NET 2.0 speed compared to NET 1.1

[Mrinal]

Hi ,

This is my take on the same :

1. .Net 2.0 , as Nicolas suggested will be a definite improvement in certain
areas , but that will be more related to optimization , but how much human
perception will catch it , totally depends on application .

2. There are cases , where actually un managed code can perform better ,
especially when .net code has to go for some kind of P Invoking with win32
API's, then
there are sure shot overheads , but then it is offset by good things like :

1. Automatic memory management , as unmanaged code are main reason for mem.
and handle leaks .

There could be some performance loss at IL -- Native Code level , but good
part is easy management of IL stuff and various other security benefits that
couple along .

3. Ngen , by defnition it seems to make good differences , but then there
are certain caveats :

1. By defaults it generates the code that is LCD for various compatible
processor architectures or otherwise for optimizing we need to know about
special calls and use appropriate flags , however jit , cretaes a real
optimized code with all benefits .

2. It invariably needs IL code (Manifest ) , at runtime , you maynot think
of protecting intellectual properties .

3. Any issues with installed native cache images , execution reverts to JIT
..

Actually JIT in theory may look like having some issues , but it's benefits
offsets all of them and again it's a first time hit after that there's no
repetitive compilation till the point source code is changed .

i hope this helps upto some extent ,

Mrinal

 

[Joanna Carter \(TeamB\)]

Hi gurus.

Does anyone have any experience as to whether .NET 2.0 will be any faster
executing than .NET 1.1 ?

We are looking at starting a new project in .NET and are getting feedback
from other developers that they have found problems with apparent speed of
apps being slower than their clients were expecting having come from Win32
Delphi apps.

Does NGen'ing an app help with the speed ?

Joanna

 

[Nicholas Paldino [.NET/C# MVP]]

Joanna,

It all depends on the application you are writing. All things
considered, I don't see any reason why a .NET program can't perform as well
as a non-managed program.

.NET 2.0 will be faster in some areas, but not across the board. These
kind of improvements are not really going to be realized in such a general
manner.

There are also people out there that are writing code for the managed
environments that are doing it wrong, which can contribute to this.

The problem here is the use of the word "apparent". Unless you have
numbers indicating that there is a massive differential in performance, I
wouldn't bother trying to guess. Get some hard numbers, making sure that
best practices are being observed, and then make a determination based on
that.

Hope this helps.

 

[clintonG]

Having never used it before I recently suggested the consideration of ngen
to somebody asking the same type of questions but one of the fellas who also
replied to the topic stated ngen has 'issues.'

Study the benefits of using aspnet_compiler which is the term you can submit
to search.

<%= Clinton Gallagher
METROmilwaukee (sm) "A Regional Information Service"
NET csgallagher AT metromilwaukee.com
URL http://metromilwaukee.com/
URL http://clintongallagher.metromilwaukee.com/

Joanna,

It all depends on the application you are writing. All things
considered, I don't see any reason why a .NET program can't perform as
well as a non-managed program.

.NET 2.0 will be faster in some areas, but not across the board. These
kind of improvements are not really going to be realized in such a general
manner.

There are also people out there that are writing code for the managed
environments that are doing it wrong, which can contribute to this.

The problem here is the use of the word "apparent". Unless you have
numbers indicating that there is a massive differential in performance, I
wouldn't bother trying to guess. Get some hard numbers, making sure that
best practices are being observed, and then make a determination based on
that.

Hope this helps.


--
- Nicholas Paldino [.NET/C# MVP]
- (e-mail address removed)

Hi gurus.

Does anyone have any experience as to whether .NET 2.0 will be any faster
executing than .NET 1.1 ?

We are looking at starting a new project in .NET and are getting feedback
from other developers that they have found problems with apparent speed
of
apps being slower than their clients were expecting having come from
Win32
Delphi apps.

Does NGen'ing an app help with the speed ?

Joanna

 

[Joanna Carter \(TeamB\)]

It all depends on the application you are writing. All things
considered, I don't see any reason why a .NET program can't perform as well
as a non-managed program.

.NET 2.0 will be faster in some areas, but not across the board. These
kind of improvements are not really going to be realized in such a general
manner.

What areas tend to induce speed problems ?

We will be using our own data persistence layer and connecting UI controls
to objects and lists rather than directly to data tables.

The data layer will dynamically create SQL statements for submission to a
single query component, so we will not be using datasets of multiple tables
which, I gather, can be a potential bottleneck.

Once the code has been run once, I gather that the JIT "penalty" is removed
from the equation and that I should expect native or better ^performance ?

There are also people out there that are writing code for the managed
environments that are doing it wrong, which can contribute to this.

Don't worry, as a Delphi consultant, I get to see a fair amount of bad code
that can make Delphi apps run like Java <gd&r>

Joanna