Stereo Tool

Hi Chris,

That's a bit worse than what I see (+13%) - but still not very extreme. The problem is that 2 people reported an increase of more than a factor 2 (!) when comparing 6.20 BETA028/030 vs. version 6.10. And the only big change I can find is this S-protection filter, so I'd like to know if for them the CPU load is back to normal (close to that of 6.10) without it.

I will also optimize the S-protection filter (when I created it I had NO time at all to look at the performance, and I call it everywhere where it COULD be useful - I still need to check if it is really necessary to call it that often).

yes hans... actually the diff is about +7-9%. My cpu works on 48-52% all the time...!
You have to optimize as much as you can !

_________________
Sexy Girls in your city

I meant relative diff. In your case: 48/38 = ~ 26%.

I'll try to optimize it.

@Hans:

Yes better CPU usage now in BETA 31. Thanks.

Only thing I see a problem with now is reaction time. And it is NOT the AGC, it is something else I can't figure out.

Good work fixing CPU load issue.

Baires, what if you keep S protection turned on? There was also a bug in the previous version which caused crashes, hence unpredictable behavior, hence - maybe - also performance issues.

Not Baires, but it is definitely the advanced highs that's causing a large CPU load increase. I just tried B31, and with it enabled, CPU usage is running 65-70%, but with it disabled, back down to 38-43%.

Edit: That's also perhaps why I so strongly disliked the declipper, as I think both of those got implemented at the same time (?), and it would absolutely choke my computer when I enabled the declipper. Since I couldn't disable the Advanced Highs, it stands to reason that it may have been a cummulative effect...

btw - 'advanced extreme highs protection' is usefull only for FM mode because pre-emphasis! It is needed to work different for non-FM mode.

Hm, you're right, I completely overlooked that. So I guess I can just turn it off for non-FM mode.

More good news: I've checked the S-protection code, and it turns out that there's a rather complex calculation (involving, among others, 18 (!) Fourier transforms), which with the default setting ALWAYS results in exactly the same value. So I can completely remove this calculation!

- Oops. It IS used.

If that filter will stay only for FM mode, maybe is better to be part of MPX Compposite clipper/embedder or whatever name is, ofcourse if is still plan for that.