Stereo Tool

I've seen the same here also red output meters sometimes.

O, have you ever tried that in Win7 when you had performance issues?

Actually, does this have any effect in WinXP? That's surprising - based on the numbers that I see here (but that's Win7) but also numbers reported by others, it appears that ST's processing is running on 2 cores; 1 core for the main thread and one core for the 'offloaded' threads. Which shouldn't change by setting affinity. I have not tested this in a WinXP install though (I only have a WinXP VMware image, but that might behave differently).

That shouldn't be possible, so I suppose it indicates a bug somewhere.

I have only 2 cores. I suggested him to try to set affinity to 2 cores because his Q6600. As i said erlier.. Window7 works with cpu totally different then XP.
Also.. i noticed with other programs and games.. that sometimes even on my 2 cores some programs/games works better with with one core then with both.

What i think for W7 that it still makes "higher priority then realtime" for some system services. Also 64bit OS is just not not for ST!

If ST works wih 2 cores Q6600 will work with affinity to 2 cores better - if i am right.

And i also saw few times red bars at output. Maybe 2 or 3 times. So very rarely.

Cant really correlate to any specific program material, yet. Happens here about 5 to 6 times a minute. Will be able to test on a fast i7 in a few days. Will watch it more tomorrow. I have screen refresh set low. Let me watch at higher speed tomorrow.

I am running 75 preemph, stereo, rds, output maxed (0vu checked). No obvious artifact in the output, when the output meter turns red, at least so far.

Restricting affinity to 2 cores definitely helps. I learned and saw something new in Win XP today: 'affinity'...had no idea!

I meant to make it fully user configurable since I sometimes need filters in different order.

Small cosmetic glitch: the order of buttons isn't consistent for all sidechains.

For older versions of ST there was the recommendation to set the upspeed of the AGC to 10 dB/sec for highly dynamic content since there was a brake build in anyhow. Is this still valid or what else should one do?

The relative threshold of the AGC compressor can only be lowered to x0.5 (-6 dB) which is sometimes still a bit to high for me. Can this please be lowered or isn't it intended to be used that way?

I suddenly don't think the output flashing red is any bug in the software, I think it's the limitation of the core2quad E6600.

I'm only seeing the output flash red when SSB is activated...I can get it to disappear when the quality slider is reduced to around 30 while SSB is on. If SSB is off, I can get the quality slider up to 70% and no appearance flashing red.

Affinity in task manager: 2 cores.

may be a known issue...when composite clipper is on...the meter below clipper drive (loudness) towards the left of (performance: no composite for low frequencies) freezes/disables. at least here, it does.

I was relying on MpxTool for the channel separation measurements, but it seems now that the actual separation is not 30 dB as MpxTool reports, but close to 60 dB in SSB mode. In multicore mode, at 60 dB below the level of the other channel there are clicks present every 46 ms or so (for any real world purpose these can be considered to be inaudible), and these peak to about -60 dB below the audio level of the other channel - without these peaks stereo separation would even be a lot better.

I added a different type of SSB mode using Hilbert transforms, the advantage is that I can make a real SSB mode (the upper sideband is really empty, no area around 0 Hz that's DSB encoded). It works fine in single core mode, but with multiple cores there are issues - I'll try if I can fix them tomorrow, if that fails I'll just forget about this Hilbert transform (the advantages, if any, appear to be very small, on top of that the CPU load is higher).

no red flashing red output issue with ssb on since just upping to an i7...now testing for real.