Forum Discussion

Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
13 years ago

Is feasible to load both Altera and Xilinx libraries to same Modelsim？

I need use Quartus II and ISE for Altera and Xilinx FPGA respectively. I have never load libraries from Xilinx and Altera, both of them, into Modelsim. I just wonder whether this is feasible？ Will it cause some problems？

Thanks very much.

6 Replies

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    13 years ago

    --- Quote Start ---

    Yes you can. It shouldn't be a problem

    --- Quote End ---

    Thanks,Tricky. But I realized a problem. For loading Xilinx libraries, I use Simulation Library Compilation Wizard. After compilation, it generates a "modelsim.ini", it stores the directory for compiled libraries. For Altera libraries, I can use tcl script, or it seems latest Quartus also provide tool to compile library. Similarly, a "modelsim.ini" will be generated.

    My approach is copying new "modelsim.ini" to Modelsim directory to replace older one, then the compiled libraries will appear in Modelsim when I run Modelsim. However, now there are two new "modelsim.ini" for Altera and Xilinx, how can I do in Modelsim to map all these libraries?

    Thanks very much.
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    13 years ago

    At the top of modelsim.ini are the mappings to Modelsim libraries. You can ignore the generated modelsim.ini files, and create your own Tcl script that sets up these mappings.

    The Modelsim command is

    vmap <libname> <libpath>

    If you wanted to automate the process, have your Tcl script parse the generated modelsim.ini files, and issue vmap for each of the generated libraries. The calls to vmap will edit your master modelsim.ini file.

    Cheers,

    Dave
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    13 years ago

    --- Quote Start ---

    At the top of modelsim.ini are the mappings to Modelsim libraries. You can ignore the generated modelsim.ini files, and create your own Tcl script that sets up these mappings.

    The Modelsim command is

    vmap <libname> <libpath>

    If you wanted to automate the process, have your Tcl script parse the generated modelsim.ini files, and issue vmap for each of the generated libraries. The calls to vmap will edit your master modelsim.ini file.

    Cheers,

    Dave

    --- Quote End ---

    Thanks, Dave.

    I have another question. In each simulation work directory, it needs a "modelsim.ini" to tell Modelsim where the compiled libraries are. Why we still need a "modelsim.ini" in Modelsim installation directory? Or say in another way, what is the usage of "modelsim.ini" in Modelsim installation directory?

    Thanks very much.
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    13 years ago

    There should be a 'master' copy of modelsim.ini that is read-only in the install area.

    When you call vmap the first time, Modelsim will detect that the master copy is read-only and create a local version for your project.

    What this means however is that you end up with a modelsim.ini per simulation. I personally don't like this, so have setup environment variables to force Modelsim to use a single copy of modelsim.ini. The variables are discussed in an Appendix in the Modelsim manual.

    For your tests, just deal with the fact that you need to have a modelsim.ini in the area that you are working. Make sure that the versions generated by Altera and Xilinx are somewhere else. Use vmap to populate your local modelsim.ini (or manually copy and paste the mappings).

    Cheers,

    Dave