GemStone is an innovative full system performance that consists of five software tools that, together, identify and evaluate the sources of error in gem5 models against a reference hardware (HW) platform by utilising statistical and machine learning approaches. Independently, the tools serve many purposes: Automating the running of experiments and collection of Performance Monitoring Counters […]
PRiME has made available another free, open source software tool developed as part of its research activities. PRiME Code Generation tool is a Rodin* plug-in that facilitates automatic code generation of Runtime Management software from verified Event-B formal models. The tool is an extension to the existing Rodin code generation plugin (Tasking Event-B). The tool has […]
PRiME is pleased to announce the release of PRiME Framework which enables application- and platform-agnostic runtime management. Being able to balance the competing needs of application performance with the minimisation of system power is essential for future systems and one way to achieve this is with application-aware runtime management of system behaviour. There are many runtime management […]
PRiME has made available another free, open source software tool developed as part of its research activities. KOCL (“KAPow for OpenCL”) is PRiME’s automated tool flow for generating FPGA designs capable of accelerating OpenCL kernel execution whilst reporting their own power consumption. The tool was created by PRiME researchers at Imperial College as part of […]
PRiME’s collaborative research has been featured in the latest issue of the prestigious IEEE Computer Magazine (August 2017). The article describes the work performed by three of PRiME’s research institutes, led by Newcastle University, studying the interplay between the performance, energy, and reliability (PER) of parallel-computing systems. It describes methods supporting the meaningful crossplatform analysis […]
On 21st September 2016, Dr. Domenico Balsamo presented a paper on PRiME’s research on embedded processor power modelling at the “PATMOS 2016” conference (Power And Timing Modeling, Optimization and Simulation), held in Bremen, Germany. PATMOS is an annual conference that has been running for over 25 years, and was one of the first research conferences […]
PRiME is pleased to announce the release of PRiME Stereo Match software – a heterogeneous and fully parallel stereo matching algorithm for video depth estimation, available in both OpenCL and C++. Stereo Matching is based on the disparity estimation algorithm. This is an algorithm which is designed to calculate 3D depth information about a scene […]
PRiME PhD researcher and PowMon developer, Matt Walker, hosted a day long tutorial, ‘Building Online Power Models from Real Data’ at 2016 IEEE International Symposium on Performance Analysis of Systems and Software, Uppsala, Sweden. Working in collaboration with colleagues; Vasileios Spiliopoulos (Uppsala University, Sweden) and Stephan Diestelhorst (Staff Research Engineer at ARM, Cambridge, UK), Matt presented an […]
Modelling the interplay between Performance, Energy and Reliability (PER) in multi- and many-core systems has been one of the main objectives within the PRiME project. This has led to the development of the idea to represent PER interplay as a region of reliable operation in the throughput-voltage space. This region is defined by platform-specific constraints […]
Accurately estimating CPU power consumption is a key requirement for controlling CPUs – e.g. for implementing energy saving techniques – and for exploring the CPU design space. PRiME researchers have developed a tool , PowMon, that uses models built and validated from real, measured data from an actual device. This means that the tool accuracy […]