Smalltalk benchmarks for Composite systems?

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

Smalltalk benchmarks for Composite systems?

Christoph Thiede

Hi all,


Two years ago, I started a project for a Parser library - unfortunately, at this time, I was not aware of the marvels of Squeak, so I wrote it in C#. Now I've resumed that project and feel the classical, un-live development process more cumbersome than ever. I would love to migrate my project to Smalltalk, but I'm worrying how much slower it will be compared to C#, as I am doing some quite CPU-intensive computations (that take even C# several seconds for small instances).


So I'm searching for some meaningful benchmarks to get an idea of the speed losses to expect. The main computations affect composite traversal, deep copying, etc., so I think the workload should mainly address message-sending. I did not found really good benchmarks searching for "(squeak | smalltalk) vs (c# | java) benchmark" (assuming that C# and Java play in about the same league). Can someone give me a few pointers? In short, under what circumstances would you recommend me to port a computation-intensive library to Smalltalk?

Thanks in advance!


Best regards,

Christoph


_______________________________________________
Beginners mailing list
[hidden email]
http://lists.squeakfoundation.org/mailman/listinfo/beginners
Reply | Threaded
Open this post in threaded view
|

Re: Smalltalk benchmarks for Composite systems?

Hannes Hirzel
Hi Christoph

As you did not find a good benchmark I suggest that you design a suite
of benchmarks for your own purpose. Use as target the  field of
application.

http://rosettacode.org/wiki/Rosetta_Code

Contains many algorithms for different languages.

http://rosettacode.org/wiki/Tree_traversal   (unfortunaltely Smalltalk
version not implemented yet)

But there are over 200 tasks done in Smalltalk

http://rosettacode.org/wiki/Category:Smalltalk

Kind regards
Hannes


P.S. Also note that the SistaV1 VM will offer a substantial
improvement in performance https://wiki.squeak.org/squeak/676

On 9/25/19, Thiede, Christoph
<[hidden email]> wrote:

> Hi all,
>
>
> Two years ago, I started a project for a Parser library - unfortunately, at
> this time, I was not aware of the marvels of Squeak, so I wrote it in C#.
> Now I've resumed that project and feel the classical, un-live development
> process more cumbersome than ever. I would love to migrate my project to
> Smalltalk, but I'm worrying how much slower it will be compared to C#, as I
> am doing some quite CPU-intensive computations (that take even C# several
> seconds for small instances).
>
>
> So I'm searching for some meaningful benchmarks to get an idea of the speed
> losses to expect. The main computations affect composite traversal, deep
> copying, etc., so I think the workload should mainly address
> message-sending. I did not found really good benchmarks searching for
> "(squeak | smalltalk) vs (c# | java) benchmark" (assuming that C# and Java
> play in about the same league). Can someone give me a few pointers? In
> short, under what circumstances would you recommend me to port a
> computation-intensive library to Smalltalk?
>
> Thanks in advance!
>
>
> Best regards,
>
> Christoph
>
_______________________________________________
Beginners mailing list
[hidden email]
http://lists.squeakfoundation.org/mailman/listinfo/beginners
Reply | Threaded
Open this post in threaded view
|

Re: Smalltalk benchmarks for Composite systems?

Christoph Thiede

Hi Hannes,


thanks for your tips! Rosetta Code is a good idea, I will have a look at this.


How many tests would you say a benchmark should contain for being meaningful? Could you also recommend any other task from this page that suits for message-send-stressing?


I have no experience at all with VMs, yet. Can I already run Squeak on SistaV1 anywhere? Does it affect the overall speed greatly if I am in CommandLine mode only?


Best,

Christoph


Von: Beginners <[hidden email]> im Auftrag von H. Hirzel <[hidden email]>
Gesendet: Donnerstag, 26. September 2019 10:32:17
An: A friendly place to get answers to even the most basic questions about Squeak.
Betreff: Re: [Newbies] Smalltalk benchmarks for Composite systems?
 
Hi Christoph

As you did not find a good benchmark I suggest that you design a suite
of benchmarks for your own purpose. Use as target the  field of
application.

http://rosettacode.org/wiki/Rosetta_Code

Contains many algorithms for different languages.

http://rosettacode.org/wiki/Tree_traversal   (unfortunaltely Smalltalk
version not implemented yet)

But there are over 200 tasks done in Smalltalk

http://rosettacode.org/wiki/Category:Smalltalk

Kind regards
Hannes


P.S. Also note that the SistaV1 VM will offer a substantial
improvement in performance https://wiki.squeak.org/squeak/676

On 9/25/19, Thiede, Christoph
<[hidden email]> wrote:
> Hi all,
>
>
> Two years ago, I started a project for a Parser library - unfortunately, at
> this time, I was not aware of the marvels of Squeak, so I wrote it in C#.
> Now I've resumed that project and feel the classical, un-live development
> process more cumbersome than ever. I would love to migrate my project to
> Smalltalk, but I'm worrying how much slower it will be compared to C#, as I
> am doing some quite CPU-intensive computations (that take even C# several
> seconds for small instances).
>
>
> So I'm searching for some meaningful benchmarks to get an idea of the speed
> losses to expect. The main computations affect composite traversal, deep
> copying, etc., so I think the workload should mainly address
> message-sending. I did not found really good benchmarks searching for
> "(squeak | smalltalk) vs (c# | java) benchmark" (assuming that C# and Java
> play in about the same league). Can someone give me a few pointers? In
> short, under what circumstances would you recommend me to port a
> computation-intensive library to Smalltalk?
>
> Thanks in advance!
>
>
> Best regards,
>
> Christoph
>
_______________________________________________
Beginners mailing list
[hidden email]
http://lists.squeakfoundation.org/mailman/listinfo/beginners

_______________________________________________
Beginners mailing list
[hidden email]
http://lists.squeakfoundation.org/mailman/listinfo/beginners
Reply | Threaded
Open this post in threaded view
|

Re: Smalltalk benchmarks for Composite systems?

Nicola Mingotti

About one month ago I tested a simple loop speed comparing:
C, Squeak, Python, Ruby.

C of course is the kind, but that is not really a fair comparison.

Of the others 3, which are comparable languages, Squeak was the fastest.

bye
Nicola




On 9/26/19 4:29 AM, Thiede, Christoph wrote:

Hi Hannes,


thanks for your tips! Rosetta Code is a good idea, I will have a look at this.


How many tests would you say a benchmark should contain for being meaningful? Could you also recommend any other task from this page that suits for message-send-stressing?


I have no experience at all with VMs, yet. Can I already run Squeak on SistaV1 anywhere? Does it affect the overall speed greatly if I am in CommandLine mode only?


Best,

Christoph


Von: Beginners [hidden email] im Auftrag von H. Hirzel [hidden email]
Gesendet: Donnerstag, 26. September 2019 10:32:17
An: A friendly place to get answers to even the most basic questions about Squeak.
Betreff: Re: [Newbies] Smalltalk benchmarks for Composite systems?
 
Hi Christoph

As you did not find a good benchmark I suggest that you design a suite
of benchmarks for your own purpose. Use as target the  field of
application.

http://rosettacode.org/wiki/Rosetta_Code

Contains many algorithms for different languages.

http://rosettacode.org/wiki/Tree_traversal   (unfortunaltely Smalltalk
version not implemented yet)

But there are over 200 tasks done in Smalltalk

http://rosettacode.org/wiki/Category:Smalltalk

Kind regards
Hannes


P.S. Also note that the SistaV1 VM will offer a substantial
improvement in performance https://wiki.squeak.org/squeak/676

On 9/25/19, Thiede, Christoph
[hidden email] wrote:
> Hi all,
>
>
> Two years ago, I started a project for a Parser library - unfortunately, at
> this time, I was not aware of the marvels of Squeak, so I wrote it in C#.
> Now I've resumed that project and feel the classical, un-live development
> process more cumbersome than ever. I would love to migrate my project to
> Smalltalk, but I'm worrying how much slower it will be compared to C#, as I
> am doing some quite CPU-intensive computations (that take even C# several
> seconds for small instances).
>
>
> So I'm searching for some meaningful benchmarks to get an idea of the speed
> losses to expect. The main computations affect composite traversal, deep
> copying, etc., so I think the workload should mainly address
> message-sending. I did not found really good benchmarks searching for
> "(squeak | smalltalk) vs (c# | java) benchmark" (assuming that C# and Java
> play in about the same league). Can someone give me a few pointers? In
> short, under what circumstances would you recommend me to port a
> computation-intensive library to Smalltalk?
>
> Thanks in advance!
>
>
> Best regards,
>
> Christoph
>
_______________________________________________
Beginners mailing list
[hidden email]
http://lists.squeakfoundation.org/mailman/listinfo/beginners

_______________________________________________
Beginners mailing list
[hidden email]
http://lists.squeakfoundation.org/mailman/listinfo/beginners


_______________________________________________
Beginners mailing list
[hidden email]
http://lists.squeakfoundation.org/mailman/listinfo/beginners