TV Guide very slow vertical scrolling since RC3 (1 Viewer)

[disaster123]

As i don't know how to modify the for for more speed i tried to fix the double processing.
Simple changing:

      if (updateScreen)
      {
        if (iCurOff == _channelOffset)
        {
          Correct();
          UpdateCurrentProgram();
          return;
        }

        Correct();
        Update(false);
        SetFocus();

to

      if (updateScreen)
      {
        if (iCurOff == _channelOffset)
        {
          Correct();
          UpdateCurrentProgram();
          return;
        }

        Correct();
        // removed as update is already called before
        // Update(false);
        SetFocus();

works fine. No issues with tvguie until now and already a lot faster.

OK the slowest thing in the for loop is the RenderChannel function - and this one sets the retdot - so hopefully this will be solved by gibman's patch.

 

[disaster123]

nobody interested in my findings?

 

[doornjoostje]

I think it is great if the processing speed doubled.
The only remaining question is, why the Correct is performed there for the 2nd time.

 

[disaster123]

It's not the correct() it's the Update() function - but i don't know why it is called twice. I've remove the 2nd update from code since 3 days and hadn't any problem with epg.

 

[doornjoostje]

It's not the correct() it's the Update() function - but i don't know why it is called twice. I've remove the 2nd update from code since 3 days and hadn't any problem with epg.

You are right, Update() function.

Sounds very promising!
Would be nice if this could go into the next SVN release, so it can be thoroughly tested.

 

[framug]

As i don't know how to modify the for for more speed

Hello,
Did you try to track what instruction take time, into the for loop ?

i tried to fix the double processing.
No issues with tvguie until now and already a lot faster.

And just for knowing more about it, are you absolutely sure, there is no case where the first Update(false); is not called ?

OK the slowest thing in the for loop is the RenderChannel function - and this one sets the retdot - so hopefully this will be solved by gibman's patch.

Sorry but, there is something I don't understand here.
Maybe the gibman's patch could help the robby3 problem and, is probably a good enhancement but for your...
The retdot treatment in the RenderChannel function is in the foreach (Schedule record in _recordingList) and, you already tried without entering in this loop (because of empty _recordingList) and, you wrote it was not faster.
What do you think about ?

 

[disaster123]

i tried to fix the double processing.
No issues with tvguie until now and already a lot faster.

And just for knowing more about it, are you absolutely sure, there is no case where the first Update(false); is not called ?

No i'm not. I simply tested all buttons and variations i could imagine in tv guide - like scrolling, right, left, up, down, throught days, skip, .... I hoped somebody knows why it is called twice.

OK the slowest thing in the for loop is the RenderChannel function - and this one sets the retdot - so hopefully this will be solved by gibman's patch.

Sorry but, there is something I don't understand here.
Maybe the gibman's patch could help the robby3 problem and, is probably a good enhancement but for your...
The retdot treatment in the RenderChannel function is in the foreach (Schedule record in _recordingList) and, you already tried without entering in this loop (because of empty _recordingList) and, you wrote it was not faster.
What do you think about ?

[/QUOTE]
Yeah that's a little bit strange. I haven't measured the time and haven't tested it after removing the 2nd update. So perhaps my answer should be: also with an empty _recordingList is wasn't as fast as normal scrolling or so. Sorry for that. As gibman is already working on that i'll post my results when the patch is ready.

If you need anything more - ask me.

 

[robby3]

I'm a typical user, not a programmer so I can't say if this is a solution. Is it possible to send me a changed file so I can test the behaviour as well ? I opened the thread so I'm highly interested in any results

Thanks
Robby

 

[robby3]

Sorry for asking, but what is the satus on this? Will the problem be resolved in a future release?

Thanks
Robby