Forum Discussion
Altera_Forum
Honored Contributor
14 years agoalternate command for 'set_case_analysis'
I want to know if there is any alternate command for 'set_case_analysis' as this tcl command is not working in altera(time quest). This command is used to constrain a particular pin/port to a particular value when doing timing analysis.
eg: "set_case_analysis 0 port1" This command assigns 0 to port1 in timing analysis. Is there any alternate way to do this in altera?3 Replies
- Altera_Forum
Honored Contributor
TimeQuest doesn't support it. Could you elaborate what you're trying to do and maybe there's another approach?
- Altera_Forum
Honored Contributor
Take the example
"set_case_analysis 0 port1" This pin,port1 is a static pin. It wont change for one complete run(but it is not tied to 0). But above all it is in a clock path and not a data path. So i do not need a false path there. This pin is the select line of the mux in the clock path. So i need to constraint this pin to a particular value(0 here) so that only one particular path is taken while analyzing. Since it is in the clock path i cant use false path command. One alternate method i can do is to tie that port to 0 and 1 and do the complete set up(synthesizing, place & route) for both and see if timing is met when tied to 0. But here the problem here is if there are "n" such situations, i will be having 2^n bit maps at the end. - Altera_Forum
Honored Contributor
The alternate method won't work because you'll be analyzing completely different place-and-routes, where you'll really only be running one image.
For clock muxes, the select line can generally come from: - A top-level port - A regisiter - A combinatorial node based off a clock(and perhaps a combination of the previous two) If it comes from a port, then it won't be analyzed as a clock unless you put a clock constraint on it, so you should be fine. If it comes from a register, clocks don't propogate through registers unless you put a generated clock assignment on it, so again, you should be fine. If it comes from a combinatorial node, then the registers/ports upstream from that comb node will not get analyzed as clocks for the reasons above, and I doubt an actual clock would drive directly into the select line of a clock mux(let me know if I'm wrong). I've analyzed numerous clock muxes without this constraint. Please let me know if I'm overlooking something.