MapForums

I have an application that uses the Mappoint2004 Control to plot pushpins in the mappoint application. We have found that mappoint preforms very very slowly on Dual core or dual processor machines running XP, 2000 server or server 2003. I have tried setting the process affinity for Mappoint.exe and it still seems to be very slow on these machines. If I disable the processor in hardware, then it runs fine.

Has anyone else experienced mappoint performance problems on dual core machines?

Wow, no, this is new to me.. good finding.. can you explain it more?

Eric

__________________

I have the same problem with an application built using the MapPoint 2006 control.

On a Pentium 4 machine, Windows XP Pro with SP2, 1 GB of RAM, the application can draw 500 line shapes in a few seconds.

On a "faster" dual core T7700 machine, Windows XP Pro with SP2, 2 GB of RAM, the same application draws the same 500 line shapes in about 45 seconds.

The dual core machine performs most other operations faster than the Pentium 4 machine.

Also, the draw time on the Pentium 4 seems linearly proportional to the number of shapes, but the dual core machine draws the 500th shape much slower than the draw time of the first shape.

Has anyone else seen this behavior? Any ideas on what is happening or how to resolve? Thanks -JW

Hi,

Very strange. Maybe video card slow? There is an windows API call LockWindowUpdate(). You can use this to prevent writing to the window until last line is draw. Btw this speed up a lot things if you have to draw many.

__________________
rgds, Wilfried Mestdagh
www.mestdagh.biz
MapPoint coding demo
Order MapPoint 2009 with Routing and User Tools Spreadsheet

Wilfried,

Here is some interesting related information. When I enable hyper-threading on the Pentium 4 machine, the shapes draw more than an order of magnitude slower than when HT is disabled. So a draw time of 500 shapes goes from seconds with HT disabled to a couple of minutes with HT enabled. Nothing else is changed - just enabling or disabling HT.

Other draw and display operations appear unaffected. Drawing 1000 pushpins in a dataset takes about one second whether HT is enabled or disabled. Shape drawing is the only case I can see so far where HT has an impact on MapPoint performance.

I suspect the HT problem has the same underlying cause as the slow shape draw time on dual core machines. I don't understand much about multi-threading so I'm at a loss to determine a solution. I could try and disable one of the processors on the dual core machine but that seems like a silly solution.

Do the HT timing results provide any insight into what is happening?

Thanks -JW

Hi,

No sorry this gives me not more insight. My advice is to try LockWindowUpdate() while drawing. It speedup performance enormoes for every program that has to write a lot of graphical data.

__________________
rgds, Wilfried Mestdagh
www.mestdagh.biz
MapPoint coding demo
Order MapPoint 2009 with Routing and User Tools Spreadsheet

I have encountered the same issue with the mp ocx being slow at drawing on a dual core machine. MP simply cannot keep up with my requests (unless I set the App and ocx to run on the same CPU). I also noticed the performance is significantly worse when using the flat view vs globe view...

Since my customer is still running MP2004 on a Pentium 4 (single core) machine, I havent had to provide him with a solution yet, but he is planning to upgrade his systems later this year, so my clock is now ticking...It will kind of defeat the purpose of the upgrade if he can only utilize one of the CPU cores.

Does anyone have any more info? Any word on if this might be fixed in the next version? The feeling I get is that MS is not doing much to fix or enhance the desktop version of the product.

My experience was with MP02 on a dual P3 machine- I did not have any noticable performance changes running on 1 or 2 CPUs

__________________
Mike Mueller
Lannon Fire Department

However, if there is a locking problem then it makes sense that Hyperthreading is worse than single cpu or double cpu.

HT adds any potential slowness from threading (eg. locking) and at the same time adds only a small potential increase (from parallel processing). When HT was first released, there were a lot of reports of it making no or negative difference for many applications.

Richard

__________________
Winwaed Software Technology LLC
http://www.winwaed.com
See http://www.mapping-tools.com for MapPoint Tools
Pre-Order MapPoint 2009 today: http://www.mapping-tools.com/mappoint2009

It is almost certainly the same issue between HT and dual core. I first saw this when I tried to run on an XP machine with HT enabled a few years ago. The only way it would run was to disable HT. With DC, I must set the affinity of both the app and mappoint to the same CPU (or of course disable DC).

I have been hoping eventually a MP update would be able to run in such an environment (let alone take advantage of additional resources), but so far that is not the case. Maybe the silence on the next MP release means MS is working feverishly to address this... or more likely, they havent decided whether to ever release another version...