Sanger Benchmarking Suite – an application specific benchmarking suite for genomics
Monday, May 13, 2024 3:00 PM to Wednesday, May 15, 2024 4:00 PM · 2 days 1 hr. (Europe/Berlin)
Foyer D-G - 2nd floor
Project Poster
Bioinformatics and Life SciencesPerformance MeasurementSustainability and Energy Efficiency
Information
Poster is on display.
Application specific benchmark suites allow users and vendors in the specific field to evaluate hardware and software in a fair and transparent manner. There presently exists several genomics specific synthetic benchmarks but there are relatively few full application benchmarks with relevant and appropriate publicly available and shareable data. This deficiency is problematic as genomics workloads are inherently high throughput and an understanding of a whole system’s performance rather than just the CPU or GPU components is key. To that end we present the Sanger Benchmarking Suite, a modular and extensible genomics focused benchmarking suite which includes many of the programs used in modern genomics, open datasets to test them and validated results to ensure correctness. In line with the developing needs of the field and in response to rising power demands over recent years, we have included in addition to the usual CPU and IO performance statistics, perf driven energy and CO2 measurements to demonstrate the hidden costs of these relative to performance. Mindful of the sensitivity associated with genomics data, we have selected open datasets which have: the appropriate consent for public access; can be distributed freely and are modern enough to represent the challenges of modern genomics. This ensures that the benchmark can be performed anywhere and is available to both scientists and commercial entities so that vendors can run these benchmarks themselves and understand the reasons for their performance.
Contributors:
Application specific benchmark suites allow users and vendors in the specific field to evaluate hardware and software in a fair and transparent manner. There presently exists several genomics specific synthetic benchmarks but there are relatively few full application benchmarks with relevant and appropriate publicly available and shareable data. This deficiency is problematic as genomics workloads are inherently high throughput and an understanding of a whole system’s performance rather than just the CPU or GPU components is key. To that end we present the Sanger Benchmarking Suite, a modular and extensible genomics focused benchmarking suite which includes many of the programs used in modern genomics, open datasets to test them and validated results to ensure correctness. In line with the developing needs of the field and in response to rising power demands over recent years, we have included in addition to the usual CPU and IO performance statistics, perf driven energy and CO2 measurements to demonstrate the hidden costs of these relative to performance. Mindful of the sensitivity associated with genomics data, we have selected open datasets which have: the appropriate consent for public access; can be distributed freely and are modern enough to represent the challenges of modern genomics. This ensures that the benchmark can be performed anywhere and is available to both scientists and commercial entities so that vendors can run these benchmarks themselves and understand the reasons for their performance.
Contributors:
Format
On-site