the benchmarks we have undertaken. Furthermore, the specific configuration of applications can have an impact on the runtime and performance experienced.

Memory and I/O performance bottlenecks in supercomputing simulations are two key challenges that must be addressed on the road to Exascale. The new byte-addressable persistent non-volatile memory technology from Intel, DCPMM, promises to be an exciting opportunity to break with the status quo, with unprecedented levels of capacity at near-DRAM speeds. Here, we explore the potential of DCPMM in the context of two high-performance scientific applications in terms of outright performance, efficiency and usability for both its Memory and App Direct modes. In Memory mode, we show equivalent performance and better efficiency for a CASTEP simulation that is limited by memory capacity on conventional DRAM-only systems without any changes to the application. For IFS, we demonstrate that a distributed object-store over NVRAM reduces the data contention created in weather forecasting data producer-consumer workflows. In addition, we also present the achievable memory bandwidth performance using STREAM.

In modern distributed computing, vast amounts of data are stored in many different formats, employing different storage solutions. OGSA-DAI 3.0 is a middleware software solution. It provides application developers with the means to access data distributed across multiple platforms with different native access mechanisms. Data integration can take place at the server and deliver results using a variety of protocols and mechanisms within OGSA-DAI. It accomplishes this by using a highly flexible and extensible framework which can accommodate different types of data resources, such as XML databases, relational databases or files, different operations such as transformation to different formats, selection or filter operations. The framework can be extended by a developer to provide customized functionality for project specific tasks while using generic functions for common tasks such as database querying. This paper presents an overview of OGSA-DAI and how it tackles data access and integration through a set of example use cases.

The use of FPGAs for general-purpose supercomputing is being investigated by several projects worldwide. The FPGA High Performance Computing Alliance (FHPCA) is developing HPC solutions for real-world applications using FPGAs. It has built a 64-FPGA supercomputer located in Scotland.The widespread take-up of FPGAs will only occur when they are more easily programmed. Ease of programming is the major challenge faced by this technology. This BoF will focus on reporting the experience gained in porting three real-world industrial applications to the FHPCA system and discussing the developments needed to overcome the programmability barrier. Panellists will come from the FHPCA and other organisations active in this area. There will be a strong emphasis on moderated discussion.The FHPCA involves EPCC, Alpha Data Ltd, Nallatech Ltd, Xilinx Corporation, Algotronix Ltd, ISLI and Scottish Enterprise.