I always thought that 1 to: 100 do: was faster than (1 to: 100) do:

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

I always thought that 1 to: 100 do: was faster than (1 to: 100) do:

Stéphane Ducasse
>
>
> but apparently
> http://code.google.com/p/pharo/issues/detail?id=750
> shows the contrary and this is really strange. I think that may be  
> this is a slow down in Transcript but...
>
> Stef
>
>
>
>


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: I always thought that 1 to: 100 do: was faster than (1 to: 100) do:

Nicolas Cellier
I bet it's undo history again!

reverse the order of the two tests and redo...
or deselect the preference for multipletextUndo.

2009/5/3 Stéphane Ducasse <[hidden email]>:

>>
>>
>> but apparently
>>       http://code.google.com/p/pharo/issues/detail?id=750
>> shows the contrary and this is really strange. I think that may be
>> this is a slow down in Transcript but...
>>
>> Stef
>>
>>
>>
>>
>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: I always thought that 1 to: 100 do: was faster than (1 to: 100) do:

Igor Stasenko
In reply to this post by Stéphane Ducasse
2009/5/3 Stéphane Ducasse <[hidden email]>:
>>
>>
>> but apparently
>>       http://code.google.com/p/pharo/issues/detail?id=750
>> shows the contrary and this is really strange. I think that may be
>> this is a slow down in Transcript but...
>>

i added a comment to this issue.

>> Stef
>>
>>
>>
>>
>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>



--
Best regards,
Igor Stasenko AKA sig.

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: I always thought that 1 to: 100 do: was faster than (1 to: 100) do:

hernanmd
 Igor, you're claiming that every reference to Transcript in every
package and every example should be wrapped in some way by

[ ...  ] forkAt: Processor userBackgroundPriority.

to preserve the historical behavior ?

2009/5/3 Igor Stasenko <[hidden email]>:

> 2009/5/3 Stéphane Ducasse <[hidden email]>:
>>>
>>>
>>> but apparently
>>>       http://code.google.com/p/pharo/issues/detail?id=750
>>> shows the contrary and this is really strange. I think that may be
>>> this is a slow down in Transcript but...
>>>
>
> i added a comment to this issue.
>
>>> Stef
>>>
>>>
>>>
>>>
>>
>>
>> _______________________________________________
>> Pharo-project mailing list
>> [hidden email]
>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>
>
>
>
> --
> Best regards,
> Igor Stasenko AKA sig.
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: I always thought that 1 to: 100 do: was faster than (1 to: 100) do:

Igor Stasenko
2009/5/3 Hernán Morales Durand <[hidden email]>:
>  Igor, you're claiming that every reference to Transcript in every
> package and every example should be wrapped in some way by
>
> [ ...  ] forkAt: Processor userBackgroundPriority.
>
> to preserve the historical behavior ?
>
No, i'm not. It is the Transcript which ensuring the thread-safety.
Transcript is a stream. It could be used w/o being connected to a window.
Please make difference between Transcript stream, which is updated
immediately, and Transcript window which can be updated when it is
safe to do it.

> 2009/5/3 Igor Stasenko <[hidden email]>:
>> 2009/5/3 Stéphane Ducasse <[hidden email]>:
>>>>
>>>>
>>>> but apparently
>>>>       http://code.google.com/p/pharo/issues/detail?id=750
>>>> shows the contrary and this is really strange. I think that may be
>>>> this is a slow down in Transcript but...
>>>>
>>
>> i added a comment to this issue.
>>
>>>> Stef
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>> _______________________________________________
>>> Pharo-project mailing list
>>> [hidden email]
>>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>>
>>
>>
>>
>> --
>> Best regards,
>> Igor Stasenko AKA sig.
>>
>> _______________________________________________
>> Pharo-project mailing list
>> [hidden email]
>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project



--
Best regards,
Igor Stasenko AKA sig.

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: I always thought that 1 to: 100 do: was faster than (1 to: 100) do:

Stéphane Ducasse
In reply to this post by Nicolas Cellier
but nicolas this is quite strange I unset the preference

and I get
        1 to: 1000 -> 3405 ms

        (1 to: 1000) -> 2600 ms

I'm puzzled. Should probably increase the sample....
I redid the experience with exaclty the same setup and I get
        26115 vs 25487 so this should be ok


Stef



On May 3, 2009, at 9:25 PM, Nicolas Cellier wrote:

> I bet it's undo history again!
>
> reverse the order of the two tests and redo...
> or deselect the preference for multipletextUndo.
>
> 2009/5/3 Stéphane Ducasse <[hidden email]>:
>>>
>>>
>>> but apparently
>>>      http://code.google.com/p/pharo/issues/detail?id=750
>>> shows the contrary and this is really strange. I think that may be
>>> this is a slow down in Transcript but...
>>>
>>> Stef
>>>
>>>
>>>
>>>
>>
>>
>> _______________________________________________
>> Pharo-project mailing list
>> [hidden email]
>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: I always thought that 1 to: 100 do: was faster than (1 to: 100) do:

Nicolas Cellier
{
[(1 to: 100) do: [:i | i yourself]] bench.
[1 to: 100 do: [:i | i yourself]] bench.
}.
 #('6935.01299740052 per second.' '44448.7102579484 per second.')

You observe a Transcript dependent behavior, not related to to:do:

2009/5/3 Stéphane Ducasse <[hidden email]>:

> but nicolas this is quite strange I unset the preference
>
> and I get
>        1 to: 1000 -> 3405 ms
>
>        (1 to: 1000) -> 2600 ms
>
> I'm puzzled. Should probably increase the sample....
> I redid the experience with exaclty the same setup and I get
>        26115 vs 25487 so this should be ok
>
>
> Stef
>
>
>
> On May 3, 2009, at 9:25 PM, Nicolas Cellier wrote:
>
>> I bet it's undo history again!
>>
>> reverse the order of the two tests and redo...
>> or deselect the preference for multipletextUndo.
>>
>> 2009/5/3 Stéphane Ducasse <[hidden email]>:
>>>>
>>>>
>>>> but apparently
>>>>      http://code.google.com/p/pharo/issues/detail?id=750
>>>> shows the contrary and this is really strange. I think that may be
>>>> this is a slow down in Transcript but...
>>>>
>>>> Stef
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>> _______________________________________________
>>> Pharo-project mailing list
>>> [hidden email]
>>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>>
>>
>> _______________________________________________
>> Pharo-project mailing list
>> [hidden email]
>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>
>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: I always thought that 1 to: 100 do: was faster than (1 to: 100) do:

Stéphane Ducasse
Great!

This is a nice example to show all the tools we have for benchmarking.

On May 6, 2009, at 8:46 PM, Nicolas Cellier wrote:

> {
> [(1 to: 100) do: [:i | i yourself]] bench.
> [1 to: 100 do: [:i | i yourself]] bench.
> }.
> #('6935.01299740052 per second.' '44448.7102579484 per second.')
>
> You observe a Transcript dependent behavior, not related to to:do:
>
> 2009/5/3 Stéphane Ducasse <[hidden email]>:
>> but nicolas this is quite strange I unset the preference
>>
>> and I get
>>       1 to: 1000 -> 3405 ms
>>
>>       (1 to: 1000) -> 2600 ms
>>
>> I'm puzzled. Should probably increase the sample....
>> I redid the experience with exaclty the same setup and I get
>>       26115 vs 25487 so this should be ok
>>
>>
>> Stef
>>
>>
>>
>> On May 3, 2009, at 9:25 PM, Nicolas Cellier wrote:
>>
>>> I bet it's undo history again!
>>>
>>> reverse the order of the two tests and redo...
>>> or deselect the preference for multipletextUndo.
>>>
>>> 2009/5/3 Stéphane Ducasse <[hidden email]>:
>>>>>
>>>>>
>>>>> but apparently
>>>>>     http://code.google.com/p/pharo/issues/detail?id=750
>>>>> shows the contrary and this is really strange. I think that may be
>>>>> this is a slow down in Transcript but...
>>>>>
>>>>> Stef
>>>>>
>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Pharo-project mailing list
>>>> [hidden email]
>>>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>>>
>>>
>>> _______________________________________________
>>> Pharo-project mailing list
>>> [hidden email]
>>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>>
>>
>>
>> _______________________________________________
>> Pharo-project mailing list
>> [hidden email]
>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project