Changeset 339 for anr/section-etat-de-art.tex
- Timestamp:
- Jan 31, 2011, 3:32:59 PM (13 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
anr/section-etat-de-art.tex
r319 r339 44 44 researches on HPC-FPGA are mainly conducted in the USA. 45 45 None of the approaches developed in these researches are fulfilling entirely the 46 challenges described above. For example, Convey Computer proposes application-specific instruction set extension of x86 cores in FPGA accelerator,46 challenges described above. For example, Convey Computer proposes application-specific instruction set extension of x86 cores in an FPGA accelerator, 47 47 but extension generation is not automated and requires hardware design skills. 48 48 Mitrionics has an elegant solution based on a compute engine specifically … … 74 74 tool will then automatically generate synthesizable Hardware Description 75 75 Language (HDL) code mapped to \xilinx pre-optimized algorithms. 76 However, this tool targets only DSP basedalgorithms, \xilinx FPGAs and76 However, this tool targets onlysignal processing algorithms, \xilinx FPGAs and 77 77 cannot handle a complete SoC. Thus, it is not really a system synthesis tool. 78 78 \\ 79 79 In the opposite, SOPC Builder~\cite{spoc-builder} from \altera and \xilinx 80 Platform Studio XPS from \xilinx allow sto describe a system, to synthesize it,80 Platform Studio XPS from \xilinx allow to describe a system, to synthesize it, 81 81 to program it into a target FPGA and to upload a software application. 82 82 Both SOPC Builder and XPS, allow designers to select and parameterize components from … … 126 126 Low power consumption constraint which is mandatory for embedded systems is not yet 127 127 well handled or not handled at all by the HLS tools already available, 128 \item The parallelism is extracted from initial specification.128 \item The parallelism is extracted from the initial specification. 129 129 To get more parallelism or to reduce the amount of required memory in the SoC, the user 130 130 must re-write the algorithmic specification while there are techniques such as polyedric … … 166 166 This approach however has a severe weakness, since it also significantly reduces 167 167 opportunities for achieving good speedups (most speedups remain between 1.5x and 168 2.5x), since ISEs performance is generally tied downby I/O constraints as168 2.5x), since ISEs performance is generally limited by I/O constraints as 169 169 they generally rely on the main CPU register file to access data. 170 170 … … 219 219 As a side effect, it has been observed that the polyhedral model is a useful 220 220 tool for many other optimization, like memory reduction and locality 221 improvement. Another point is221 improvement. Itshould be noted 222 222 that the polyhedral model \emph{stricto sensu} applies only to 223 223 very regular programs. Its extension to more general programs is … … 245 245 are members of the consortium and the board is incorporating top actors 246 246 (STM, NXP, TI, ARM, FREESCALE, LSI, Mentor, Synopsys and Cadence), ensuring the 247 wide adoption by industry. Initiatives have already % work for (paul)247 wide adoption by industry. Initiatives have already 248 248 attempted to extend this standard 249 to AMS IPs packaging domain (MEDEA+ Beyond Dreams Project) and to Hardware Dependent249 to the AMS IPs packaging domain (MEDEA+ Beyond Dreams Project) and to Hardware Dependent 250 250 Software layers (MEDEA+ SoftSoc project) and Accellera is reusing these results for 251 251 further releases.
Note: See TracChangeset
for help on using the changeset viewer.