Stereo Tool

Thanks Guys. Bojcha - I did not immediately notice you had Post Amp set below 0 for R-Type preset. Look forward to the FM version of this preset - and indeed many other presets designed around the new version.

I assume BPP may have composite limiting/clipping enabled, and that (and RDS) does explain it. Also looking forward to new composite limiter in ST, which as I understand it's likely design, results in less loudness/modulation loss when enabling RDS.

There seems to be some better spectral and 'source to source' consistency in latest beta compared to the beta of a couple of weeks ago. Sounding better.

Sugestion about RDS phase to pilot is not optimum now, recommended is shift 90 degrees from actual. It provide about less 3% peak2peak apmlitude without sound.

_________________
Urban Junglist

No! You cannot just change this, in Germany there is (was?) some different standard that uses a signal placed at a 90 degree angle at 57 kHz, so on some German radio's RDS might not work it it's not locked correctly to the stereo pilot.

I'll check in more detail later, but from your video I see:
- Frequencies of band 1 and 2 are both very low (don't know how steep the separation is, could you tell me this?)
- Input is EXTREMELY loud, no AGC etc.

I know this issue exists, and it exists in all the filters. I know how to fix it and I will - but unless you can actually hear this effect in music, I would say there are far more urgent things to do.

I was looking at optimization opportunities and it looks like I can replace a logarithm (responsible for 2% of the total CPU load) and 2 table lookups by a single lookup! That should lower the CPU load by a few %... And based on what I'm reading the difference should be biggest on older systems.

I'm trying to squeeze every last bit of performance out of the compressors because I intend to use them for many more things in the future.

Stereo Tool V8 ?
Now I understand why we have the drive parameter in the new multiband.

I don't think so, i have many many schematics where hardware RDS and stereo decoders are completly independent, maybe in Germany was something different... I don't know... In Poland it's doesn't matter 90 or 0, some, station has 0 some 90. Specification allows RDS subcarrier 0/180 and +/-90 to 3rd harmonic of pilot tone, +/-90 is better.

_________________
Urban Junglist

Band 1 is 40 hz , band 2 is 90 hz and steep is 32... just like all others and looks good and same at the graphic display of the bands . ( In fact there is extremely high CPU usage when mouse is over that display) . So in this case there is no AGC( just to see the effect clearly) but if i turn it on things are just the same ( only the effect is a little bit decreased ) . I do this just for good picture of the wave shaking. There is some wrong in the first two bands , cause if i disable the first one and enable only the second, still this effect is there . I test it first band with the last available steep of 64, no luck ( it`s just i don`t change nothing ).. so it`s not from the steepness. Cause AGC is also a type of filter( first band) but no any shaking from it ? signal from it is just clean. Look what code you type for it and may be implement it to the MB bands ?!? i don`t know

Yes i agree that there are other things more important ... like allow us to use more "Hold" in the MB..I have a settings that needs more of it.

Have you made sure that you are not running through extra / unneeded loop iterations, ala the previous multiband?

What about the segments of code I posted the other day, which are what you posted on the Intel forum around 3 months ago?

Have you taken any time to look into what Process Explorer and Process Monitor can do for you?

^ Those tools don't just have the potential to help me with my older system, but they can help you for all systems, even current systems. They dig deeper than Task Manager, and can see some things that debuggers cannot, because they are written by Microsoft employees, and those employees have access to both disclosed and undisclosed kernel / API information, including the HAL (Hardware Abstraction Layer).

@Brian: I have checked all the vectorization reports of performance sensitive functions, and for nearly all of them I either made them vectorize or I understand why they aren't.

In the current version (tested with phantoms 80s preset) the 'unnecessary code' uses less than 0.5% of the total ST CPU load. So that's really not much.

The changes I made today should give a reduction of about 4% in the total CPU load (on the most active CPU core; reduction should be bigger on a single core system).