Comments for Excel and UDF Performance Stuff

Comment on TEXT vs VALUE vs VALUE2 – Slow TEXT and how to avoid it by fastexcel

fastexcel — Tue, 04 Jun 2013 17:38:35 +0000

Hi Dan, It works Ok for me: put =1/0 in cell A1 and then run this.
Sub Dan() Dim var As Variant Dim var2 As Variant var = Range("A1").Value var2 = Range("A1").Value2 MsgBox IsError(var2) MsgBox IsError(var) End Sub

Comment on TEXT vs VALUE vs VALUE2 – Slow TEXT and how to avoid it by Dan

Dan — Tue, 04 Jun 2013 14:48:29 +0000

Hi Fastexcel – your method doesnt work as .Value2 converts errors such as #DIV/0! into a number (-2146826281) and if you test IsError(-2146826281) you get false … so it doesnt flag as an error

Comment on Table Extension recalculation bug in Excel 2010 but not 2007 by fastexcel

fastexcel — Fri, 24 May 2013 09:40:08 +0000

As of 24 May 2013 this bug is still there in Excel 2010 and Excel 2013

Comment on SpeedTools AVLOOKUP2 & MEMLOOKUP versus VLOOKUP – Performance, Power and Ease-of-Use Shootout Part 2 by kalx

kalx — Tue, 09 Apr 2013 20:02:59 +0000

Hi Charles,

Have you done any testing to compare with DAX? That seems to be a similar tool. Microsoft seems to be leaving MDX to benign neglect and pushing this now.

Best,
Keith

Comment on SpeedTools Beta 3 – Win a SpeedTools Coffee Mug and SpeedTools License by fastexcel

fastexcel — Tue, 19 Mar 2013 13:39:03 +0000

Download and install SpeedTools from http://www.decisionmodels.com/fastexcelV3SpeedTools.htm then you will find it in the Speedtools Function library on the Formulas Tab

Comment on SpeedTools Beta 3 – Win a SpeedTools Coffee Mug and SpeedTools License by BlackBerry Curve 9220

BlackBerry Curve 9220 — Tue, 19 Mar 2013 13:17:13 +0000

How to find “MEMLOOKUP” formula in excel2007. I did many times but not reflect in my ms-excel. can anybody suggest me? How to get it.

Comment on SpeedTools AVLOOKUP2 & MEMLOOKUP versus VLOOKUP – Performance, Power and Ease-of-Use Shootout Part 2 by fastexcel

fastexcel — Fri, 15 Mar 2013 13:15:54 +0000

Hi Jan Karel – excellent question!

If the data changes to create an additional duplicated match then memory lookup will return the result from the same row as last time (row 47 in the example. This is not the same behaviour as VLOOKUP/MATCH).

If you tell MEMLOOKUP/AVLOOKUP that the data is sorted then SpeedTools does not use memory match anyway so the behaviour is the same as VLOOKUP/MATCH (except that the default is to do an exact match with sorted data).

If you need to control which of the duplicated matches to use you can use AVLOOKUPNTH or AMATCHNTH, but in that case the lookup memory cannot be used. The data can be sorted or unsorted with AVLOOKUPNTH (sorted is much faster than unsorted).

Comment on SpeedTools AVLOOKUP2 & MEMLOOKUP versus VLOOKUP – Performance, Power and Ease-of-Use Shootout Part 2 by Jan Karel Pieterse

Jan Karel Pieterse — Fri, 15 Mar 2013 12:55:03 +0000

Hi Charles,

What happens if data other than the one on row 47 changes, causing a new match to occur on -say- row 10?

Comment on TEXT vs VALUE vs VALUE2 – Slow TEXT and how to avoid it by Peter Bartholomew

Peter Bartholomew — Thu, 14 Mar 2013 10:29:55 +0000

I tried the .Text property to move zip codes (especially those with a leading zero) in the knowledge that the code would be validated as text in the receiving workbook. It was ill-conceived though because, on input, the .Value2 property recognised the text format correctly and I still couldn’t write the text property to the output range. I finally settled for assigning an apostrophe as String, concatenated with the Value2 using “&”.

Any better ideas?

Comment on SpeedTools AVLOOKUP2 & MEMLOOKUP versus VLOOKUP – Performance, Power and Ease-of-Use Shootout Part 1 by fastexcel

fastexcel — Thu, 28 Feb 2013 18:54:15 +0000

Hi Sam,
Thats very clever: I have not seen the hi-lo trick before.
But as you say it is very slow.

The closest SpeedTools equivalents are either

2 AMATCHNTH formulas, one finding the first and the second finding the last,

or a better solution is FILTER.MATCH which returns start row index and number of consecutive rows for each cluster of consecutive matches so that its easy to use OFFSET to get the filtered subsets.

Both of these are much faster (about 30 times faster on a test over 70K rows) than your clever MATCH array formula.
And FILTER.MATCH allows for multiple criteria matching like SUMIFS but more powerful.

The FastExcel SpeedTools functions use Linear Search on unsorted data, but preceded with an indexed Lookup Memory search where appropriate.