Changeset 56 for anr/section-3.1.tex
- Timestamp:
- Feb 1, 2010, 6:07:27 PM (14 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
anr/section-3.1.tex
r31 r56 1 % vim:set spell: 2 % vim:spell spelllang=en: 3 1 4 Our project covers several critical domains in system design in order 2 5 to achieve high performance computing. Starting from a high level description we aim … … 4 7 5 8 \subsubsection{High Performance Computing} 6 Accelerating high-performance computing (HPC) applications with field-programmable 7 gate arrays (FPGAs) can potentially improve performance. 9 % Un marché bouffé par les archi GPGPU tel que le FERMI de NvidiaCUDA programming language 10 High-Performance Computing (HPC) world is composed of three main families of architectures: 11 many-core, GPGPU (General Purpose computation on Graphics Unit Processing) and FPGA. 12 The two first families are dominating the market by taking benefit 13 of the strength and influence of mass-market leaders (Intel, Nvidia) 14 %such as Intel for many-core CPU and Nvidia for GPGPU. 15 In this market, FPGA architectures are emerging and very promising. 16 By adapting architecture to the software, % (the opposite is done in the others families) 17 FPGAs architectures enable better performance 18 (typically between x10 and x100 accelerations) 19 while using smaller size and less energy (and heat). 8 20 However, using FPGAs presents significant challenges~\cite{hpc06a}. 9 21 First, the operating frequency of an FPGA is low compared to a high-end microprocessor. 10 22 Second, based on Amdahl law, HPC/FPGA application performance is unusually sensitive 11 23 to the implementation quality~\cite{hpc06b}. 12 Finally, High-performance computing programmers are a highly sophisticated but scarce 13 resource. Such programmers are expected to readily use new technology but lack the time 14 to learn a completely new skill such as logic design~\cite{hpc07a} . 15 \\ 24 % Thus, the performance strongly relies on the detected parallelism. 25 % (pour résumer les 2 derniers points) 26 Finally, efficient design methodology are required in order to 27 hide FPGA complexity and the underlying implantation subtleties to HPC users, 28 so that they don't have to change their habits and can have equivalent design productivity 29 than in others families~\cite{hpc07a}. 30 31 %état de l'art FPGA 16 32 HPC/FPGA hardware is only now emerging and in early commercial stages, 17 33 but these techniques have not yet caught up. 34 Industrial (Mitrionics~\cite{hpc08}, Gidel~\cite{hpc09}, Convey Computer~\cite{hpc10}) and academic (CHREC) 35 researches on HPC-FPGA are mainly conducted in the USA. 36 None of the approaches developed in these researches are fulfilling entirely the 37 challenges described above. For example, Convey Computer proposes application-specific instruction set extension of x86 cores in FPGA accelerator, 38 but extension generation is not automated and requires hardware design skills. 39 Mitrionics has an elegant solution based on a compute engine specifically 40 developed for high-performance execution in FPGAs. Unfortunately, the design flow 41 is based on a new programming language (mitrionC) implying designer efforts and poor portability. 42 % tool relying on operator libraries (XtremeData), 43 % Parle t-on de l'OPenFPGA consortium, dont le but est : "to accelerate the incorporation of reconfigurable computing technology in high-performance and enterprise applications" ? 44 18 45 Thus, much effort is required to develop design tools that translate high level 19 46 language programs to FPGA configurations. 47 Moreover, as already remarked in~\cite{hpc11}, Dynamic Partial Reconfiguration~\cite{hpc12} 48 (DPR, which enables changing a part of the FPGA, while the rest is still working) 49 appears very interesting for improving HPC performance as well as reducing required area. 20 50 21 51 \subsubsection{System Synthesis}
Note: See TracChangeset
for help on using the changeset viewer.