Last Updated 8 Mar 2022 Difficulty level : Moderate
This is the first in a series of articles discussing various tests done to compare the efficiency of different approaches to coding.
Example databases are provided in each case so the same tests can be done on your own workstations
1. Handling Nulls: Trim / Len / Nz
Three different approaches to checking for 'nothing' in a control
2. CurrentDB vs DBEngine(0)(0)
Comparison of two methods used to reference the current database
3. DoEvents vs DBIdle.RefreshCache vs Sleep
Comparison of three methods used to pause code.
The purpose is to give the processor time to complete the previous task before continuing
4. HAVING vs WHERE
Compares the efficiency of using a WHERE clause or a HAVING clause in aggregate queries.
5. Conditional Updates
This compares 5 different ways of doing a conditional update where there are multiple conditions:
If/ElseIf/End If ; Select Case ; Nested IIf ; Switch ; Lookup table
6. Query vs SQL vs QueryDef
These tests compare the execution times using SQL statements, saved queries and query
definitions and discusses the advantages of each approach
7. Check Record Exists
These tests compare the time required to check the existence of a specified record in a large
dataset using 4 different methods.
8. Optimise Queries
This compares the effect of different ways of improving query performance
9. SELECT DISTINCT vs GROUP BY
This compares the efficiency of two methods of grouping data
10. Regex Or Not
This compares the efficiency of two methods of filtering data
11. $ Or No $
This compares the efficiency of two types of function for processing data e.g. Left$ vs Left
Also see these related articles:
a) Timer Comparison Tests
This article compares the accurary and consistency in the times measured using six different
methods including those used in the speed comparison tests
b) Show Plan - Run Faster
This article explains how the little documented Jet ShowPlan feature can be used to assist
with optimising queries & VBA SQL statements
c) Synchronise Data
This article discusses various ways of synchronising data with external tables. The times are
compared as well as the increase in file sizes associated with each method
1. Handling nulls: Trim / Len / Nz
Last Updated 27 Feb 2019 Return To Top
This was originally written in response to a question by Mister Chips at Utter Access forum:
Best Way To Check For 'nothing' In A Form Control
The OP wanted to know the best approach to checking if there is anything at all in a control (textbox, combo etc).
Three methods were suggested by the OP / Phil C & myself respectively
a) Trim e.g. If Trim(txtControl & "") = vbNullString Then
b) Len e.g. If Len(txtControl.Value & vbNullString) <> 0 Then
c) Nz e.g. If Nz(txtControl,"") < > " " Then
It was strongly suggested by several forum members that Len would be faster as
"'VBA handles numbers better then strings especially in comparing, so checking if Len(...) <> 0 is easier for VBA than comparing two strings'"
I was sceptical about this claim so decided to test all 3 methods on a local table with approx 2.6 million records.
In each case a text field was modified where it was null or a zero length string.
These were approximately 20% of the total records
I made sure nothing else was running during each test to try & ensure there were no other factors influencing the results. I ran the tests repeatedly for each of the methods but found little variation between each set of results
Tests were repeated on several other workstations. In each case, the order was the same
Nz was marginally faster than Len with Trim being the slowest
However, the differences were relatively small.
To my mind, this perhaps indicates that it is easy to get too bound up in optimisation worries due to the processing power of most modern computers.
I also ran the tests with the Accuracy field used in the test first unindexed & then again after indexing. The outcome was consistently slower for the indexed field – indexing speeds up searches but significantly slows down updates
Each result is the average of 3 tests though there was minimal variation in each test
Times were measured using the system timer (updated 64 times per second approx.) and rounded off to 2 d.p. (centiseconds)
I had intended to compare all 3 approaches looping through a recordset which I knew would be much slower. The first one was indeed VERY SLOW and eventually crashed the database as the file size approached the 2 GB limit. I abandoned the remaining recordset tests!
Click to download : Empty Fields Comparison Tests v1.2 Approx 25 MB (zipped)
NOTE This is a VERY LARGE download due to the table containing 2.6 million records
Dim sngStart As Single, sngEnd As Single
'first reset any existing data from previous tests
CurrentDb.Execute "UPDATE Postcodes SET Postcodes.Accuracy = Null WHERE Postcodes.Accuracy ='OK';"
sngStart = GetCurrentSystemTime
'run test using one of the following:
CurrentDb.Execute "UPDATE Postcodes SET Postcodes.Accuracy = 'OK' WHERE (((Trim([Accuracy] & ''))=''));"
'CurrentDb.Execute "UPDATE Postcodes SET Postcodes.Accuracy = 'OK' WHERE (((Len([Accuracy] & ''))=0));"
' CurrentDb.Execute "UPDATE Postcodes SET Postcodes.Accuracy = 'OK' WHERE ((Nz(Accuracy,'') =''));"
sngEnd = GetCurrentSystemTime
MsgBox "Time taken = " & Round((sngEnd - sngStart), 2) & " seconds"
The system time code is:
Option Compare Database
Private Type SYSTEMTIME
wYear As Integer
wMonth As Integer
wDayOfWeek As Integer
wDay As Integer
wHour As Integer
wMinute As Integer
wSecond As Integer
wMilliseconds As Integer
Declare Sub GetSystemTime Lib "kernel32" (lpSystemTime As SYSTEMTIME)
Function GetCurrentSystemTime() As Single
'Use this when accurate time differences to milliseconds are required
Dim tSystem As SYSTEMTIME
GetCurrentSystemTime = (1000 * Int(Timer) + tSystem.wMilliseconds) / 1000
As previously mentioned, system time is updated about 60 times per second - around 0.16 second intervals. So there is some potential error in the values but not as much as the time differences between the methods.
However, to me, the results indicate that in real life situations, there is minimal difference in the methods.
There are other ways of measuring time with precision
For example, you can use GetTickCount but that is also based on system time
The built in Timer function can be used to give time to centisecond accuracy though with the same limitation
You could also use the StopClock class code but this also appears to be millisecond accuracy
For a detailed comparison of the different approaches available for measuring time intervals, see my article Timer Comparison Tests
Colin Riddington Mendip Data Systems Last Updated 8 Mar 2022