Curious what is going on when MC does a full redraw.

Questions about MultiCharts and user contributed studies.
bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Curious what is going on when MC does a full redraw.

Postby bowlesj3 » 15 Jul 2008

I just added two studies to all my charts. I may be pushing MC a bit because now on the 1 minute chart if I make an adjustment that requires the recalculate command it does a full redraw many times (meaning all text and plots the study is putting out blank out then they come back in). If I turn these new studies off the full redraw does not occur when the recalculate is issued.

MC seems fine over all with this happening but I am just curious what MC is doing and why it is doing this. I suspect many may have the same curiosity.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

1 study off redraws only part.

Postby bowlesj3 » 16 Jul 2008

I was experimenting with this.

If both new studies are turned on, the original study fully blanks and redraws when I issue the recalculate in the original study.

If only one new study is turned on, the original study only partially blanks and redraws when I issue the recalculate in the original study.

If both new studies are turned off, the original study does not blank at all when I issue the recalculate in the original study.

All the above is done on the one minute bars and the original study has a lot of text, and lines.

If I apply the same 3 studies to the 5 minute bars to repeat this experiment where on that chart there is a lot less text and no lines at all, turning on the two new studies and issuing exactly the same recalculate to that 5 minute chart does not trigger a redraw at all.

So it seems to be a memory (or whatever) resource limitation relating to the amount of text and lines one can place on the chart.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Another episode in this soap opera

Postby bowlesj3 » 16 Jul 2008

For those who may be interested, here is episode #3 in this soap opera :-)

I went back to the original study and took out some code that was displaying missing bar text (text which is no longer useful too me and now I have a good reason to get rid of it). When this text was removed and both of those new studies are turned on, the recalculate will many times only do partial redraws. It is somewhat random now but it seems there is never a full redraw. Prior to removing that unused text, having both studies on would always trigger a full redraw of everything on the chart.

Again the conclusion is there is a memory resource limit on the amount of text, lines, arrows, etc that can be put on a chart and when it is exceeded recalculate will trigger a screen blanking and redraw. So the question is does it cause any other problems such as unexplained aborts, unexplained bog downs in speed here and there? Last but not least, when a new release comes out are these resource limitions in need of expansion. I am running a gig of memory on a AMD dual core chip. MC and all my other apps are only using about 50% of that. So it appears that this restriction is internal to MC.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Final Episode and question.

Postby bowlesj3 » 16 Jul 2008

Final soap opera episode #4.

I tried reducing a related pair of newly installed arrays in the original study from 10 elements to 5 but this did not produce any noticable change from the last episode. The two new studies are actually the same study duplicated in the chart with different parameter settings. I could not reduce the array sizes in that study.

I tried reducing the number of bars of price data way down. It speeds the recalculation but does not change the blanking out and redrawing at all.

The question is, I do not suppose anyone knows of any other ways to reduce memory usage specific to a chart or even generally. A forum search does not find anything.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

A little more

Postby bowlesj3 » 17 Jul 2008

Maybe this is a fluck.

I did a recalculate today on the 1 minute bars with this new study on the chart (two copies of it that is) and I did not get a blank and redraw but just a normal recalculate. What was different however is the fact that I had turned this study off on a few of the other charts. Seeing as this study is new I have left both copies of it turned on for all the charts (01, 05, 10, 15,30,60).

Maybe these new studies will increase the load and cause more aborts when the recalculate is issued (increasing it from 1 a day roughly to a higher amount). I will let you know if I notice a difference. This is why I turned it off on a few of them. I thought that maybe this was happening since I had more aborts today than normal.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Seems to have fixed it.

Postby bowlesj3 » 18 Jul 2008

The "recalculate command full blanking (and partial blanking) and redraw" seems to have stopped. The only thing that I have changed that may have taken it back to normal from partial blanking is only turning on one copy of this new study on the larger charts (05,10,15,30 and 60 minutes). I have to leave both copies of this study turned on for the 1 minute chart. Putting the second copy on the other charts is needed less often.

So maybe this "limited resource issue" is workspace specific rather than chart specific.
Last edited by bowlesj3 on 18 Jul 2008, edited 1 time in total.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

LiveChat Referene thread.

Postby bowlesj3 » 18 Jul 2008

Pertaining to (contacting TS-Support via livechat and getting on helpdesk to inspect these aborts I get which appear to be related to the recalculate command), I will dump any info that may have relavance to this issue into this thread so all the info will be in one place and known before I contact you.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Last theory out the window.

Postby bowlesj3 » 23 Jul 2008

It seems that my last theory is out the window regarding why MC at times does a blank and redraw when the recalculate is issued.

Today it started doing a partial blank and redraw after not having done it for about 3 or 4 days at least (since my last post actually). The reason I say the theory is out the window is I only have this new study turned on once for the 30 minute bars, once for the 5 minute bars and twice for the 1 minute bars yet it is still doing it. Since it was not doing it the last few days I left these studies on all the bars at least once so I tried turning as many of them off as I could when it started to blank and redraw again. There is nothing new either that I can think of. I am purposely trying to avoid all programming and get back to trading now that all my systems are finished.

As far as aborts when using the recalculate, I have only been getting them in the afternoon (1 or 2 of them a day). Normally I figure the recalculate runs about 50 to 100 times a day.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Few aborts but 2 in a row.

Postby bowlesj3 » 14 Aug 2008

I am not sure if this will have any value but MC 2.1.999.999 has been running without any aborts for several days at a time lately over the last few weeks. However yesterday it aborted two times in a row at its typical location in my trading cycle where it is forced to execute the recalculate command. I have never seen it do that before over the last year. Prior to the last few weeks it was aborting about once every day or two. For the most part the same EL code is running as stated in prior posts. Outside of MC I am also running exactly the same programs too. Today it ran all day without problem so yesterday was very unusual.

User avatar
Marina Pashkova
Posts: 2758
Joined: 27 Jul 2007

Postby Marina Pashkova » 15 Aug 2008

Hi John,

Have you tried the latest 3.1 and 4.0 versions? I mean, if you're comfortable with 2.1 you don't really need to upgrade. However, 3.1 and 4.0 have more feauters and more bugs fixed. They work faster as well.

Regards.

bowlesj3
Posts: 2180
Joined: 21 Jul 2007
Has thanked: 227 times
Been thanked: 429 times

Postby bowlesj3 » 15 Aug 2008

Hi Marina,

I haven't tried MC 3.0 since beta 2. MC2.1.999.999 has actually been performing well except that one unusual double abort (before and after for two days). I am a little too busy to try and upgrade at the moment. I also find the features you are adding at the moment are mostly for full computer system traders and I have no interest in this (I have not run a backtest since 1998 and will never do so again). I am a specialist and have no interest in the scanner. I figure that in time TS-Support will finish these types of changes and at that time the features going in may be of greater interest. For now I am really quite satisfied with MC. I have worked around many of the shortcomings with my programming experience. As long as it does not abort very often I am happy.

I am also too busy to get around to clearing the item stopping me from okaying the case study.

Thanks,
John.


Return to “MultiCharts”