arXiv subjects

Nick Brown

Publications and source records attributed to Nick Brown.

At least 37 records · Page 2Linked to original sources

arXiv2024

RISC-V for HPC: Where we are and where we need to go

Funded by the UK ExCALIBUR H&ES exascale programme, since early 2022 we have provided a RISC-V testbed for HPC to offer free access for scientific software developers to experiment with RISC-V for their workloads. Based upon our experiences of providing access to RISC-V for the HPC community, and our involvement with the RISC-V community at large, in this extended abstract we summarise the current state of RISC-V for HPC and consider the high priority areas that should be addressed to help drive adoption.

cs.DC↗

arXiv2024

Evaluating Versal AI Engines for option price discovery in market risk analysis

Whilst Field-Programmable Gate Arrays (FPGAs) have been popular in accelerating high-frequency financial workload for many years, their application in quantitative finance, the utilisation of mathematical models to analyse financial markets and securities, is less mature. Nevertheless, recent work has demonstrated the benefits that FPGAs can deliver to quantitative workloads, and in this paper, we study whether the Versal ACAP and its AI Engines (AIEs) can also deliver improved performance. We focus specifically on the industry standard Strategic Technology Analysis Center's (STAC) derivatives risk analysis benchmark STAC-A2. Porting a purely FPGA-based accelerator STAC-A2 inspired market risk (SIMR) benchmark to the Versal ACAP device by combining Programmable Logic (PL) and AIEs, we explore the development approach and techniques, before comparing performance across PL and AIEs. Ultimately, we found that our AIE approach is slower than a highly optimised existing PL-only version due to limits on both the AIE and PL that we explore and describe.

cs.DC↗

arXiv2024

Fast and energy-efficient derivatives risk analysis: Streaming option Greeks on Xilinx and Intel FPGAs

Whilst FPGAs have enjoyed success in accelerating high-frequency financial workloads for some time, their use for quantitative finance, which is the use of mathematical models to analyse financial markets and securities, has been far more limited to-date. Currently, CPUs are the most common architecture for such workloads, and an important question is whether FPGAs can ameliorate some of the bottlenecks encountered on those architectures. In this paper we extend our previous work accelerating the industry standard Securities Technology Analysis Center's (STAC\textregistered) derivatives risk analysis benchmark STAC-A2\texttrademark{}, by first porting this from our previous Xilinx implementation to an Intel Stratix-10 FPGA, exploring the challenges encountered when moving from one FPGA architecture to another and suitability of techniques. We then present a host-data-streaming approach that ultimately outperforms our previous version on a Xilinx Alveo U280 FPGA by up to 4.6 times and requiring 9 times less energy at the largest problem size, while outperforming the CPU and GPU versions by up to 8.2 and 5.2 times respectively. The result of this work is a significant enhancement in FPGA performance against the previous version for this industry standard benchmark running on both Xilinx and Intel FPGAs, and furthermore an exploration of optimisation and porting techniques that can be applied to other HPC workloads.

cs.DC↗

arXiv2024

Interactive and Urgent HPC: Challenges and Opportunities

As a broader set of applications from simulations to data analysis and machine learning require more parallel computational capability, the demand for interactive and urgent high performance computing (HPC) continues to increase. This paper overviews the progress made so far and elucidates the challenges and opportunities for greater integration of interactive and urgent HPC policies, techniques, and technologies into HPC ecosystems.

cs.DC↗

arXiv2023

Fortran performance optimisation and auto-parallelisation by leveraging MLIR-based domain specific abstractions in Flang

MLIR has become popular since it was open sourced in 2019. A sub-project of LLVM, the flexibility provided by MLIR to represent Intermediate Representations (IR) as dialects at different abstraction levels, to mix these, and to leverage transformations between dialects provides opportunities for automated program optimisation and parallelisation. In addition to general purpose compilers built upon MLIR, domain specific abstractions have also been developed. In this paper we explore complimenting the Flang MLIR general purpose compiler by combining with the domain specific Open Earth Compiler's MLIR stencil dialect. Developing transformations to discover and extracts stencils from Fortran, this specialisation delivers between a 2 and 10 times performance improvement for our benchmarks on a Cray supercomputer compared to using Flang alone. Furthermore, by leveraging existing MLIR transformations we develop an auto-parallelisation approach targeting multi-threaded and distributed memory parallelism, and optimised execution on GPUs, without any modifications to the serial Fortran source code.

cs.DC↗

arXiv2023

Is RISC-V ready for HPC prime-time: Evaluating the 64-core Sophon SG2042 RISC-V CPU

The Sophon SG2042 is the world's first commodity 64-core RISC-V CPU for high performance workloads and an important question is whether the SG2042 has the potential to encourage the HPC community to embrace RISC-V. In this paper we undertaking a performance exploration of the SG2042 against existing RISC-V hardware and high performance x86 CPUs in use by modern supercomputers. Leveraging the RAJAPerf benchmarking suite, we discover that on average, the SG2042 delivers, per core, between five and ten times the performance compared to the nearest widely available RISC-V hardware. We found that, on average, the x86 high performance CPUs under test outperform the SG2042 by between four and eight times for multi-threaded workloads, although some individual kernels do perform faster on the SG2042. The result of this work is a performance study that not only contrasts this new RISC-V CPU against existing technologies, but furthermore shares performance best practice.

cs.DC↗

arXiv2023

Stencil-HMLS: A multi-layered approach to the automatic optimisation of stencil codes on FPGA

The challenges associated with effectively programming FPGAs have been a major blocker in popularising reconfigurable architectures for HPC workloads. However new compiler technologies, such as MLIR, are providing new capabilities which potentially deliver the ability to extract domain specific information and drive automatic structuring of codes for FPGAs. In this paper we explore domain specific optimisations for stencils, a fundamental access pattern in scientific computing, to obtain high performance on FPGAs via automated code structuring. We propose Stencil-HMLS, a multi-layered approach to automatic optimisation of stencil codes and introduce the HLS dialect, which brings FPGA programming into the MLIR ecosystem. Using the PSyclone Fortran DSL, we demonstrate an improvement of 14-100$\times$ with respect to the next best performant state-of-the-art tool. Furthermore, our approach is 14 to 92 times more energy efficient than the next most energy efficient approach.

cs.DC↗

arXiv2023

Fortran High-Level Synthesis: Reducing the barriers to accelerating HPC codes on FPGAs

In recent years the use of FPGAs to accelerate scientific applications has grown, with numerous applications demonstrating the benefit of FPGAs for high performance workloads. However, whilst High Level Synthesis (HLS) has significantly lowered the barrier to entry in programming FPGAs by enabling programmers to use C++, a major challenge is that most often these codes are not originally written in C++. Instead, Fortran is the lingua franca of scientific computing and-so it requires a complex and time consuming initial step to convert into C++ even before considering the FPGA. In this paper we describe work enabling Fortran for AMD Xilinx FPGAs by connecting the LLVM Flang front end to AMD Xilinx's LLVM back end. This enables programmers to use Fortran as a first-class language for programming FPGAs, and as we demonstrate enjoy all the tuning and optimisation opportunities that HLS C++ provides. Furthermore, we demonstrate that certain language features of Fortran make it especially beneficial for programming FPGAs compared to C++. The result of this work is a lowering of the barrier to entry in using FPGAs for scientific computing, enabling programmers to leverage their existing codebase and language of choice on the FPGA directly.

cs.DC↗

arXiv2023

Experiences of running an HPC RISC-V testbed

Funded by the UK ExCALIBUR H\&ES exascale programme, in early 2022 a RISC-V testbed for HPC was stood up to provide free access for scientific software developers to experiment with RISC-V for their workloads. Here we report on successes, challenges, and lessons learnt from this activity with a view to better understanding the suitability of RISC-V for HPC and important areas to focus RISC-V HPC community efforts upon.

cs.DC↗

arXiv2023

Test-driving RISC-V Vector hardware for HPC

Whilst the RISC-V Vector extension (RVV) has been ratified, at the time of writing both hardware implementations and open source software support are still limited for vectorisation on RISC-V. This is important because vectorisation is crucial to obtaining good performance for High Performance Computing (HPC) workloads and, as of April 2023, the Allwinner D1 SoC, containing the XuanTie C906 processor, is the only mass-produced and commercially available hardware supporting RVV. This paper surveys the current state of RISC-V vectorisation as of 2023, reporting the landscape of both the hardware and software ecosystem. Driving our discussion from experiences in setting up the Allwinner D1 as part of the EPCC RISC-V testbed, we report the results of benchmarking the Allwinner D1 using the RAJA Performance Suite, which demonstrated reasonable vectorisation speedup using vendor-provided compiler, as well as favourable performance compared to the StarFive VisionFive V2 with SiFive's U74 processor.

cs.DC↗

arXiv2023

Backporting RISC-V Vector assembly

Leveraging vectorisation, the ability for a CPU to apply operations to multiple elements of data concurrently, is critical for high performance workloads. However, at the time of writing, commercially available physical RISC-V hardware that provides the RISC-V vector extension (RVV) only supports version 0.7.1, which is incompatible with the latest ratified version 1.0. The challenge is that upstream compiler toolchains, such as Clang, only target the ratified v1.0 and do not support the older v0.7.1. Because v1.0 is not compatible with v0.7.1, the only way to program vectorised code is to use a vendor-provided, older compiler. In this paper we introduce the rvv-rollback tool which translates assembly code generated by the compiler using vector extension v1.0 instructions to v0.7.1. We utilise this tool to compare vectorisation performance of the vendor-provided GNU 8.4 compiler (supports v0.7.1) against LLVM 15.0 (supports only v1.0), where we found that the LLVM compiler is capable of auto-vectorising more computational kernels, and delivers greater performance than GNU in most, but not all, cases. We also tested LLVM vectorisation with vector length agnostic and specific settings, and observed cases with significant difference in performance.

cs.DC↗

arXiv2023

Morpheus unleashed: Fast cross-platform SpMV on emerging architectures

Sparse matrices and linear algebra are at the heart of scientific simulations. Over the years, more than 70 sparse matrix storage formats have been developed, targeting a wide range of hardware architectures and matrix types, each of which exploit the particular strengths of an architecture, or the specific sparsity patterns of the matrices. In this work, we explore the suitability of storage formats such as COO, CSR and DIA for emerging architectures such as AArch64 CPUs and FPGAs. In addition, we detail hardware-specific optimisations to these targets and evaluate the potential of each contribution to be integrated into Morpheus, a modern library that provides an abstraction of sparse matrices (currently) across x86 CPUs and NVIDIA/AMD GPUs. Finally, we validate our work by comparing the performance of the Morpheus-enabled HPCG benchmark against vendor-optimised implementations.

cs.DC↗

arXiv2023

Task-based preemptive scheduling on FPGAs leveraging partial reconfiguration

FPGAs are an attractive type of accelerator for all-purpose HPC computing systems due to the possibility of deploying tailored hardware on demand. However, the common tools for programming and operating FPGAs are still complex to use, especially in scenarios where diverse types of tasks should be dynamically executed. In this work we present a programming abstraction with a simple interface that internally leverages High-Level Synthesis, Dynamic Partial Reconfiguration and synchronisation mechanisms to use an FPGA as a multi-tasking server with preemptive scheduling and priority queues. This leads to an improved use of the FPGA resources, allowing the execution of several different kernels concurrently and deploying the most urgent ones as fast as possible. The results of our experimental study show that our approach incurs only a 10% overhead in the worst case when using two reconfigurable regions, whilst providing a significant performance improvement of at least 24% over the traditional full reconfiguration approach.

cs.DC↗

arXiv2022

Web-based volunteer distributed computing for handling time-critical urgent workloads

Urgent computing workloads are time critical, unpredictable, and highly dynamic. Whilst efforts are on-going to run these on traditional HPC machines, another option is to leverage the computing power donated by volunteers. Volunteer computing, where members of the public donate some of their CPU time to large scale projects has been popular for many years because it is a powerful way of delivering compute for specific problems, with the public often eager to contribute to a good cause with societal benefits. However, traditional volunteer computing has required user installation of specialist software which is a barrier to entry, and the development of the software itself by the projects, even on-top of existing frameworks, is non-trivial. As such, the number of users donating CPU time to these volunteer computing projects has decreased in recent years, and this comes at a time when the frequency of disasters, often driven by climate change, are rising fast. We believe that an alternative approach, where visitors to websites donate some of their CPU time whilst they are browsing, has the potential to address these issues. However, web-based distributed computing is an immature field and there are numerous questions that must be answered to fully understand the viability of leveraging the large scale parallelism that website visitors represent. In this paper we describe our web-based distributed computing framework, Panther, and perform in-depth performance experiments for two benchmarks using real world hardware and real world browsing habits for the first time. By exploring the performance characteristics of our approach we demonstrate that this is viable for urgent workloads, but there are numerous caveats, not least the most appropriate visitor patterns to a website, that must be considered.

cs.DC↗

arXiv2022

Exploring the Versal AI engines for accelerating stencil-based atmospheric advection simulation

AMD Xilinx's new Versal Adaptive Compute Acceleration Platform (ACAP) is an FPGA architecture combining reconfigurable fabric with other on-chip hardened compute resources. AI engines are one of these and, by operating in a highly vectorized manner, they provide significant raw compute that is potentially beneficial for a range of workloads including HPC simulation. However, this technology is still early-on, and as yet unproven for accelerating HPC codes, with a lack of benchmarking and best practice. This paper presents an experience report, exploring porting of the Piacsek and Williams (PW) advection scheme onto the Versal ACAP, using the chip's AI engines to accelerate the compute. A stencil-based algorithm, advection is commonplace in atmospheric modelling, including several Met Office codes who initially developed this scheme. Using this algorithm as a vehicle, we explore optimal approaches for structuring AI engine compute kernels and how best to interface the AI engines with programmable logic. Evaluating performance using a VCK5000 against non-AI engine FPGA configurations on the VCK5000 and Alveo U280, as well as a 24-core Xeon Platinum Cascade Lake CPU and Nvidia V100 GPU, we found that whilst the number of channels between the fabric and AI engines are a limitation, by leveraging the ACAP we can double performance compared to an Alveo U280.

cs.DC↗

arXiv2022

Performance of the Vipera framework for DSLs on micro-core architectures

Vipera provides a compiler and runtime framework for implementing dynamic Domain-Specific Languages on micro-core architectures. The performance and code size of the generated code is critical on these architectures. In this paper we present the results of our investigations into the efficiency of Vipera in terms of code performance and size.

cs.DC↗

arXiv2022

Programming abstractions for preemptive scheduling in FPGAs using partial reconfiguration

FPGAs are an attractive type of accelerator for all-purpose HPC computing systems due to the possibility of deploying tailored hardware on demand. However, the common tools for programming and operating FPGAs are still complex to use, specially in scenarios where diverse types of tasks should be dynamically executed. In this work we present a programming abstraction with a simple interface that internally leverages High-Level Synthesis, Dynamic Partial Reconfiguration and synchronisation mechanisms to use an FPGA as a multi-tasking server with preemptive scheduling and priority queues. This leads to a better use of the FPGA resources, allowing the execution of several kernels at the same time and deploying the most urgent ones as fast as possible. The results of our experimental study show that our approach incurs only a 1.66% overhead when using only one Reconfigurable Region (RR), and 4.04% when using two RRs, whilst presenting a significant performance improvement over the traditional non-preemptive full reconfiguration approach.

cs.AR↗

arXiv2022

TensorFlow as a DSL for stencil-based computation on the Cerebras Wafer Scale Engine

The Cerebras Wafer Scale Engine (WSE) is an accelerator that combines hundreds of thousands of AI-cores onto a single chip. Whilst this technology has been designed for machine learning workloads, the significant amount of available raw compute means that it is also a very interesting potential target for accelerating traditional HPC computational codes. Many of these algorithms are stencil-based, where update operations involve contributions from neighbouring elements, and in this paper we explore the suitability of this technology for such codes from the perspective of an early adopter of the technology, compared to CPUs and GPUs. Using TensorFlow as the interface, we explore the performance and demonstrate that, whilst there is still work to be done around exposing the programming interface to users, performance of the WSE is impressive as it out performs four V100 GPUs by two and a half times and two Intel Xeon Platinum CPUs by around 114 times in our experiments. There is significant potential therefore for this technology to play an important role in accelerating HPC codes on future exascale supercomputers.

cs.DC↗