Do you read the website "Daily Dose of Excel"? If not, it is a very interesting website about Excel. Frequently updated and many interesting discussions.
Today I want to share an article with you:
This week Dick Kusleika posted a very interesting article where he quoted Charles Williams from DecisionModels.com. Charles Williams is an absolute expert in the field of the performance of Excel models.
Recently Charles found out that Google Desktop can significantly slow down Excel:
Charles Williams - For the last few weeks I have been trying to find out why clearing a large range of cells caused Excel to hang on some PCs but not on others. This does not happen with Excel 97, but does with Excel 2000, 2002, 2003 and 2007.
It turns out that its not just Clear, but also Delete or transferring data from a variant to a range, or even just selecting a large range of cells. The further down the sheet and to the right the slower it gets. And the more recent the Excel version the slower it gets.
Someone on the newsgroups discovered that, when using VBA, you could bypass this problem by switching off EnableEvents, and then someone else discovered that the culprit was Google Desktop Search.
The problem actually turns out to be the Google Desktop Office Com Addin. When you deactivate this you get a miraculous speedup.
With Excel 2007 it is fairly easy to deactivate:
Office Button>Excel Options>Addins>Com Addins and deselect Google Desktop Office Addin
With earlier versions of Excel you have to customise a toolbar and add the Com Addins dialog to it.
View>Toolbars>Customise>Commands tab>Tools then about halfway down you will find COM Addins, select and drag to the toolbar of your choice.
Then you can uncheck Google Desktop Office addin.
If you have multiple versions of Excel installed you only have to do this once.
Presumably this COM addin sets up one or more application-level events to monitor things like Selection Change and Worksheet change and then tries to trap the change in order to index it.
If you want to measure this effect you can download a Variant Benchmark Timer from my website that allows you to run a read and write benchmark with and without EnableEvents.
This represents an interesting new twist in the Google-Microsoft wars!
I could reproduce the delay when I removed data/cleared cells in a (very) large dataset. However with small datasets of no more then 10.000 cells I didn't see any significant differences in speed. Removing over 1 million cells (thank you Excel 2007) however took more then several minutes with Google Desktop, compared to next-to-nothing without having it installed.
If you have any comments on this subject, please post them in the original article.