Richard,
Please see inline...
Best regards,
Ilya
This posting is provided "AS IS" with no warranties, and confers no rights.
--------------------
> From: "Richard Kucia" <(E-Mail Removed)>
> Subject: CF SP2 and ReadXML performance
> Date: Tue, 2 Mar 2004 17:18:05 -0500
> Lines: 42
> X-Priority: 3
> X-MSMail-Priority: Normal
> X-Newsreader: Microsoft Outlook Express 6.00.2800.1158
> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165
> Message-ID: <(E-Mail Removed)>
> Newsgroups: microsoft.public.dotnet.framework.compactframework
> NNTP-Posting-Host: 146.cleveland-11-13rs.oh.dial-access.att.net
12.75.70.146
> Path: cpmsftngxa06.phx.gbl!TK2MSFTNGP08.phx.gbl!TK2MSFTNGP10.phx.gbl
> Xref: cpmsftngxa06.phx.gbl
microsoft.public.dotnet.framework.compactframework:47475
> X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework
>
> My app's startup time has always been horrible, so I was looking forward
> with great anticipation to CF SP2.
> The app reads a single XML file once during its startup, and it stores the
> contents in various objects.
>
> I installed CF SP2 (non-developer) on my Dell Axim X5 and timed ReadXML
in a
> variety of ways. I created a two files, one with and one without an
embedded
> schema. Then I tried various combinations of ReadXML and XMLReadMode. Here
> are the results, in milliseconds:
>
> File and XMLReadMode Elapsed Time in ms
> -------------------------- -------------------
> NoSchema + Auto 14.129
Schema inferred, this is the slowest way to go and it should be generally
avoided.
However, in case of very small XML it could be faster than loading schema.
> Schema + Auto 6.782
Schema is loaded first; data is loaded based on schema.
This is second best to use. Best one is to have schema already in the
DataSet.
Schema code is a big and complicated one; it takes few seconds to JIT and
execute it.
> Schema + Ignore .358
This is fast as it's doing nothing.
Since you do not have schema in the DataSet and you choose to ignore schema
in XML, no data is actually loaded.
> Schema + ReadSchema 5.630
Same as Schema + Auto. If you running tests in a batch, it might appear a
bit faster as code is preJITed already.
> NoSchema + Auto 12.711
Schema inferred. Again, might appear faster because code has been preJITed.
> NoSchema + Infer 12.146
Same as above, done in about the same time.
>
> Typical code for any one of these cases is:
> RaiseEvent InitializationStepEvent("User: Reference file: No Schema +
Auto")
>
> ReferenceDataSet = New System.Data.DataSet("reference")
>
> fsXML = New System.IO.FileStream(clsCommon.AppPath & "Reference With No
> Schema.xml", System.IO.FileMode.Open, IO.FileAccess.Read)
>
> myXmlReader = New System.Xml.XmlTextReader(fsXML) ' Create an
XmlTextReader
> to read the file
>
> ReferenceDataSet.ReadXml(myXmlReader, System.Data.XmlReadMode.InferSchema)
>
>
> I believe the Axim is running CF CLR 1.0.3316.0
>
> Can anyone offer an explanation for the truly wide disparity among these
> results? I'd be happy to send the files to anyone wishing to reproduce
these
> test results.
There's no disparity, it's in fact very consistent.
It takes 12-14 seconds to infer schema, 5-7 seconds to load schema and
data, .4 seconds to scan the file and do nothing.
If you preJIT the code, time will probably match to a fractions of a second.
>
> Richard Kucia
>
>
>
|
Similar Threads
