comp.arch.fpga - Field Programmable Gate Array based computing systems.
Hi, All! Need more info about system generator hardware co-simulation interface. Using JTAG for custom board very simply but extremely slow. How to support hardware co-simulation on custom FPGA boards with high speed interfaces: Ethernet, PCI, USB. For example: ML506 board Ethernet co-simulation files, from system generator plugins folder: eth_cosim_bitgen.opt eth_cosim_core.lna eth_cosim_core_synopsis eth_cosim_impl.opt eth_cosim_top.bmm eth_cosim_top.ucf ML506_PPEth_NonMmPorts.m ML506_PPEth_PostGeneration.m ML506_PPEth_Target.m xltarget.m How to build such plugin for custom Virtex board?
I am trying to run a simulink model simulation in definite time steps from MATLAB so that I can extract data from simulink, do some caluclations and send these new data back to Simulink for further simulation from the point I left off. The process can be descirbed below: Note that total simulation time is 600 seconds. 1. Start simulation from MATLAB for 100 seconds and then pause the simulation 2. Extract all the data I need from simulink and bring into MATLAB, do some calculations to changes simulink model parameters and then pass this new calculated data back to simulink. 3. Continue the simulation where I left off (i.e. from 100 seconds) but this time in 15 second intervals, up to total simulation time end (repeating step 2 and 3 after every 15 seconds). Can someone please advise if they have done this before and what command structures they used. Thank you. True.
Hello, My simulation group is setting up an Easy5-Simulink co-simulation for the first time. We have been running tests to compare the co-simulations and stand-alone Easy5 models. In one test, I set up the xi/EZ block to transmit 8 signals from the Easy5 model to the Simulink model. Four signals were sent from Simulink to the Easy5 model, but these signals were not used by the Easy5 model. In this test, ideally the stand-alone Easy5 model would produce the same simulation results as the Easy5-part of the Simulink-Easy5 co-simulation. Here is more info about the tests: Software: Easy5 2008 r1 MATLAB/Simulink r2008b Microsoft Visual Studio 2008 Intel Visual Fortran v10.1.030 Windows XP Easy5 integration method: Euler (10 ms time step) Simulink integration method: ode23tb (variable time step) There seem to be differences between the stand-alone simulation and the co-simulation (measured using gp/WR blocks). At time = .05 sec, Easy5 signal w_1_vl is 0.00991 in the co-simulation. The same signal is 0.05580 in the stand-alone simulation at .05 sec. In similar tests, we've noted oscillations in the co-simulation, but not in the equivalent stand-alone Easy5 simulation. We've arrived at these questions: 1. In the co-simulation, does Simulink know what kind of integration method the Easy5 model is using? 2. In the co-simulation, does Simulink use its own solver to integrate the Easy5 model? 3. Are there known differences (e.g. in terms of round-off error, integration methods) between using the co-simulation and running a stand-alone Easy5 model? In other words, are there limitations to be aware of when we run the Easy5-Simulink co-simulation? We usually run models using the Euler integration method, at 10 ms time step in real time. Any thoughts you have regarding these questions or general feedback (e.g. set-up suggestions) would be appreciated. Thank you, Venkatesh
Hello. I do not succeed to make the Simulink Co-simulation through parallel-door or with platform cable USB. How it is made?