Do POI files impact performance?

 

Does the number of POI files impact performance of the TOMTOM while driving? Or maybe the number of POI catagories that you selected to be displayed?

I've been adding quite a few POI files to my TOMTOM Start 55TM and noticed that it doesn't track as closely as it used to. When I make a turn I'm usually about a half a block on the new road before TOMTOM makes the turn and is pointing me on the new road.

Just curious if there is a trade off between performance and convenience.

BTW, I currently havd 58 POI files loaded on my TOMTOM.

Jim

--
Jim F.

Loaded Vs Displayed

I doubt that the number of POI files you have loaded will make any difference, but the number of categories to be displayed could. That is because it is trying to re-locate and draw those icons on your screen.

--
Shooter N32 39 W97 25 VIA 1535TM, Lexus built-in, TomTom Go

What he said

Shooter wrote:

I doubt that the number of POI files you have loaded will make any difference, but the number of categories to be displayed could. That is because it is trying to re-locate and draw those icons on your screen.

Yup. Nice explanation.

With best wishes,
- Tom -

--
XXL540, GO LIVE 1535, GO 620

OV2 Performance

The specifications for OV2 files allow for the use of "skipper" records to improve performance in situations where a file has a large number of locations. When skipper records are used, closely grouped locations inside the file are bundled together and preceded by a skipper record. This allows your GPS to quickly skip over a bunch of locations if they're nowhere near your current location.

Right now, the OV2 files from POI factory have just one skipper record for the entire file; so you might see some performance issues if you load several files that have a very large number of locations.

VersatileGuy did some stress testing last year with 40 copies of the WiFi file and found very substantial performance degradation with a single skipper record.

There's an online OV2 optimizer tool at this web page:

http://www.samposoft.com/tools/default.aspx

You might try reloading some of the larger files after running them through the online optimizer to see if it makes a difference in your situation.

Eventually, I'd like to incorporate a feature like this into the program that generates the OV2 files here at POI factory.

Jonathan

Thanks !!!

JM wrote:

The specifications for OV2 files allow for the use of "skipper" records to improve performance in situations where a file has a large number of locations. When skipper records are used, closely grouped locations inside the file are bundled together and preceded by a skipper record. This allows your GPS to quickly skip over a bunch of locations if they're nowhere near your current location.

Right now, the OV2 files from POI factory have just one skipper record for the entire file; so you might see some performance issues if you load several files that have a very large number of locations.

VersatileGuy did some stress testing last year with 40 copies of the WiFi file and found very substantial performance degradation with a single skipper record.

There's an online OV2 optimizer tool at this web page:

http://www.samposoft.com/tools/default.aspx

I was unaware of that factor, let alone the optimization software. I have quite a few POI files loaded on my GPS, but normally have almost all of them selected "Off" except when I expect to use a specific file, because of the performance impact.

Thank you very, very much for the tip.

With best wishes,
- Tom -

--
XXL540, GO LIVE 1535, GO 620

Thanks JM!

I just ran my largest files through the process (both largest in size (KB) and number of records).

I'll see if there is a noticeable difference in performance.

It is strange that the output file was the same size as the input file. Normally when I update an ov2 file (delete or add a location) and copy it to my TOMTOM, I get a warning that the file already exists and Windows gives me the option to overwrite the file or not. Widows will give stats on both files and identify which is larger. When I copied the optimized files it didn't say either the old or the new file was larger. It just seemed odd to me, like nothing was changed.

Jim

--
Jim F.

?????????????????????

?????????????????????

--
nuvi 250 --> 1250T --> 265T Lost my 1250T

optimized file

jim407 wrote:

...It is strange that the output file was the same size as the input file. Normally when I update an ov2 file (delete or add a location) and copy it to my TOMTOM, I get a warning that the file already exists and Windows gives me the option to overwrite the file or not. Widows will give stats on both files and identify which is larger. When I copied the optimized files it didn't say either the old or the new file was larger. It just seemed odd to me, like nothing was changed.

Jim

The new optimized file should be slightly larger and will have "_fixed" appended to the end of the file name. If your old and new files are exactly the same size, you might want to double check everything and run the process again.

Jonathan

Retested...

I ran the McDonalds POI file through the optimizer again and the output had the "_fixed" at the end of the file name, but when I compared file sizes the input and output were exactly the same.

This time I looked at the properties of the two files and compared the size in actual bytes.

Makes me wonder if it does anything or if the distribution of locations need to be a certain way. Groups of locations separated by empty space, East cost vs West cost with little in between? Or maybe locations on various islands? The McDonalds locations are everywhere!!

--
Jim F.

different results

jim407 wrote:

I ran the McDonalds POI file through the optimizer again and the output had the "_fixed" at the end of the file name, but when I compared file sizes the input and output were exactly the same. ...

Just ran the McDonalds .ov2 through, right clicked both files, and chose properties from the pop-up menus. Here are the sizes I found.

Original .ov2: 603,292 bytes
Optimized .ov2: 646,258 bytes

slightly larger, as I would expect.

Jonathan

Maybe this is why

I typically download the Garmin csv file from the POI factory because it usually has more information and I can format the data how I like. (I typically do Name@Address>Phone, so the address is included in the ov2 format.)

Once I have the data formatted and in a csv file I run it through POIEdit 2007 to convert it to an ov2 formmat file. Maybe POIEdit 2007 does the optimization so nothing is changed when I use the online tool.

JM, did you get my icons? Are they just low on the priority list? Is there a way I can post them myself?

Jim

--
Jim F.

POIEdit / Skippers

Hmm. I'm not sure what POIEdit does for skipper records, so it still might be worth trying out the online optimizer tool on your larger files to see if it makes any difference.

I did get the icons, and I'll get them posted later this week after the camera research is done and the updated files are published.

Thanks,

Jonathan

Mystery solved

Apparently POIEdit optimizes ov2 files by adding skipper records, exactly the same way the web page does.

I took the csv file that I have for Subway and McDonalds and created an ov2 from it using makeov2.exe. I surprized how long it took to process the file and create the output, but I imagine it is an old process/program.

I then optimized the output ov2 file using the web page and compared it to an ov2 created with POIEdit using the same csv file.

The file sizes of the ov2 produced by POIEdit and optimized by the web site were exactly the same and the ov2 file created by makeov2.exe was actually larger. OK, so maybe the mystery isn't completely solved wink

--
Jim F.

.

jim407 wrote:

the ov2 file created by makeov2.exe was actually larger. OK, so maybe the mystery isn't completely solved wink

The OV2 optimizer at

http://www.samposoft.com/tools/default.aspx

does indeed create multiple skipper records. I confirmed this by downloading the Future Shop OV2 file at

http://www.poi-factory.com/poifile/download/ov2?node=23687

and then running it through the optimizer:

11,191 Future_Shop.ov2
11,485 Future_Shop_fixed.ov2

The optimizer added 14 skipper records at 21 bytes each = 294 bytes, which is the difference in the size of the files.

The old makeov2.exe utility added considerably more skipper records (49 of them) because it uses a different algorithm for segmenting the POI records. If you still see performance degradation with your POIEdit OV2 files (which are apparently the same as the OV2 optimizer files) then you could try using makeov2.exe to create the OV2 files from your CSV files and see if they work better. That information could be very helpful to JM if he decides to modify the OV2 algorithm here at The Factory.

Performance is better

I reduced the number of POIs that are displayed and it seemed to improve performance. Since I used POIEdit to create the ov2 files they were already optimized so that wasn't a factor.

--
Jim F.

.

jim407 wrote:

Since I used POIEdit to create the ov2 files they were already optimized so that wasn't a factor.

We don't know that for sure because the POIEdit OV2 files and the "optimized" OV2 files have far fewer skipper records than the OV2 files produced by makeov2.exe. As JM mentioned, I tested my GO720 with over 2.5 million POIs in categories that were set to display and it worked fine when the OV2 files were created by makeov2.exe.

Interesting

jim407 wrote:

I reduced the number of POIs that are displayed and it seemed to improve performance. Since I used POIEdit to create the ov2 files they were already optimized so that wasn't a factor.

Very interesting... who would have thunk it?

--
http://www.poi-factory.com/node/21626 - red light cameras do not work

More testing...

I just converted 55 of my csv files to ov2 files using makeov2.exe. I added all of them as icons to be displayed. Now it is time for a road test smile to see how it performs.

--
Jim F.

Road test....

It is difficult to make an objective observation about something as subjective as performance.

But after loading the ov2 file generated from makeov2.exe and marking over 60 icons to be displayed, I thought the performace was better than when I did the same with ov2 files generated from POIEdit.

The performace wasn't that much better, but it seemed to be better. It wasn't as good as when I reduced the number of icons to be displayed with ov2 files generated from POIEdit.

So based on my limited subjective testing I would say there may be a slight performace gain from using makeov2.exe over POIEdit to generate ov2 files.

Jim

--
Jim F.

Sometimes and to Varying Degrees

If I installed huge POI files on my 60CSx, it would slow down. I then trimmed the POI file(s) and all was well.

Recently, on the Montana 600, it appears to be very picky. My POI, waypoints and routes that I have used for over 5 years on the 60CSx aren't working on the Montana 600, as if it isn't backward compatible.

--
When you are dead, you don’t know that you are dead. It is only difficult for the others. It is the same when you are stupid.

@jim407

Thanks for taking the time to do some testing and report the results. When I get home I'll try to do another torture test using "optimized" OV2 files and see if I can quantify the difference in performance between those and OV2 files created by makeov2.exe.

.

VersatileGuy wrote:

I'll try to do another torture test using "optimized" OV2 files and see if I can quantify the difference in performance between those and OV2 files created by makeov2.exe.

I did a quick test this afternoon, and even with ~2.75 million POIs the difference in performance between the two variants of OV2 files was negligible on my GO 720. Of course, "your mileage may vary"....

@VersatileGuy

Thanks for testing and reporting back.

Jonathan

OV2 Optimizer utility for Windows

TomTom owners using Windows might be interested in my little "ov2optimizer" utility:

http://www.poi-factory.com/node/37978

@JM

JM wrote:

Thanks for testing and reporting back.

Jonathan

Doesn't he get a "light bulb" for his work?

--
Illiterate? Write for free help.

I gave him one!

Box Car wrote:
JM wrote:

Thanks for testing and reporting back.

Jonathan

Doesn't he get a "light bulb" for his work?

Thank you, VersatileGuy! I know you and JM have been talking about this for awhile. This will be so helpful!

~Angela

Note to JM (all your badges are belong to me!)razz

Thanks!

Glad to help.

been waiting for that one

~Angela wrote:

Note to JM (all your badges are belong to me!)razz:

all your base are belong to us A declaration of victory or superiority. The phrase stems from a 1991 adaptation of Toaplan's "Zero Wing" shoot-'em-up arcade game

--
Illiterate? Write for free help.

Light bulb?

GlobeTurtle wrote:
Box Car wrote:
JM wrote:

Thanks for testing and reporting back.

Jonathan

Doesn't he get a "light bulb" for his work?

Thank you, VersatileGuy! I know you and JM have been talking about this for awhile. This will be so helpful!

~Angela

Note to JM (all your badges are belong to me!)razz

Is the light bulb a temporary badge? I don't see it associated with my user name.

--
Jim F.

Software Contributor Badge

jim407 wrote:

Is the light bulb a temporary badge? I don't see it associated with my user name.

The "light bulb" (Software Contributor) badge is a permanent badge, like the one for POI Contributors. It goes to people who create apps/programs for the community (like ov2optimizer or EPE).

Jonathan