Stereo Tool

To be honest, both of the things you're working on right now do not have any significant value for me personally. I don't have a real large audio collection, but out of hundreds of tracks, the declipper only has any noticable impact on one track - an overdriven rip of "Goodbye Horses" by Q Lazzarus. As for the Dynamics Booster, while you all that are audiophiles might knock X-Fi cards, the fact remains that Creative's Crystalizer does what the Dynamics Booster is attempting to do, but does not consume 50-60% of my CPU.

So, I could really choose to ignore testing of those two things, but I don't. I feel it's my duty to report to you that the performance issues make it to where the value of those items are so far out of line with any value added that they should be investigated for inefficiency.

That said, you have mentioned that you feel that multiband is in need of a rework, and that you feel that there are problems in the Loudness filter. Both of those items would impact everyone, not just a small segment of your users.
Your CPU has more L2 cache. I suspect I'm right about the FFT ops running within the cache, or if they need main memory access, you have more CPU-to-RAM bandwidth available, so the CPU doesn't have as many stalls waiting on memory.

Also, there is still the possibility of an AMD penalty with the Intel compiler and/or Intel libraries.
I'll look. Odds are my system will handle it much more efficiently. The positions are reversed. My AMD has a larger L2 cache vs. a P4 (1M vs. 256K or 512K), as well as an on-die memory controller. Athlon64 processors ran circles around Pentium 4. Now, Intel processors have larger L2 and even L3 caches, and with the Core i series, the memory controller has been brought on-die.

Edit: The oldest version I've been able to find is 2.xx...

I do understand what you're saying, but if you look at the discussion between me and Hans, you'll note that he agrees with looking into optimization. Based on what he said about the expected final load of the dynamics booster being around what Multiband's load is, instead of a 50-60% consumption, I'm anticipating 2-5% consumption for the final product. I have no complaints about 2-5%, but I do about 50-60%. I hope you get it...

I understand that. My plan was to build a new system in late 2009. Unemployment put a stop to those plans. I have to work with what I have. Keeping health insurance and paying my bills until I am able to find a company that will hire me is much more important than acquiring a new computer. Finding a company that will hire is more difficult than it might seem, for a variety of reasons. I was recently rejected by a company for an entry-level system technician job that handled data backups. I was given a reason of that it was felt I didn't have the skillset. Given that my unemployment was self-inflicted (I made a mistake in judgement), it's my belief that the reason I was given was not the real reason, especially given that my previous job duties were intertwined with people doing backups, to the point of the fact that I specifically worked with them to compress data so that the backup process took less time...
If Michi95 hadn't taken the aggressive tone that he took, this discussion would've ended when Hans told me that he anticipated the load to be around what multiband's load is... Could I have just ignored Michi? Perhaps. I was just not in the mood to be mocked that day.

Again, mea culpa.

Exactly! but... is it automatically ? I think not.

You cannot do that without some Dynamics detection/detecor (?) You don't know my input signal level! Maybe it's around -18dB and maybe it's at around -3dB...

If it's Highly compressed material - add more this filter
If it's already dynamic - add ass little ass possible
Then it's all usable.

Leif explained that very well! His Dynamic detection filter is even 5-band. + expander

Bojcha, the idea is this: If you play 2 CD's, on the same player, and one is loud and compressed (recent CD) and the other isn't (1980's CD), then the Stereo Tool input level is already very different - much higher for the louder CD. This assumes of course that no gain has been changed.

What I do is basically making loud sounds even louder. So if the input is louder, the effect is bigger.

I might re-introduce the detection (that was there earlier, in a version that I never placed online) later, but first I want to get the processing right. And right now it's not.

ok.
Phail xD

Just maybe stupid idea...

Is it possible to split and put 'Dynamic Boost' every band in MB? How that would sound ?
Ofcourse this same adjustment for every band separate.

or to 5bands...

Bojcha: If I would place it after the AGC, that's exactly the same as adding it in Multiband (actually it's already in there, I have several Multiband instances, one does normal multiband, one does stereo boost and one does dynamics boost. Combining those functions is as easy as typing 'true' twice when creating the instance...

Ok, I do confess.
I am guilty.
I am the bad boy here.

That's the reason why I think that this: is a very good idea.
More fun for Brian's CPU !