Elapsed time when logging

[michelpons]

When using COLTT 2.1 Development Build on a CAPE-OPEN UO in PRO/II 8.3, I have been surprised by the time elapsed between mouse actions. Just the fact of dropping the UO on the flowsheet leads to several minutes of hour-glass presence. Same after connecting one stream: it takes several minutes before being able to make another action. And so on.
I have compared this situation with the time elapsed when enabling COLTT 1.08.4 on the same UO. Everything takes then place within seconds so there is no much difference in overall elapsed time with and without COLTT enabled on the UO with that earlier version of COLTT.
For sure the log file with COLTT 2.1 is approximately twice the size of the log file with COLTT 1.08.4. Consequently I would expect an elapsed time of twice the size but the ratio is here between 100 and 1000.
The CAPE-OPEN UO is exhibiting a large number of parameters. Apart from that I don't see any other specificity. May be worth investigating because the situation with COLTT 2.1 leads to a too large difference in overall behaviour because of the delays between actions.

[michelpons]

Zipped log file obtained with COLTT 2.1 Development Build

[michelpons]

Zipped log file obtained with COLTT 1.08.4

[michelpons]

PRO/II 8.3 input file used as starting point

[michelpons]

Scenario used: open Unconnected.prz, drop a CAPE-OPEN UO on the flowsheet, choose GLCC UO, connect leftmost stream to UO inlet, connect each of the rightmost streams to an outlet, run the simulation. Success should be obtained.

[michaelhalloran]

Michel,

have you seen this problem with any other combinations of PME and PMC? I have removed all the \"release\" messages from COLTT output but I still see this problem. Further investigation shows:

• if I switch off parameter logging by disabling it in code, connection to GLCC in PRO/II are very fast
• With parameter logging on connections to GLCC in PRO/II are very slow
• With parameter logging on connections to GLCC in COFE are fast
• Connecting in the UI in PRO/II does not actually call Connect on a Port! That only happens when running the simulation
• Connecting to GLCC in the UI causes PRO/II to repeatedly loop through the whole parameter collection many times over. It looks as though the code loops through the whole collection from the start each time it wants to get to the next item. The GLCC UO has 44 parameters so this becomes very expensive. In the log attached there are 49 loops through the parameter collection between the last call to Save and the end of the file. As far as I can tell everything in the log between those two points was in response to connecting one stream.

So I think the problem is in PRO/II and it shows up because GLCC has a large parameter collection. Who can we contact at simsci to find out what's happening?

[michaelhalloran]

Log file which attempts to isolates what happens when PRO/II connects a stream to a UO. Everything after the last \"Save\" is due to connecting one stream

[michaelhalloran]

Michel,

I'm markied this Resolved - Not Fixable - PME Issue. I hope you are in agreement. I think Simsci need to comment on the behaviour accessing parameters that we see here.