HIFI examples overflow audio range on avr about mozzi HOT 3 CLOSED

Note that this part of the code is for STM32 (see #elif on line 491). The AVR HIFI code is around line 780, and was not touched, recently.

I'll have to admit that I never included HIFI in my testing, so I cannot tell whether it may have been overflowing, before I started. But I do not think I have touched any code for that while adding the STM and ESP ports, or merging any of the other recent PRs.

Could it have been this one from three years ago: b22a84c ? While that should not change the scale in itself, it will certainly cause values above the range to be handled badly. ((B1000000 00000000 << 1) >> 8 == B0, while previously B10000000 00000000 >> 7 == B00000000 10000000).

from mozzi.

Thanks Thomas. Apologies for dumb questions, as I'm out of touch with the code after a long time adapting the Mozzi web page to work better with GitHub (which I hope will eventually make it easier for others to modify the pages).

I went back and tried the release version, then the current code again and it worked no problem, without any change... wierd.

When adding the <<1 then >>8, I assumed it to be faster then >>7 after tests years ago when >>7 produced 7 single shifts... but this may no longer be the case with more recent compilers. Anyway, I'll put the issue down to corrupt memory or something, and hopefully do a speed comparison sometime.

from mozzi.

No problem. Glad to have you back!

My theory about the shift was probably wrong, too, as this problem would only kick in when the output is already at the topmost bit. That would seem unlikely to become relevant, as the output value would need to exceed 14bits a lot in order to reach that point. Well, glad the problem went away by itself...

from mozzi.