Vania Marangozova-Martin

Performance analysis of HPC applications on low-power embedded platforms

This paper presents performance evaluation and analysis of well-known HPC applications and benchmarks running on low-power embedded platforms. The performance to power consumption ratios are compared to classical x86 systems. Scalability studies have been conducted on the Mont-Blanc Tibidabo cluster.We have also investigated optimization opportunities and pitfalls induced by the use of these new platforms, and proposed optimization strategies based on auto-tuning.

Towards a Component-Based Observation of MPSoC

Motivated by the increasing heterogeneity and complexity of MPSoC systems, we propose a component-based generic approach for MPSoC observation. We show that components help in observing all software levels from system to application. We present the EMBera prototype and relate our experience in implementing it on two different platforms: a Linux-based 16-core SMP machine and a 5-core embedded system developed by STMicroelectronics.

Analysis and Tracing of Applications Based on Software Transactional Memory on Multicore Architectures

Transactional Memory (TM) is a new programming paradigm that offers an alternative to traditional lock-based concurrency mechanisms. It offers a higher-level programming interface and promises to greatly simplify the development of correct concurrent applications on multicore architectures. However, simplicity often comes with an important performance deterioration and given the variety of TM implementations it is still a challenge to know what kind of applications can really take advantage of TM. In order to gain some insight on these issues, helping developers to understand and improve the performance of TM applications, we propose a generic approach for collecting and tracing relevant information about transactions. Our solution can be applied to different Software Transactional Memory (STM) libraries and applications as it does not modify neither the target application nor the STM library source codes. We show that the collected information can be helpful in order to comprehend the performance of TM applications.

Optimizing 3d convolutions for wavelet transforms on CPUs with SSE units and GPUs

Optimizing convolution operators is an important issue as they are used in numerous domains including electromagnetic computations, image processing and nanosimuations. In this paper we present our optimizations for 3D convolutions in the BigDFT nanosimulation software. We focus on processors with vector units and on GPU acceleration and experiment with several architectures. Exploiting the relation between algorithmic specifics and hardware architecture, we obtain performance gains of around x2 on CPU and up to x20 on GPU.

Debugging component-based embedded applications

With the advent of parallel architectures in the domain of embedded systems, developing applications fully adapted to their underlying platform becomes more and more complicated. Finding and fixing bugs in such environments is even trickier. Furthermore, current complex applications cannot be developed from scratch, only based on programming language primitives. They have to follow advanced programming models and their execution will be driven by the APIs of the key underlying libraries. In this paper, we propose a new approach for source-level debuggers. Going beyond their long-established ability to support sequential programming languages, we describe the functionalities a debugger should be able to provide to debug embedded and parallel component-based applications. Then we demonstrate our solution to this problem with a debugger targeting the component framework used on an MPSoC platform. We also explain the development challenges we faced during the implementation of this GDB-based debugger and illustrate its efficiency though a case study of an image processing application.

A Generic Component-Based Approach to MPSoC Observation

Observing the execution of MPSoC is a key activity in debugging and optimizing embedded applications. However, providing suitable observation tools becomes a real challenge with the fast evolution of embedded platforms. We propose a generic component-based approach which allows for partial and configurable MPSoC observation. Genericity is obtained by encapsulating specific embedded features and exporting a generic observation API. Partial observation is achieved by instantiating and attaching observation components only to the entities of interest: application modules, HW components or different levels of the SW stack. Configurability is related to the possibility to organize and parametrize observation treatments (data collection, filtering, storing) according to the target system. The approach has been validated in several application and platform contexts including an embedded platform built by STMicroelectronics.