We have long believed information is power, and with the emergence of FileMaker Go as the one of the best, if not the best information management platform for the iOS, we thought we would test to see just how much better it performs on the new iPad2. This has become particularly relevant for us with the release of our new product for non profit organizations, CCNonProfit.
We have a testing database that we use for internal purposes which we used to run these tests. It can run a sequence of routines, namely:
- Create Records
- Create a specified number of records with the window either frozen or not.
- Custom Function Loop
- A self-referential custom function looping a specified number of times.
- Empty Loop
- A looping script incrementing a local variable a specified number of times.
- Find Records
- Perform a find on all records in the sample data set on either an indexed or unindexed field, either locally or through a relationship, a specified number of times.
- Layout Change
- Jump between two layouts, either refreshing the layout or not, a specified number of times.
- Record Loop
- Loop through all the records is the sample data set, either refreshing the window on each new record or not.
- Sort Records
- Perform a sort on all records in the sample data set on either an indexed or unindexed field a specified number of times.
We have the database process a long loop of each event, and record the time it takes to process. We long the event in the database, and that way we can compare it across multiple platforms. All functions are done in a straightforward manner, and use only native FileMaker functions.
We built the tests using sample tables copied and pasted from the contact management template provided by FileMaker Inc. to keep the table structure simple and repeatable. All tests on the iPad and iPad2 were done with the device plugged in and no other applications running in the background, in case that ends up affecting anything.
My hope is that this can start a trend towards finding ways to efficiently code and design on the iOS platform.
Our results from our test are as follows, each cell is the time in loops per second the device could generate and the second chart is how many times faster than the iPad the device was. One important note, FileMaker can’t tell which iPad it is on, so the test results in the database are combined in the report, if you look, the faster ones are on the iPad2 😉
Test Results (Loops per second)
Bigger numbers are better
Machine Tested |
Create Records | Custom Function Loop | Empty Loop | Find Records | Layout Change | Record Loop | Sort Records |
---|---|---|---|---|---|---|---|
iPad | 51.72 | 1.3 | 2,439.02 | 1.46 | 1.09 | 1.10 | 0.79 |
iPad2 | 65.22 | 1.67 | 3,448.28 | 2.00 | 1.71 | 1.97 | 1.11 |
iMac 27″ (i7) | 600 | 10.71 | 38,974.36 | 20.46 | 30 | 29.70 | 10.34 |
Test Results (Times faster than original iPad)
Bigger numbers are betterMachine Tested | Create Records | Custom Function Loop | Empty Loop | Find Records | Layout Change | Record Loop | Sort Records |
---|---|---|---|---|---|---|---|
iPad | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
iPad2 | 1.26 | 1.28 | 1.42 | 1.37 | 1.57 | 1.79 | 1.41 |
iMac 27″ (i7) | 11.6 | 9.48 | 16 | 14.01 | 27.52 | 27 | 13.09 |
Of course we don’t expect you to want to compare the new iPad2 to our machine, we figured you’d want to see how your systems will compare relative to your machine, so below you’ll find a download link for the database we use to do the testing. It contains our baseline tests on the iMac, the iPad and the iPad2 for your own comparisons.
Grab the zipped FileMaker database here…
Enjoy 😉
Court, does the Layout Change test reflect the redraw speed change? I.e. does every layout change in the test wait for the layout to completely redraw? What is your subjective feeling of the redraw speed when rotating the iPad between landscape and portrait orientation with the preinstalled Inventory database open?
HOnza,
It doesn’t wait in our test, it fires the next script step as soon as FileMaker will do so natively. I think the layout draw speed is really one of the least efficient features. The interesting (and very helpful) feature is that the layout doesn’t redraw on rotate like it does on record or layout change.
Subjectively, the redraw doesn’t feel much faster, although the tests do show a very measurable bump…