Emulator Performance

[Robert Bouillon]

I'm on a 2.4 GHz P4, 1GB DDR3200 2-3-2-2, I875P, SATA RAID0.

My emulator performs most operations at the speed of a 233MHz ARM I have.

Stepping into the program is slow: Loading the stack or local/watch windows
can take up to 30 seconds. Accessing variables can take up to 10 seconds
from within Visual Studio. I get a 1-2 second delay each step, stepping
through the program.

I'm running the latest service packs on 2kServer.

Is this performance typical? Is there any way to tweak this performance? I
would think a full-power x86 proc would perform significantly faster than a
low-power ARM. If not, are there any plans to improve the speed of the
Emulator for .NET?

Thanks.

--ROBERT

 

[Paul G. Tobey [eMVP]]

It's emulating the entire machine (there's a virtual PC running as a process
under Win2k, emulating hardware and running a second OS, Windows CE, on top
of which the CLR is running, etc.); it shouldn't be a surprise that it
doesn't run at native code speeds. All performance testing should be done
on the real device.

Paul T.

 

[Robert Bouillon]

Thanks for the response, but you really didn't answer any of my questions.

Is this performance typical? Can I tweak the emulator for better
performance? Are there plans to improve the speed (i.e. has anyone played
with .NET CF 2.0?).

When stepping through the application, my processor hangs around 10%, so I
don't think it's a matter of the emulator having "too much to do."

--ROBERT

 

[Ilya Tumanov [MS]]

I'd say this is typical. There's nothing you can tweak besides using more
powerful PC.
Emulation is very slow in general, full power x86 is slowed down 10-20
times because of it.
Also, emulator is x86, so it's running slower version of CF's IJIT instead
of highly optimized ARM version sJIT.
There's a new ARM emulator coming with VS 2005 which would be way better.

Best regards.

Ilya

This posting is provided "AS IS" with no warranties, and confers no rights.

--------------------

 

[Paul G. Tobey [eMVP]]

Yes, I'd say that the performance is fairly typical. Remote debugging of
managed code is noticeable slow in some cases, even on the real devices. I
don't think that you're doing anything wrong to deserve bad performance.

I'm sure that there will be a few improvements in the future versions, but
the emulator is just an emulator and if I had to choose between work to make
the OS itself and the tools themselves better or making the emulator faster,
I'd choose the former.

Paul T.

 

[Alex Feinman [MVP]]

What you are seeing is not poor emulator performance but rather poor
debugger link performance.

This performance is not typical. Among the things that can drastically
affect the emulator performance under debugger there are such as
1) Having the memory inspector window opened
2) Having too many watched variables

 

[Robert Bouillon]

Thanks.

Yes, I'd say that the performance is fairly typical. Remote debugging of
managed code is noticeable slow in some cases, even on the real devices. I
don't think that you're doing anything wrong to deserve bad performance.

I'm sure that there will be a few improvements in the future versions, but
the emulator is just an emulator and if I had to choose between work to make
the OS itself and the tools themselves better or making the emulator faster,
I'd choose the former.

Paul T.

 

[Robert Bouillon]

Awesome. Just what I needed to know

Thanks.

I'd say this is typical. There's nothing you can tweak besides using more
powerful PC.
Emulation is very slow in general, full power x86 is slowed down 10-20
times because of it.
Also, emulator is x86, so it's running slower version of CF's IJIT instead
of highly optimized ARM version sJIT.
There's a new ARM emulator coming with VS 2005 which would be way better.

Best regards.

Ilya

This posting is provided "AS IS" with no warranties, and confers no rights.

--------------------
From: "Robert Bouillon" <[email protected]>
Subject: Emulator Performance
Date: Tue, 26 Oct 2004 13:53:35 -0400
Lines: 21
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1437
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1441
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 24-161-37-111.hvc.rr.com 24.161.37.111
Path:

cpmsftngxa10.phx.gbl!TK2MSFTFEED02.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP0

8.phx.gbl!TK2MSFTNGP14.phx.gbl
Xref: cpmsftngxa10.phx.gbl microsoft.public.dotnet.framework.compactframework:63942
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

I'm on a 2.4 GHz P4, 1GB DDR3200 2-3-2-2, I875P, SATA RAID0.

My emulator performs most operations at the speed of a 233MHz ARM I have.

Stepping into the program is slow: Loading the stack or local/watch windows
can take up to 30 seconds. Accessing variables can take up to 10 seconds
from within Visual Studio. I get a 1-2 second delay each step, stepping
through the program.

I'm running the latest service packs on 2kServer.

Is this performance typical? Is there any way to tweak this performance? I
would think a full-power x86 proc would perform significantly faster

than

a
low-power ARM. If not, are there any plans to improve the speed of the
Emulator for .NET?

Thanks.

--ROBERT

 

[Robert Bouillon]

I usually only have 1 watch window open with 2 items @ most in it (That I'm
watching).

I don't have the memory inspector window opened.

I do, however, have most windows available to me, but hidden in tabs, such
as Locals, Autos, etc, but I usually only keep the ouput and command windows
focused unless I need the others.

--ROBERT

 

[Alex Feinman [MVP]]

Allow me to disagree. In my experience the emulator running on a 2.54 GHz
Pentium 4 machine outperforms a PPC2002 device with 206GHz Intel CPU (Symbol
2800) 2-3:1. This is already very good. Where emulator is slower is anything
graphic-intensive, but as long as you stick with intrinsic controls and
regular forms, it is better if anything

The new emulator offers comparable performance, so I don't see why would you
want to instill a hope that it would magically fix things

The issue at hands is clearly a transport issue rather than emulator itself.
In myt experience it is typically resilced by reducing the number of open
debug windows, cold-resetting the emulator and making sure the computer IP
configuration is straightforward

--
Alex Feinman
---
Visit http://www.opennetcf.org

I'd say this is typical. There's nothing you can tweak besides using more
powerful PC.
Emulation is very slow in general, full power x86 is slowed down 10-20
times because of it.
Also, emulator is x86, so it's running slower version of CF's IJIT instead
of highly optimized ARM version sJIT.
There's a new ARM emulator coming with VS 2005 which would be way better.

Best regards.

Ilya

This posting is provided "AS IS" with no warranties, and confers no
rights.

--------------------

From: "Robert Bouillon" <[email protected]>
Subject: Emulator Performance
Date: Tue, 26 Oct 2004 13:53:35 -0400
Lines: 21
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1437
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1441
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 24-161-37-111.hvc.rr.com 24.161.37.111
Path: cpmsftngxa10.phx.gbl!TK2MSFTFEED02.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP0
8.phx.gbl!TK2MSFTNGP14.phx.gbl
Xref: cpmsftngxa10.phx.gbl microsoft.public.dotnet.framework.compactframework:63942
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

I'm on a 2.4 GHz P4, 1GB DDR3200 2-3-2-2, I875P, SATA RAID0.

My emulator performs most operations at the speed of a 233MHz ARM I have.

Stepping into the program is slow: Loading the stack or local/watch windows
can take up to 30 seconds. Accessing variables can take up to 10 seconds
from within Visual Studio. I get a 1-2 second delay each step, stepping
through the program.

I'm running the latest service packs on 2kServer.

Is this performance typical? Is there any way to tweak this performance?
I
would think a full-power x86 proc would perform significantly faster than a
low-power ARM. If not, are there any plans to improve the speed of the
Emulator for .NET?

Thanks.

--ROBERT

 

[Ilya Tumanov [MS]]

I'd say it depends on what exactly code is doing.
Say, emulator running on my P4EE 3.2 populates list view control at half
the speed of e740.
At the same time, it's two times faster looping in SqlCeDataReader.
Personally, I do not have issues with emulator performance in code
execution or debugging.
However, I'm pretty sure somebody running it on P3-600 will.
In any case, it's a good idea to run performance tests on real target, not
the emulator.

Best regards,

Ilya

This posting is provided "AS IS" with no warranties, and confers no rights.
--------------------

From: "Alex Feinman [MVP]" <[email protected]>
References: <[email protected]>

Subject: Re: Emulator Performance
Date: Tue, 26 Oct 2004 14:37:29 -0700
Lines: 82
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2900.2180
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180
X-RFC2646: Format=Flowed; Original
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 204.249.181.133
Path: cpmsftngxa10.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP08.phx.gbl!TK2MSFTNGP14
.phx.gbl
Xref: cpmsftngxa10.phx.gbl microsoft.public.dotnet.framework.compactframework:63974
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

Allow me to disagree. In my experience the emulator running on a 2.54 GHz
Pentium 4 machine outperforms a PPC2002 device with 206GHz Intel CPU (Symbol
2800) 2-3:1. This is already very good. Where emulator is slower is anything
graphic-intensive, but as long as you stick with intrinsic controls and
regular forms, it is better if anything

The new emulator offers comparable performance, so I don't see why would you
want to instill a hope that it would magically fix things

The issue at hands is clearly a transport issue rather than emulator itself.
In myt experience it is typically resilced by reducing the number of open
debug windows, cold-resetting the emulator and making sure the computer IP
configuration is straightforward

--
Alex Feinman
---
Visit http://www.opennetcf.org

I'd say this is typical. There's nothing you can tweak besides using more
powerful PC.
Emulation is very slow in general, full power x86 is slowed down 10-20
times because of it.
Also, emulator is x86, so it's running slower version of CF's IJIT instead
of highly optimized ARM version sJIT.
There's a new ARM emulator coming with VS 2005 which would be way better.

Best regards.

Ilya

This posting is provided "AS IS" with no warranties, and confers no
rights.

--------------------

From: "Robert Bouillon" <[email protected]>
Subject: Emulator Performance
Date: Tue, 26 Oct 2004 13:53:35 -0400
Lines: 21
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1437
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1441
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 24-161-37-111.hvc.rr.com 24.161.37.111
Path:

cpmsftngxa10.phx.gbl!TK2MSFTFEED02.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP0
8.phx.gbl!TK2MSFTNGP14.phx.gbl

Xref: cpmsftngxa10.phx.gbl microsoft.public.dotnet.framework.compactframework:63942
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

I'm on a 2.4 GHz P4, 1GB DDR3200 2-3-2-2, I875P, SATA RAID0.

My emulator performs most operations at the speed of a 233MHz ARM I have.

Stepping into the program is slow: Loading the stack or local/watch windows
can take up to 30 seconds. Accessing variables can take up to 10 seconds
from within Visual Studio. I get a 1-2 second delay each step, stepping
through the program.

I'm running the latest service packs on 2kServer.

Is this performance typical? Is there any way to tweak this performance?
I
would think a full-power x86 proc would perform significantly faster

than
a

low-power ARM. If not, are there any plans to improve the speed of the
Emulator for .NET?

Thanks.

--ROBERT

 

[Robert Bouillon]

When you say "New Emulator", are you referring to the PPC2003 SE images, or
the .NET emulator vs. eVC?

--ROBERT

Allow me to disagree. In my experience the emulator running on a 2.54 GHz
Pentium 4 machine outperforms a PPC2002 device with 206GHz Intel CPU (Symbol
2800) 2-3:1. This is already very good. Where emulator is slower is anything
graphic-intensive, but as long as you stick with intrinsic controls and
regular forms, it is better if anything

The new emulator offers comparable performance, so I don't see why would you
want to instill a hope that it would magically fix things

The issue at hands is clearly a transport issue rather than emulator itself.
In myt experience it is typically resilced by reducing the number of open
debug windows, cold-resetting the emulator and making sure the computer IP
configuration is straightforward

--
Alex Feinman
---
Visit http://www.opennetcf.org

I'd say this is typical. There's nothing you can tweak besides using more
powerful PC.
Emulation is very slow in general, full power x86 is slowed down 10-20
times because of it.
Also, emulator is x86, so it's running slower version of CF's IJIT instead
of highly optimized ARM version sJIT.
There's a new ARM emulator coming with VS 2005 which would be way better.

Best regards.

Ilya

This posting is provided "AS IS" with no warranties, and confers no
rights.

--------------------

From: "Robert Bouillon" <[email protected]>
Subject: Emulator Performance
Date: Tue, 26 Oct 2004 13:53:35 -0400
Lines: 21
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1437
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1441
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 24-161-37-111.hvc.rr.com 24.161.37.111
Path:

cpmsftngxa10.phx.gbl!TK2MSFTFEED02.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP0
8.phx.gbl!TK2MSFTNGP14.phx.gbl

Xref: cpmsftngxa10.phx.gbl microsoft.public.dotnet.framework.compactframework:63942
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

I'm on a 2.4 GHz P4, 1GB DDR3200 2-3-2-2, I875P, SATA RAID0.

My emulator performs most operations at the speed of a 233MHz ARM I have.

Stepping into the program is slow: Loading the stack or local/watch windows
can take up to 30 seconds. Accessing variables can take up to 10 seconds
from within Visual Studio. I get a 1-2 second delay each step, stepping
through the program.

I'm running the latest service packs on 2kServer.

Is this performance typical? Is there any way to tweak this performance?
I
would think a full-power x86 proc would perform significantly faster

than
a

low-power ARM. If not, are there any plans to improve the speed of the
Emulator for .NET?

Thanks.

--ROBERT

 

[Alex Feinman [MVP]]

Well, the OP indicated that he is using P4 2.4 (which is fast enough).
You are absolutely correct about doing performance tests on a real hardware
rather than on the emulator, but this is beyond the point. The OP complained
about the debugger low performance to the point of being unusable. That has
nothiung to do with performance testing and clearly not a normal behavior

--
Alex Feinman
---
Visit http://www.opennetcf.org

I'd say it depends on what exactly code is doing.
Say, emulator running on my P4EE 3.2 populates list view control at half
the speed of e740.
At the same time, it's two times faster looping in SqlCeDataReader.
Personally, I do not have issues with emulator performance in code
execution or debugging.
However, I'm pretty sure somebody running it on P3-600 will.
In any case, it's a good idea to run performance tests on real target, not
the emulator.

Best regards,

Ilya

This posting is provided "AS IS" with no warranties, and confers no
rights.
--------------------

From: "Alex Feinman [MVP]" <[email protected]>
References: <[email protected]>

Subject: Re: Emulator Performance
Date: Tue, 26 Oct 2004 14:37:29 -0700
Lines: 82
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2900.2180
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180
X-RFC2646: Format=Flowed; Original
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 204.249.181.133
Path: cpmsftngxa10.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP08.phx.gbl!TK2MSFTNGP14
phx.gbl
Xref: cpmsftngxa10.phx.gbl microsoft.public.dotnet.framework.compactframework:63974
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

Allow me to disagree. In my experience the emulator running on a 2.54 GHz
Pentium 4 machine outperforms a PPC2002 device with 206GHz Intel CPU (Symbol
2800) 2-3:1. This is already very good. Where emulator is slower is anything
graphic-intensive, but as long as you stick with intrinsic controls and
regular forms, it is better if anything

The new emulator offers comparable performance, so I don't see why would you
want to instill a hope that it would magically fix things

The issue at hands is clearly a transport issue rather than emulator itself.
In myt experience it is typically resilced by reducing the number of open
debug windows, cold-resetting the emulator and making sure the computer IP
configuration is straightforward

--
Alex Feinman
---
Visit http://www.opennetcf.org

I'd say this is typical. There's nothing you can tweak besides using more
powerful PC.
Emulation is very slow in general, full power x86 is slowed down 10-20
times because of it.
Also, emulator is x86, so it's running slower version of CF's IJIT instead
of highly optimized ARM version sJIT.
There's a new ARM emulator coming with VS 2005 which would be way better.

Best regards.

Ilya

This posting is provided "AS IS" with no warranties, and confers no
rights.

--------------------
From: "Robert Bouillon" <[email protected]>
Subject: Emulator Performance
Date: Tue, 26 Oct 2004 13:53:35 -0400
Lines: 21
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1437
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1441
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 24-161-37-111.hvc.rr.com 24.161.37.111
Path:
cpmsftngxa10.phx.gbl!TK2MSFTFEED02.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP0
8.phx.gbl!TK2MSFTNGP14.phx.gbl
Xref: cpmsftngxa10.phx.gbl
microsoft.public.dotnet.framework.compactframework:63942
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

I'm on a 2.4 GHz P4, 1GB DDR3200 2-3-2-2, I875P, SATA RAID0.

My emulator performs most operations at the speed of a 233MHz ARM I have.

Stepping into the program is slow: Loading the stack or local/watch
windows
can take up to 30 seconds. Accessing variables can take up to 10 seconds
from within Visual Studio. I get a 1-2 second delay each step,
stepping
through the program.

I'm running the latest service packs on 2kServer.

Is this performance typical? Is there any way to tweak this performance?
I
would think a full-power x86 proc would perform significantly faster than
a
low-power ARM. If not, are there any plans to improve the speed of the
Emulator for .NET?

Thanks.

--ROBERT

 

[Robert Bouillon]

OT: How is the P4EE versus the standard P4? Benchmarks really don't do
justice to real-world apps, especially with regards to development. I hear
she runs cooler than the comperable P4 775 with the new stepping, too.

--ROBERT

I'd say it depends on what exactly code is doing.
Say, emulator running on my P4EE 3.2 populates list view control at half
the speed of e740.
At the same time, it's two times faster looping in SqlCeDataReader.
Personally, I do not have issues with emulator performance in code
execution or debugging.
However, I'm pretty sure somebody running it on P3-600 will.
In any case, it's a good idea to run performance tests on real target, not
the emulator.

Best regards,

Ilya

This posting is provided "AS IS" with no warranties, and confers no rights.

Subject: Re: Emulator Performance
Date: Tue, 26 Oct 2004 14:37:29 -0700
Lines: 82
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2900.2180
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180
X-RFC2646: Format=Flowed; Original
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 204.249.181.133
Path:

cpmsftngxa10.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP08.phx.gbl!TK2MSFTNGP14

phx.gbl
Xref: cpmsftngxa10.phx.gbl microsoft.public.dotnet.framework.compactframework:63974
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

Allow me to disagree. In my experience the emulator running on a 2.54 GHz
Pentium 4 machine outperforms a PPC2002 device with 206GHz Intel CPU (Symbol
2800) 2-3:1. This is already very good. Where emulator is slower is anything
graphic-intensive, but as long as you stick with intrinsic controls and
regular forms, it is better if anything

The new emulator offers comparable performance, so I don't see why would you
want to instill a hope that it would magically fix things

The issue at hands is clearly a transport issue rather than emulator itself.
In myt experience it is typically resilced by reducing the number of open
debug windows, cold-resetting the emulator and making sure the computer IP
configuration is straightforward

--
Alex Feinman
---
Visit http://www.opennetcf.org

I'd say this is typical. There's nothing you can tweak besides using more
powerful PC.
Emulation is very slow in general, full power x86 is slowed down 10-20
times because of it.
Also, emulator is x86, so it's running slower version of CF's IJIT instead
of highly optimized ARM version sJIT.
There's a new ARM emulator coming with VS 2005 which would be way better.

Best regards.

Ilya

This posting is provided "AS IS" with no warranties, and confers no
rights.

--------------------
From: "Robert Bouillon" <[email protected]>
Subject: Emulator Performance
Date: Tue, 26 Oct 2004 13:53:35 -0400
Lines: 21
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1437
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1441
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 24-161-37-111.hvc.rr.com 24.161.37.111
Path:

cpmsftngxa10.phx.gbl!TK2MSFTFEED02.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP0

 

[Robert Bouillon]

So do we have any recommendations? Little SQL errors are killing me because
my app initialization is heavy due to form caching and DB preparations.
Between app startup and debugging, I'm losing a lot of time.

As a note, I'm using referenced DLLs in the app, and my app uses about 8 MB
of RAM after it's loaded. Maybe that has something to do with it (Reaching
out to a referenced dll is slowing the link down?). Three referenced dlls to
be exact (A Business layer, Form Controls for the Business Layer, and a
Toolset dll that extends the mscorlib).

--ROBERT

Well, the OP indicated that he is using P4 2.4 (which is fast enough).
You are absolutely correct about doing performance tests on a real hardware
rather than on the emulator, but this is beyond the point. The OP complained
about the debugger low performance to the point of being unusable. That has
nothiung to do with performance testing and clearly not a normal behavior

--
Alex Feinman
---
Visit http://www.opennetcf.org

I'd say it depends on what exactly code is doing.
Say, emulator running on my P4EE 3.2 populates list view control at half
the speed of e740.
At the same time, it's two times faster looping in SqlCeDataReader.
Personally, I do not have issues with emulator performance in code
execution or debugging.
However, I'm pretty sure somebody running it on P3-600 will.
In any case, it's a good idea to run performance tests on real target, not
the emulator.

Best regards,

Ilya

This posting is provided "AS IS" with no warranties, and confers no
rights.
--------------------

From: "Alex Feinman [MVP]" <[email protected]>
References: <[email protected]>

Subject: Re: Emulator Performance
Date: Tue, 26 Oct 2004 14:37:29 -0700
Lines: 82
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2900.2180
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180
X-RFC2646: Format=Flowed; Original
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 204.249.181.133
Path:

cpmsftngxa10.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP08.phx.gbl!TK2MSFTNGP14
phx.gbl

Xref: cpmsftngxa10.phx.gbl microsoft.public.dotnet.framework.compactframework:63974
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

Allow me to disagree. In my experience the emulator running on a 2.54 GHz
Pentium 4 machine outperforms a PPC2002 device with 206GHz Intel CPU (Symbol
2800) 2-3:1. This is already very good. Where emulator is slower is anything
graphic-intensive, but as long as you stick with intrinsic controls and
regular forms, it is better if anything

The new emulator offers comparable performance, so I don't see why

would
you

want to instill a hope that it would magically fix things

The issue at hands is clearly a transport issue rather than emulator itself.
In myt experience it is typically resilced by reducing the number of open
debug windows, cold-resetting the emulator and making sure the computer IP
configuration is straightforward

--
Alex Feinman
---
Visit http://www.opennetcf.org
I'd say this is typical. There's nothing you can tweak besides using more
powerful PC.
Emulation is very slow in general, full power x86 is slowed down 10-20
times because of it.
Also, emulator is x86, so it's running slower version of CF's IJIT instead
of highly optimized ARM version sJIT.
There's a new ARM emulator coming with VS 2005 which would be way better.

Best regards.

Ilya

This posting is provided "AS IS" with no warranties, and confers no
rights.

--------------------
From: "Robert Bouillon" <[email protected]>
Subject: Emulator Performance
Date: Tue, 26 Oct 2004 13:53:35 -0400
Lines: 21
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1437
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1441
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 24-161-37-111.hvc.rr.com 24.161.37.111
Path:

cpmsftngxa10.phx.gbl!TK2MSFTFEED02.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP0

8.phx.gbl!TK2MSFTNGP14.phx.gbl
Xref: cpmsftngxa10.phx.gbl
microsoft.public.dotnet.framework.compactframework:63942
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

I'm on a 2.4 GHz P4, 1GB DDR3200 2-3-2-2, I875P, SATA RAID0.

My emulator performs most operations at the speed of a 233MHz ARM I have.

Stepping into the program is slow: Loading the stack or local/watch
windows
can take up to 30 seconds. Accessing variables can take up to 10 seconds
from within Visual Studio. I get a 1-2 second delay each step,
stepping
through the program.

I'm running the latest service packs on 2kServer.

Is this performance typical? Is there any way to tweak this performance?
I
would think a full-power x86 proc would perform significantly faster than
a
low-power ARM. If not, are there any plans to improve the speed of the
Emulator for .NET?

Thanks.

--ROBERT

 

[Paul G. Tobey [eMVP]]

The device is probably much faster. Use a network connection, not
ActiveSync, if possible...

Paul T.

So do we have any recommendations? Little SQL errors are killing me
because
my app initialization is heavy due to form caching and DB preparations.
Between app startup and debugging, I'm losing a lot of time.

As a note, I'm using referenced DLLs in the app, and my app uses about 8
MB
of RAM after it's loaded. Maybe that has something to do with it (Reaching
out to a referenced dll is slowing the link down?). Three referenced dlls
to
be exact (A Business layer, Form Controls for the Business Layer, and a
Toolset dll that extends the mscorlib).

--ROBERT

Well, the OP indicated that he is using P4 2.4 (which is fast enough).
You are absolutely correct about doing performance tests on a real hardware
rather than on the emulator, but this is beyond the point. The OP complained
about the debugger low performance to the point of being unusable. That has
nothiung to do with performance testing and clearly not a normal behavior

--
Alex Feinman
---
Visit http://www.opennetcf.org

I'd say it depends on what exactly code is doing.
Say, emulator running on my P4EE 3.2 populates list view control at
half
the speed of e740.
At the same time, it's two times faster looping in SqlCeDataReader.
Personally, I do not have issues with emulator performance in code
execution or debugging.
However, I'm pretty sure somebody running it on P3-600 will.
In any case, it's a good idea to run performance tests on real target, not
the emulator.

Best regards,

Ilya

This posting is provided "AS IS" with no warranties, and confers no
rights.
--------------------
From: "Alex Feinman [MVP]" <[email protected]>
References: <[email protected]>
<[email protected]>
Subject: Re: Emulator Performance
Date: Tue, 26 Oct 2004 14:37:29 -0700
Lines: 82
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2900.2180
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180
X-RFC2646: Format=Flowed; Original
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 204.249.181.133
Path:
cpmsftngxa10.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP08.phx.gbl!TK2MSFTNGP14
phx.gbl
Xref: cpmsftngxa10.phx.gbl
microsoft.public.dotnet.framework.compactframework:63974
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

Allow me to disagree. In my experience the emulator running on a 2.54 GHz
Pentium 4 machine outperforms a PPC2002 device with 206GHz Intel CPU
(Symbol
2800) 2-3:1. This is already very good. Where emulator is slower is
anything
graphic-intensive, but as long as you stick with intrinsic controls
and
regular forms, it is better if anything

The new emulator offers comparable performance, so I don't see why would
you
want to instill a hope that it would magically fix things

The issue at hands is clearly a transport issue rather than emulator
itself.
In myt experience it is typically resilced by reducing the number of open
debug windows, cold-resetting the emulator and making sure the
computer
IP
configuration is straightforward

--
Alex Feinman
---
Visit http://www.opennetcf.org
I'd say this is typical. There's nothing you can tweak besides using
more
powerful PC.
Emulation is very slow in general, full power x86 is slowed down 10-20
times because of it.
Also, emulator is x86, so it's running slower version of CF's IJIT
instead
of highly optimized ARM version sJIT.
There's a new ARM emulator coming with VS 2005 which would be way
better.

Best regards.

Ilya

This posting is provided "AS IS" with no warranties, and confers no
rights.

--------------------
From: "Robert Bouillon" <[email protected]>
Subject: Emulator Performance
Date: Tue, 26 Oct 2004 13:53:35 -0400
Lines: 21
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1437
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1441
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 24-161-37-111.hvc.rr.com 24.161.37.111
Path:

cpmsftngxa10.phx.gbl!TK2MSFTFEED02.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP0
8.phx.gbl!TK2MSFTNGP14.phx.gbl
Xref: cpmsftngxa10.phx.gbl
microsoft.public.dotnet.framework.compactframework:63942
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

I'm on a 2.4 GHz P4, 1GB DDR3200 2-3-2-2, I875P, SATA RAID0.

My emulator performs most operations at the speed of a 233MHz ARM I
have.

Stepping into the program is slow: Loading the stack or local/watch
windows
can take up to 30 seconds. Accessing variables can take up to 10
seconds
from within Visual Studio. I get a 1-2 second delay each step,
stepping
through the program.

I'm running the latest service packs on 2kServer.

Is this performance typical? Is there any way to tweak this
performance?
I
would think a full-power x86 proc would perform significantly
faster
than
a
low-power ARM. If not, are there any plans to improve the speed of the
Emulator for .NET?

Thanks.

--ROBERT

 

[Alex Feinman [MVP]]

Is there any chance you could try your debugging on a fresh machine? I'm not
suggesting reformatting and starting from scratch, but rather "borrowing"
some other system for half hour

--
Alex Feinman
---
Visit http://www.opennetcf.org

So do we have any recommendations? Little SQL errors are killing me
because
my app initialization is heavy due to form caching and DB preparations.
Between app startup and debugging, I'm losing a lot of time.

As a note, I'm using referenced DLLs in the app, and my app uses about 8
MB
of RAM after it's loaded. Maybe that has something to do with it (Reaching
out to a referenced dll is slowing the link down?). Three referenced dlls
to
be exact (A Business layer, Form Controls for the Business Layer, and a
Toolset dll that extends the mscorlib).

--ROBERT

Well, the OP indicated that he is using P4 2.4 (which is fast enough).
You are absolutely correct about doing performance tests on a real hardware
rather than on the emulator, but this is beyond the point. The OP complained
about the debugger low performance to the point of being unusable. That has
nothiung to do with performance testing and clearly not a normal behavior

--
Alex Feinman
---
Visit http://www.opennetcf.org

I'd say it depends on what exactly code is doing.
Say, emulator running on my P4EE 3.2 populates list view control at
half
the speed of e740.
At the same time, it's two times faster looping in SqlCeDataReader.
Personally, I do not have issues with emulator performance in code
execution or debugging.
However, I'm pretty sure somebody running it on P3-600 will.
In any case, it's a good idea to run performance tests on real target, not
the emulator.

Best regards,

Ilya

This posting is provided "AS IS" with no warranties, and confers no
rights.
--------------------
From: "Alex Feinman [MVP]" <[email protected]>
References: <[email protected]>
<[email protected]>
Subject: Re: Emulator Performance
Date: Tue, 26 Oct 2004 14:37:29 -0700
Lines: 82
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2900.2180
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180
X-RFC2646: Format=Flowed; Original
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 204.249.181.133
Path:
cpmsftngxa10.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP08.phx.gbl!TK2MSFTNGP14
phx.gbl
Xref: cpmsftngxa10.phx.gbl
microsoft.public.dotnet.framework.compactframework:63974
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

Allow me to disagree. In my experience the emulator running on a 2.54 GHz
Pentium 4 machine outperforms a PPC2002 device with 206GHz Intel CPU
(Symbol
2800) 2-3:1. This is already very good. Where emulator is slower is
anything
graphic-intensive, but as long as you stick with intrinsic controls
and
regular forms, it is better if anything

The new emulator offers comparable performance, so I don't see why would
you
want to instill a hope that it would magically fix things

The issue at hands is clearly a transport issue rather than emulator
itself.
In myt experience it is typically resilced by reducing the number of open
debug windows, cold-resetting the emulator and making sure the
computer
IP
configuration is straightforward

--
Alex Feinman
---
Visit http://www.opennetcf.org
I'd say this is typical. There's nothing you can tweak besides using
more
powerful PC.
Emulation is very slow in general, full power x86 is slowed down 10-20
times because of it.
Also, emulator is x86, so it's running slower version of CF's IJIT
instead
of highly optimized ARM version sJIT.
There's a new ARM emulator coming with VS 2005 which would be way
better.

Best regards.

Ilya

This posting is provided "AS IS" with no warranties, and confers no
rights.

--------------------
From: "Robert Bouillon" <[email protected]>
Subject: Emulator Performance
Date: Tue, 26 Oct 2004 13:53:35 -0400
Lines: 21
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1437
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1441
Message-ID: <[email protected]>
Newsgroups: microsoft.public.dotnet.framework.compactframework
NNTP-Posting-Host: 24-161-37-111.hvc.rr.com 24.161.37.111
Path:

cpmsftngxa10.phx.gbl!TK2MSFTFEED02.phx.gbl!TK2MSFTNGXA01.phx.gbl!TK2MSFTNGP0
8.phx.gbl!TK2MSFTNGP14.phx.gbl
Xref: cpmsftngxa10.phx.gbl
microsoft.public.dotnet.framework.compactframework:63942
X-Tomcat-NG: microsoft.public.dotnet.framework.compactframework

I'm on a 2.4 GHz P4, 1GB DDR3200 2-3-2-2, I875P, SATA RAID0.

My emulator performs most operations at the speed of a 233MHz ARM I
have.

Stepping into the program is slow: Loading the stack or local/watch
windows
can take up to 30 seconds. Accessing variables can take up to 10
seconds
from within Visual Studio. I get a 1-2 second delay each step,
stepping
through the program.

I'm running the latest service packs on 2kServer.

Is this performance typical? Is there any way to tweak this
performance?
I
would think a full-power x86 proc would perform significantly
faster
than
a
low-power ARM. If not, are there any plans to improve the speed of the
Emulator for .NET?

Thanks.

--ROBERT