VT 2014

[fphil]

From my experience (Samsung NP-NC110 (10"), win7 Starter) the 2014 versions of VT are deadfully slow to the point that one gets the win7 message "VemsTune does not respond" and  "Do you want to continue to wait?)

Checking for a possible over busy state, it appears that memory is 40%, process 30% ... No, VT is sleeping or having a smoke outside, talking with the secretary or busy at the coffee machine

Looking further, the log file (or the monitor when launching VT-c.exe) says:
"Execute time: 7852 ms for function: GaugeGroupManagerSQL::showGroup"
and in any case it takes 1300ms to popup the welcome message!!

Indeed win7 and VT are not best friends. So I have to use the November 2013 version and I stays with fw 1.2.17.

[mattias]

Developers should focus on code profiling to discover what modules/functions are the real cpu cycle black holes.

Last time I was in Budapest I made them check and they made a big improvement, but that was two years ago. They should have this in their development cycle, and perhaps it is but nobody is telling them how/what to fix.

Personally I've noticed that the log viewer has gotten painfully slow.

[fphil]

Developers should focus on code profiling to discover what modules/functions are the real cpu cycle black holes.

+1.

[VEMS]

Hi Guys,

Improving VemsTune performance is very high on our the priority list. After wrapping up the new VemsDisplay versions with analog channels support (raw, calibrated, using configured predefined profiles as in VT) we will have the time to start resolving these performance issues.

But please bear with us, some issues which might seem relatively easy to pinpoint are not so easy to resolve; Profiling will indeed show the most time spend in, most cpu used by, most mem consumed by, etc. However, how to rewrite complex interwoven pieces of code to make them perform better, its shows not

Best regards, Dave

[fphil]

Profiling will indeed show the most time spend in, most cpu used by, most mem consumed by, etc.

To test, I ask VT to compute the"statistics" for one MAP serie data, 6800 samples or so, (less than 400kB vemslog). After 75s I got the max, min and mean of the serie...

If I was a customer I would not give acceptance for that software. It would be a shame since this is a nice (big) piece of software with my many nice features and functions.

What about a VT lite?, do you really need to go through the SQL data base (if it is the case).

Sorry Dave, I am in a bad mood after 2-3 days playing with VT log viewer , the knock function (Eculogger should be disble") and the bootloader update.

Regards

Philippe