tutorial - basic optimizing and building tricks in sm

sunsynth · 21 Apr 2009

Yout Tuts are really usefull and good commented Michael 🙂

Keep on doing your great work!

Walter

drumma · 21 Apr 2009

Agree with Walter,

Thanks a lot!

D.

Nu Audio Science · 21 Apr 2009

Micheal is it OK if i consolidate these into a sticky ?

Cheers NAS

MichaelBenjamin · 22 Apr 2009

yes, you can do as you like.

infuzion · 22 Apr 2009

MichaelBenjamin wrote: here is another tutorial, giving some general building and optimizing advice.

Thanks for the "remove clutter"; proves my theory that module reduction does indeed speed up OSMs!

But I still hate you for giving away my secrets. 😉

infuzion · 22 Apr 2009

Nu Audio Science wrote: Micheal is it OK if i consolidate these into a sticky ?

IMHO, I hate stickies. Things like this are much better in the WiKi.

aliasant · 22 Apr 2009

GREAT STUFF Michael ! !

Keep it comin.

Nu Audio Science · 22 Apr 2009

infuzion wrote: Thanks for the "remove clutter"; proves my theory that module reduction does indeed speed up OSMs! But I still hate you for giving away my secrets. 😉

Was it you that found all these out was it Infuzion, Why didn't you post em in the WIKI man these are reaklly useful tips In answer to your other comment i prefer stickys in the forum , It's a matter of taste and if somebody else wants to put em in the WIKI thats cool 😉 I prefer a different kind of layout in WIKIs to the SM one but it's each to his own

Cheers

NAS

Tom7777 · 22 Apr 2009

These are good tips but it makes me want a decent default cpu analysis module added because half the time when i ran them from your examples it sometimes returned the unoptimized versions to be faster. The test module you use seems too random and all over the place like for example it was returning the cluttered module to be faster than the un-cluttered and then i closed SM and loaded the example again and it swapped around and then said the un-cluttered was faster and it has done this a few times after testing. It's kind of annoying to optimize in SM due to the fact the tools available don't give me decent readings, i know these tips work though.

aliasant · 22 Apr 2009

Tom7777 wrote: These are good tips but it makes me want a decent default cpu analysis module added because half the time when i ran them from your examples it sometimes returned the unoptimized versions to be faster. The test module you use seems too random and all over the place like for example it was returning the cluttered module to be faster than the un-cluttered and then i closed SM and loaded the example again and it swapped around and then said the un-cluttered was faster and it has done this a few times after testing. It's kind of annoying to optimize in SM due to the fact the tools available don't give me decent readings, i know these tips work though.

I had the same finding when looking at the uncluttered comparison. Strange. Maybe it depends on your cpu?

infuzion · 22 Apr 2009

Tom7777 wrote: ...because half the time when I ran them from your examples it sometimes returned the unoptimized versions to be faster.

Depends on the CPU... the more modern CPUs, esp the Core2, self-optimizes very well. You really have to test on a few different CPU, like I do with my laptop & desktop.

MichaelBenjamin · 22 Apr 2009

yes, thats right - sometime the cpu analyser seems to screw around, and i need to reset it, delete and readd it to show improvement as expected.

MichaelBenjamin · 22 Apr 2009

Tom7777 wrote: The test module you use seems too random .

thats the best i found. do you use another one?

Tom7777 · 23 Apr 2009

MichaelBenjamin wrote: Tom7777 wrote: The test module you use seems too random.

thats the best i found. do you use another one?

Well i have a few saved that i try, they all seem to be jumpy though but i think it's due to the fact they are made using modules that probably use cpu themselves or introduce occasional denormal like behavior due to that fact. So far i haven't seen/made one i found to give trustable results which is why i suggest a dedicated module that could possibly be far more accurate.

aliasant · 23 Apr 2009

Tom7777 wrote: MichaelBenjamin wrote: Tom7777 wrote: The test module you use seems too random.

thats the best i found. do you use another one?

Well i have a few saved that i try, they all seem to be jumpy though but i think it's due to the fact they are made using modules that probably use cpu themselves or introduce occasional denormal like behavior due to that fact. So far i haven't seen/made one i found to give trustable results which is why i suggest a dedicated module that could possibly be far more accurate.

I agree on that. Something I have wanted for a long time. Maybe Malc can realize some of our wet dreams now that the latest betas seems to be more and more final.