Changeset 304 for anr/section-project-task-schedule.tex
- Timestamp:
- Dec 23, 2010, 11:53:37 AM (14 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
anr/section-project-task-schedule.tex
r300 r304 14 14 \definecolor{gtcTaskBG1} {rgb}{0.90,0.99,0.7} 15 15 \definecolor{gtcMilestone}{rgb}{0.9,0.4,0.4} 16 \immediate\write\ganttdata{ML=6 ML=12 ML=18 ML=2 4}16 \immediate\write\ganttdata{ML=6 ML=12 ML=18 ML=27} 17 17 %\def\ganttlabelstyle#1{\begin{small}#1\end{small}} 18 18 \def\ganttlabelstyle#1{\begin{small}\hyperlink{#1}{#1}\end{small}} … … 49 49 %The figures~\ref{gantt1}~\&~\ref{gantt2} present the Gantt diagram of the project. 50 50 Before the final release (T0+36), there are 4 milestones (red lines on the figures) at 51 $T0+6$, $T0+12$, $T0+18$ and $T0+2 4$ that are rendez-vous points of the precedent51 $T0+6$, $T0+12$, $T0+18$ and $T0+27$ that are rendez-vous points of the precedent 52 52 deliverables. 53 53 \begin{description} 54 \item[Milestone 1 ($T0+6$)] Specification of COACH inputs, of the \xcoach format and of 54 \item[Milestone 1 ($T0+6$)] 55 Specification of COACH inputs, of the \xcoach format and of 55 56 the demonstatrors as a reference software. 56 \item[Milestone 2 ($T0+12$)] The first COACH release. At this step the demonstrators are 57 written in the COACH input format. This COACH release allows to prototype and to generate the FPGA-SoC. 57 \item[Milestone 2 ($T0+12$)] 58 The first COACH release. At this step the demonstrators are 59 written in the COACH input format. This COACH release allows to prototype 60 and to generate the FPGA-SoC. 58 61 The main restrictions are: 59 62 1) Only the neutral architectural template is supported, … … 61 64 3) Enhanced communication schemes are not available. 62 65 4) ASIP compilation flow is not available. 63 \item[Milestone 3 ($T0+18$)] The second COACH release. At this step most of the COACH 64 features are availables. A preliminary version of the ASIP synthesis flow is supported, for a 65 simple extensible MIPS model. The main restriction is that COACH can not yet 66 generate FPGA-SoC for \altera and \xilinx architectural templates. 66 \item[Milestone 3 ($T0+18$)] 67 The second COACH release. At this step most of the COACH features are available. 68 A preliminary version of the ASIP synthesis flow is supported, for a 69 simple extensible MIPS model. The main restriction is that COACH can not yet 70 generate FPGA-SoC for \altera and \xilinx architectural templates. 67 71 The others restriction is that the HAS tools are not yet fully operational. 68 \item[Milestone 4 ($T0+24$)] The pre-release of the COACH project. The full design flow is 69 supported. 70 The main restriction are: 71 1) The backend HAS tools have not been yet enhanced, 72 2) Dynamic partial reconfiguration is not supported, 73 3) NIOS processor instruction set extension is supported, but only for user specified patterns. 72 \item[Milestone 4 ($T0+27$)] 73 The pre-release of the COACH project. The full design flow is supported. 74 The main restrictions are: 75 1) Automatic frequency calibration of coprocessor is not available. 76 2) Automatic HPC set up is not yet available. 77 3) NIOS processor instruction set extension is supported, but only for user 78 specified patterns. 79 4) GAUT enhencements are not available. 74 80 \item[Final Release ($T0+36$)] 75 81 76 82 \end{description} 77 This organisation allows the project to globally progress step by step mixing development78 and demonstrator deliverables.79 Hence, demonstrator feed-back will arrive early and so the risk to point out incompatibility80 at the integration phase is significantly reduced.83 This organisation allows the project to globally progress step by step mixing 84 development and demonstrator deliverables. 85 Hence, demonstrator feed-back will arrive early and so the risk to point out 86 incompatibility at the integration phase is significantly reduced. 81 87 \par 82 88 The risks that have been identified at the beginning of the project are the following: 83 89 \begin{description} 84 \item[\xcoach format ({\NOVERSspecXcoachDoc}, {\NOVERSspecXcoachToCA})] 85 Partners have to agree on a convenient exchange format for all tools involved. 86 Because all the HAS tools rely on it, the \xcoach format specification is a 87 crucial step. There are no work-around but as mentionned in 88 section~\ref{xcoach-problem} (page~\pageref{xcoach-problem}) the five academic partners have worked on it 89 for a full year and a preliminary document already exists. 90 \item[\xcoach format ({\NOVERSspecXcoachDoc}, {\NOVERSspecXcoachToCA})] 91 Partners have to agree on a convenient exchange format for all tools involved. 92 Because all the HAS tools rely on it, the \xcoach format specification is a 93 crucial step. 94 There are no work-around but as mentionned in section~\ref{xcoach-problem} 95 (page~\pageref{xcoach-problem}) the five academic partners have worked on it 96 for a full year and a preliminary document already exists. 90 97 %\item[\xcoachplus format (\novers{\specXcoachDoc}, 91 98 % \novers{\specXcoachToSystemC}, \novers{\specXcoachToVhdl})] … … 93 100 % By centralizing the coprocessor generation, it guarantees their functioning 94 101 % independently of the used HAS tools. 95 % Our experience with UGH and GAUT give us confidence in the succes of this 96 % task. 97 \item[Virtual prototyping of \altera \& \xilinx architectural templates ({\NOVERScsgImplementation})] 98 The SoCLib component library contains several SystemC models used for the virtual 99 prototyping of the \altera and \xilinx architectural templates (NIOS and Microblaze processor cores). 100 Nevertheless, at this time we do not know how many IP cores SystemC simulation models have to be developped. 101 If the workload of this simulation model development is too important, virtual prototyping 102 of those architectural templates will not be directly supported. 103 The three architectural templates being quite similar, the virtual 104 prototyping will use the neutral architectural template. 105 \item[VCI/AVALON \& VCI/PLB bridges ({\NOVERShpcAvalonBridge}, {\NOVERShpcPlbBridge})] 106 If one of these tasks is impossible or too important or leads to inefficiency, 107 it will be abandoned. 108 In this case, the neutral architectural template will not be available for HPC and 109 a SystemC VCI model corresponding to the PCI/X IP will be developped to allow 110 virtual prototyping. 102 % Our experience with UGH and GAUT give us confidence in the succes of this 103 % task. 104 \item[Virtual prototyping ({\NOVERScsgImplementation})] 105 In this project, only the virtual prototyping of the neutral architectural 106 template is supported. 107 We think that this restriction is not a serious problem. 108 Indeed the \altera \& \xilinx architectural templates being architecturally close of 109 the neutral architectural template, an efficient software/hardware partition 110 on the neutral architectural template is also an efficient on the other 111 architectural templates. 112 The project will allow to verify experimentally this assumption. 113 % \item[Virtual prototyping of \altera \& \xilinx architectural templates 114 % ({\NOVERScsgImplementation})] 115 % The SoCLib component library contains several SystemC models used for the 116 % virtual prototyping of the \altera and \xilinx architectural templates 117 % (NIOS and Microblaze processor cores). 118 % Nevertheless, at this time we do not know how many IP cores SystemC 119 % simulation models have to be developped. 120 % If the workload of this simulation model development is too important, 121 % virtual prototyping of those architectural templates will not be directly 122 % supported. 123 % The three architectural templates being quite similar, the virtual 124 % prototyping will use the neutral architectural template. 125 \item[VCI/AVALON \& VCI/PLB bridges ({\NOVERShpcAvalonBridge}, {\NOVERShpcPlbBridge})] 126 These bridges may decrease the efficiency of the \altera \& \xilinx 127 architectural templates. 128 Developing the communication components (MWMR) for the AVALON and PLB buses 129 will correct this problem. 130 % If one of these tasks is impossible or too important or leads to inefficiency, 131 % it will be abandoned. 132 % In this case, the neutral architectural template will not be available for HPC and 133 % a SystemC VCI model corresponding to the PCI/X IP will be developped to allow 134 % virtual prototyping. 111 135 \end{description} 112 136 \parlf
Note: See TracChangeset
for help on using the changeset viewer.