Speed differences between MC and Rev

[Richard Gaskin]

Wilhelm Sanke wrote:
> I resend my post of last Friday, hoping the server is up again:
> ========================
> 
> On Sun, 24 Sep 2006 Richard Gaskin wrote:
> 
>>
>>> I notice no difference in the execution speed.
>>
>>
>>
>> So far this is consistent with all data except Wilhelm's. Neither 
>> logically nor empirically have we yet found a way that simply adding a 
>> library can affect execution speed.
>>
>> We'll have to wait for Wilhelm to provide more data.
>>
>> (snip)
>>
>> If instead there does seem to be some difference in the way 
>> standalones are made (though after building my own standalone maker I 
>> can honestly report that I can't imagine what that could be) it's also 
>> worthwhile determining where the performance goes.
>>
>> Ideally Wilhelm will be able to come up with a stack that presents an 
>> isolated recipe for this problem. It's my understanding that at the 
>> moment the tests are being run within a very complex collection of 
>> scripts, making it difficult to produce a generalized recipe for 
>> others to run.
>>
>> Without such an isolated example of this reported issue, I can't think 
>> of any way it could be reasonably addressed further beyond what's 
>> already be done.
>>
>> If he's able to deliver an isolated example I'm sure RunRev will jump 
>> at the chance to optimize based on it reveals.
>>
>> -- 
>>  Richard Gaskin
> 
> 
> Back again to resume the discussion.
> 
> I have produced this isolated example. Apart from the one-line script of 
> the "reset" button, it contains only a 13-line script of the test button 
> "duplicate colors".
> 
> Only the MC cursors stack is included as a substack, otherwise nothing 
> is included, not even an "answer dialog".
> 
> The performance differences are as reported about 600 milliseconds 
> between MC and Rev on a Windows 2 GHz machine, tested with Metacard and 
> Rev using the same 2.7.3 engine. No differences come up for the stacks 
> compared with standalones.
> 
> A test with 2.7.4 today shows an overall performance improvement for 
> both IDEs of about 200 milliseconds, but the speed difference between MC 
> and Rev of now about 500 milliseconds remains.
> Stack and standalone with MC show an identical performance - around 1550 
> and always below 1600 milliseconds. For Rev - 1935 to 2083 milliseconds 
> for the standalone  - there is an additional difference between 
> standalone and stack of 100 milliseconds, the stack running in the IDE  
> with a peak of 2165 milliseconds being the slower one.
> 
> You can download the test stack from here (2 MB because of the embedded 
> picture and the compressed imagedata to enable reset):
> 
> <http:/www.sanke.org/Software/mc-rev_speed_test.rev>
> 
> or type
> 
> go URL "http:/www.sanke.org/Software/mc-rev_speed_test.rev"
> in your message box.

Good work, Wilhelm.

Confirmed in the IDEs -- 5740ms/avg in MC, 6830ms/avg in Rev (I have a 
1GHz PB G4).

I haven't built standalones, though, which would be good to test.  Given 
this speed loss I'm confident the folks at RunRev will take a keen 
interest to determine what's eating performance.

Offhand I can't imagine how the execution of a single handler with no 
breaks, pauses, or sends is in any way affected by any outside script, 
but it does indeed appear to be the case.

Have you BZ'd this?  It would be good to include that file as an 
attachment to the report.

-- 
  Richard Gaskin
  Fourth World Media Corporation
  ___________________________________________________________
  Ambassador at FourthWorld.com       http://www.FourthWorld.com