Stereo Tool

I've also found some wierd (maybe a type of render or re-draw related?) anomaly when using the mouse scroll wheel after clicking on some of the drop down type menus like the RDS PTY box or sound card input selection. Wondering if anyone else has the same?

Higher math stuff is not my area of expertise, but I think 32-bit single or double precision is fine. I just don't know how everything is handled internally by the compiler and/or IPP. One thing you said though is "nearly everything". The last question asked of you on the Intel forum was about mixing different precisions.
According to the Intel documentation, that call shuffles 4 32-bit values, so most likely ok.

At any rate, more instructions in flight probably leads to impacts on the L1/L2 cache.

Might take a look through the AMD optimization guide for K8 and see if they have any suggestions that you may not have tried. Barcelona (Phenom) and newer have more SSE bandwidth available.

http://support.amd.com/us/Processor_TechDocs/25112.PDF

Report : Hans you said that preemphasis indicator is ok
But it is ok when i enable the Normal Output . When disabled - Preemphasis indicator is gone ! You probably will check that ?

Yes that's normal - if no-one uses the de-emphasized sound I don't calculate it hence I cannot display it. I would have to do extra things, which cause the CPU load to go up, just to display something a bit brighter - so I don't.

This makes sense: When you load a new preset, first all the settings are reset - including whether there's FM output. Then it's switched on again. So, depending on when it happens, a block of samples that should have been sent to the FM sound card may be lost.

I know this is bad behavior - but since normally you wouldn't be switching presets when broadcasting I don't see it as a high priority issue.
I'm going to make a new MB, and that should include this feature.

Hans, have you tried the demo of Ken Watabe's CP5 VST-plugin ?
http://www.x-buz.com/BuzComp_Beta.html

viewtopic.php?p=17936#p17936
That complex 5-band architecture includes such 'solo band' feature !

Do you have already a fixed decision about the range of individual bands for the new MB ?
3 - 7 ?
Or will it be more flexible ?

I know that 99 percent of the users have the opinion that 10 bands are too much.
But IMO 10 bands are only too complex for most processing situations.
10 bands are necessary to keep a strict octave-band scheme.
But there might be even rare situations (not for broadcasters, but for restoration purpose) when you need more bands.

http://www.trifex.de/buzzrizerlight.html
(it is a freeware VST-plugin !)

Ah, okay, thanks! Does this mean that the normal and LQ settings are not reset when switching like the FM output is?

I imagine that most people will not be switching presets regularly during broadcasting but what about STS file polling?

Have you noticed such behaviour Hans?

And why 10 bands to be too much ?
You give us an example that is even more than 10 !
10 bands is something like optimum if you want to feel real modeling of sound .. I think that 5 bands are even less for this purpose and more than 15 is just greedy. But I must be objective and say that professional equipment has even more ( i know middle class Sony that have 20 bands equalizer ) . Also in the studio you all know that bands are more than 30-35 ?!? So don`t even say that 10 bands are tooooo much ! Don`t even think about it

Happy New Year to all

Your are talking about equalizers Static processing !

But our controversy here is about multiband compression/expansion - you can call it dynamic (full parametric !) equalizer.
So the professional music production multiband compressors/expanders have 3 or 4-bands (or maximal 5-bands like Watabe's CP5).
Except some rare dynamic equalizer plugins with more bands (but less parametric options than traditional compressors).