Pan operation lags behind the cursor

[BjørnarSundsbø]

When I perform a pan operation without releasing the mouse button, and continiously drag the map around, the image lags behind the cursor, and sometimes "stutters". The reason I really noticed this was that the test application for map functionality does not experience the same issue. I did discover some heavy logic in the MouseMove event that I have now disabled if the left mouse button is down. I have been unable find any other events that I have hooked up that could be causing this. ANTS Performance Profiler does not indicate anything in my code. The only heavy load it refers to is native code related to CopyPixels.

I'm not sure this is a problem that is directly related to the API, and I realize the reason could be anything anywhere in my code, but I'm trying to find out if someone else has experienced the same, and how this may have been addressed.

I do not use accelerated display. I tried removing all overlay on the map, such as menus and other gadgets to no avail. I also tried removing all GraphicsLayers in case these were the cause. The experience is the same as if you use AcceleratedDisplayLayer to hold the background map, and have  no acceleration on the GraphicsLayers on top. The feeling then is that the objects are floating on top of the map in a very peculiar way; they do not have the same speed as the cursor and background map.

[AnttiKajanus1]

How is memory usage with the application while this happens? I remember seeing similar case on my professional services days where it ended to be related FeatureLayer and it's Snapshot mode (loads all graphics to memory while on-demand keeps only what is visible). Thou it sounds that you have already tested that.

Have you managed to reproduce the issue in simpler application with same basemap service? Does this happen if you use different basemap service? One thing that you could try to do also is to see how long tile requests takes and/or if there is any other requests running around wildly while pan occurs.

Have you managed to repro this issue with other computers?

What SDK version you are using? You could try to upgrade to latest if you aren't using it.

[BjørnarSundsbø]

Thank you for your reply. I haven't had time to look much deeper into this issue as I've been busy with 'Esri Incident #1270209' (related to this control) and a release. My simple sample application doesn't contain much of the application code I am experiencing the issue in, yet.

Memory usage is at approximately 400MB, which I don't feel is frightening, but could of course be related. I do experience this without any graphics (or graphics layers). Hiding all layers but one still produce the same effect. I haven't had time to recreate the issue in a smaller application, as mentioned before. Will do so when I find the time. This issue is (sadly) not critical to the user, but is something that is really bugging me personally 🙂

Tile request is rather quick, and I don't see any temporarily empty tiles when panning. As mentioned, I think I see the same effect when zooming (ZoomDuration is 200ms).

I am running the latest version of the SDK.

If I discover the reason I will post it here. I am still open to suggestions to other reasons for this.

[MichaelBranscomb]

Hi,

When you notice the stuttering map - are you using the accelerated display mode and have a debugger attached? Unfortunately there is a known issue with stuttering tiles in debug mode (something to do with low level .NET exceptions which we have not yet been able to resolve, although it does only occur when a debugger is attached and therefore is a relatively minor issue).

If using the AcceleratedDisplayLayers group layer unfortunately you will notice some unavoidable lag between the accelerated and non accelerated layers due to the fundamentally different rendering pipelines. This is much more apparent on older, slower machines.

Cheers

Mike

[BjørnarSundsbø]

No accelerated display in use, though the experience is the same as described in the release notes. I will confirm tomorrow that the experience is identical when enabling this in the sample application. I only have access to remote desktop at the moment, so it will not be the same.