Forum Discussion

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

Identical inputs different outputs

Hi!

I have ran Quartus II on two different computers with the same OS (Windows XP), the same version of Quartus II, identical source code, constraints and scripts.

On one of the computers everything passed (synthesis, place & route, timing analysis etc.), but on the other computer the timing failed. I have used "high effort" on both computers.

My theory for this that different seeds are used on the different computers, but shouldn't the effort level handle this?

How do I get the same result in both computers?

5 Replies

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

    Quartus II should give the same results with the same inputs on the same OS. (Linux and Windows give different versions, but just a seed affect). There is no "random" input to the process. I rely on this all the time and see it work. (On the flip side, the smallest change, such as changing the name of a signal, could give you different results).

    You might want to do a diff on the reports. The first one, .mao.rpt, is what will probably show it, and if there are any differences, it would make sense.
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor
    16 years ago

    Does this mean that the smallest change such as a constant value, ex build number could effect the place & route in such a severe manor that timing constraints not are met even though the margin in the working case is 1.5MHz on a 33MHz clk?

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

    I think that's ~2ns difference, which I would say is high, but if your datapath is 30ns, variation could account for that. So yes, it could. Note that there are bazillions of possible solutions, so any small change will send it down a different path and cause variation. All fitters have this affect.

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

    The use of design partitions and logiclock will allow you to limit the area that is re-fitted to what has changed and keep everything else the same.

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

    I think that Quartus will also give different results when run on 32-bit and 64-bit versions of Windows.