Stereo Tool

Graph is very very nice!!
BTW what type of filters was used in new MB?

_________________
Urban Junglist

@Brian: I agree that 'hard crashing' is not very polite, of course these are beta versions and the errors should be gone in the release but it might be good to add some protection around it. Don't agree about the information though.

Take for example this:

f(x) {return 1 / (x + 1);}
g(x) {return 1 / (x - 1);}
h(x) {return f(x) * g(x);}

As you can see, this can give a 'divide by 0' in both f and g. But a smart compiler, instructed to ignore exceptions (!), can, using the fact that 1/a * 1/b = 1/(a*b) which is faster because divide is much more expensive than multiply, rewrite h to:

h(x) {return 1 / ( (x+1) * (x-1) );}

Thus moving a possible exception to function h. And h is still small enough to be inlined, moving exceptions further up the chain.


Also, take this code:
f(x) {return x / x;}

If I instruct the compiler to ignore exceptions it can replace ANY reference to f() by 1. Which is of course MUCH faster.


Exceptions can also move in the opposite direction:

a = h(x)
b = 50
i(a, b)

with
i(a, b) {return (b < 20) ? a : 0;}

This will be rewrittten to:
(b < 20) ? h(x) : 0

So the exception caused by the calculation of a if x is +1 or -1 is moved into the function i and only occurs if b < 20.


If I instruct the compiler to keep track of where these exceptions occur, ALL these optimizations would immediately be disabled.

After a long time playing with Bands... program HALT It worked with sound but the ST window do not drawing.

_________________
Sexy Girls in your city

Any idea what you did when this happened? Maybe changed the number of bands?

yes i was changing the bands number for a long time.

_________________
Sexy Girls in your city

All of what you posted sounds nice, but here's the thing - I haven't been able to find any sort of documentation that explicitly backs up what you're saying over the course of about 4 hours of searching.

The only thing I have been able to find so far is a potential implicit issue if you use -fp-model fast. That said though, the documentation for that model says that exception specifications are disabled by default. If exception specifications are disabled, then I would think that the compiler would still do all the "smart" things. BTW, I've also read that exception specifications are not so good, so this might be where your reservations come from. However, there are other ways to do exception handling, specifically table generation / zero-cost.

Several weeks ago, I had asked if you could PM me your compiler options used. I got no response.

Just like with asking for the debug info, I think you're hung up on the idea that I want to steal your code. I don't.

When I looked at the adjustment sliders for the band steepness I was briefly confused by the terms 'Steepness left' and 'Steepness right' thinking you were referring to the left and right channels.

How about terms like 'steepness lower' and 'steepness upper' to refer to the steepness of the lower and upper edges of the bands?

Maybe I'm totally off the mark here...

In compressor window, the ''treshold level relative to agc'' slider isn't working like all other sliders, for example using the mousewheel

_________________
The Power of Sound

''Music is a moral law. It gives soul to the universe, wings to the mind, flight to the imagination, and charm and gaiety to life and to everything.'' - Plato

You're the second person saying this so it needs to be changed. But lower and upper... I wanted to change it, but then I thought - for me it would make sense that 'lower' is where the signal gets lowered, so the right side. So I find this term confusing as well. Any other ideas? Rising and Falling comes to mind but my logic might differ from other people's logic.

I need to remember to check all the new sliders for things like this before releasing this.