How to count number of paths in RTL-compiler group?

Hi dkos,

 Unfortunately, I currently have no idea how to count the number of paths in a group, but for checking if it's empty, you could do one of the following:

1. report timing -lint 
If the path_group is empty, it will be listed as exception, which is not affecting timing.

2. report timing -paths [eval [get_attribute paths <path-group name>]]
If you get "No paths found" the path group is empty.

Not very elegant, but those are all the ideas I have at the moment ...

Regards,
Johannes

 Thanks for answer Johannes,

 This solution works. And do you have an idea how to detect if path group empty or not in tcl script? Let me clarify my problem a little:

 I automatically create cost groups between all clocks in design in script. If there are 2 clocks in design my script will create following cost groups:

    inputs_2_clk1, clk1_to_clk1,clk1_to_outputs, clk1_to_clk2,

    inputs_2_ clk2, clk2_to_clk2,clk2_to_outputs, clk2_to_clk1

Althougth some groups are empty RTL compiler creates them. I suppose that this leads to memory usage increase and dramatically increase final output *.sdc file. So, I want to find a way to delete or not to create empty cost groups before synthesis starts...I have a feeling that I need to use linux 'grep' command on results of rc 'report timing -paths [eval [get_attribute paths <path-group name>]]' command, but I don't know how... 

Thank, dkos 

 P.S. script that I use for cost group creation:

#delete cost groups, created by create_clock commands
puts "----------------------------------------------------------------------------"
puts "Delete cost groups, created by create_clock commands and add own cost_groups"
puts "----------------------------------------------------------------------------"
rm [find / -cost *]

set CLOCK_LIST [find / -clock *]
#collect registers for all clocks
set i 0
foreach clock ${CLOCK_LIST} {
  set clock [basename $clock]
  puts "-- collect all regs for clock: ${clock}"
 
  set regs_of_clock($i) [all des seqs -clock ${clock}]
  set regs_col_clock [llength $regs_of_clock($i)]
  puts "-- number of regs: ${regs_col_clock}"
  incr i
}
  set clock_num [llength ${CLOCK_LIST}]
  puts "-- number of clocks: $clock_num"

#generate new cost groups
set i 0
foreach clock ${CLOCK_LIST} {
  set clock [basename $clock]
  puts "-- generate costgroups for clock: ${clock} --"
  puts "-------------------------------------------------------"

  set inList [all des inps -clock ${clock}]
  set outList [all des outs -clock ${clock}]
 
  if {[llength $regs_of_clock($i)] > 0} {
    #I2R paths
    if {[llength $inList] > 0} {
      puts "  -- generate costgroups for clock: I2${clock}"
      define_cost_group -name "I2${clock}"
      path_group -mode System -from $inList -to $regs_of_clock($i) -group I2${clock} -name I2${clock}
    }
    #R2O paths
    if {[llength $outList] > 0} {
      puts "  -- generate costgroups for clock: ${clock}2O"
      define_cost_group -name "${clock}2O"
      path_group -mode System -from $regs_of_clock($i) -to $outList -group ${clock}2O -name ${clock}2O
    }
    #R2R paths
    for {set j 0} {$j < $clock_num} {incr j} {
      set clockB [lindex ${CLOCK_LIST} $j]
      set clockB [basename $clockB]
      puts "  -- generate costgroups for clock: ${clock}->${clockB}"
   
      define_cost_group -name "${clock}2${clockB}"
      path_group -mode System -from $regs_of_clock($i) -to $regs_of_clock($j) -group ${clock}2${clockB} -name ${clock}2${clockB}
    }
  }
  incr i
}

Again you helped me very much Johannes! Thanks a lot!

One minor correction in order that code work correct.  Line

    if {$TMP eq "No paths found"} {

should be replaced with

    if {$TMP eq "No paths found."} {

i.e. dot must be added to expression.

 Hi dkos,

another alternative if you are up to some scripting is to rely on the timing_info attribute associated with every startpoint and endpoint. In addition you can use the startpoint and endpoint attribute to identify which pins are which. This attribute will provide you with the cost_group information for every endpoint as well as all exception associated with it. Amonst all the exception is the path_group information. As you can tell, it is essential to be clear on the distinction between path_group and cost_group to get the script to work right. Another caveat I would keep in mind is the use of multi-mode constraints since different mode could result in different cost groups as well. Anyhow, hope this helped

gh-