anatolyj's Recent Posts
Hi Randy, folks, i made some test, well, initially for myself, but maybe it can useful as some info (well, for example, for comparison with future possible performance improvements).
Immediately I must warn about my config, since slightly outdated CPU (Intel Core2Quad), but here the main thing is the approx difference ratio, right.
Other info: Win7 64bit, Studio one v3 64bit, ASIO setting 44100Hz \ buffer 512 samples.
Cpu usage value was taken from Studio one CPU meter (thus, presented values this is usage for one core, while remaining cores are idle).
About value range: minimum value was taken during closed plugin GUI and no active daw transport. Maximum value - opened GUI and running transport play (witth empty daw sequencer).
"Default state" - plugin factory setting immediately after loaded.
- default state (with max 4 voices): v181 - 25-30% ; v182 - 20-24%
- default state with 1 voice: v181 - 8-10% ; v182 - 6-8%
- default state (4 voices, resonator and body - disable): v131 - 25-32% ; v132 - 18-24%
- default state with 1 voice and enabled resonator and body: v131 - 16-19% ; v132 - 15-18%
- default state with max 8 voices: v131 - 65-80% ; v132 - 44-55%
- (extreme setting) default state with max 8 voices and resonator and body enabled: v131 - 85-100% ; v132 - 60-73%
- default state (1 voice): v131 - 16-19% ; v132 - 16-17%
- default state with max 4 voice: v131 - 38-45% ; v132 - 38-43%
So, Aalto and Kaivo results good, the difference is pretty noticeable, but strangely about Virta, for some reason no difference almost (maybe virta was not improved as aalto\kaivo? Or some fault on my side).
Thanks Randy, cpu usage drop detected. (Made some comparison, more details in separate topic).
Hi Randy, perhaps i wrote already about it, forgot. Would be good add option for change Kaivo samples location (for use in non-system partition, this will useful for add user samples to kaivo library).
Randy thanks, would be good.
Randy, i also reproduced stuck notes just now in Aalto and Kaivo (181 and 131 respectively, win7 64bit vst), in both cases in default patches. Host Studio One v3, just i walk by mouse cursor at Studio One pianoroll and really some notes in some voices randomly stuck occasionally, from time to time. I can make short video if it can help.
Added: i check just now in Virta (through instrument track) - the same issue.
And, in all cases it is convenient to watch, via stuck lightbulb(s) in internal [MIDI] indicators.
Added 2: check in Ableton ive 9 - the same. (Again, via mouse+pianoroll)
But what strange, i can't reproduce stuck's if i play through "qwerty" keyboard method (which internally exist in both cases - Studio one and Live 9). Hmm..
Probably the reason in a more rapid alternation of notes at the play via mouse cursor, which is probably more stressful test :D
Randy no no, i understand how to make it, i just wanted to make sure that described behavior is properly, i.e. i supposed that the maximum value of seq "column" does not have enough (planned) power (assumption based on the logic of impact to some other parameters), something like bug. Just was needed confirmation that no, and I will be calm :D
Some question about modulation value.
I modulate LFO rate from SEQ (In Kaivo for example, the same i guess in Aaalto).
LFO rate level set to default stage (1.00)
LFO rate modulation level (lfo_rate_p) set to minimum (-4.00)
Sequencer columns value set to maximum (1.00)
Ie this condition reproduce maximum impact from the sequencer to LFO rate, but LFO continues movement nevertheless :D So, this is should be ? (ie normal behavior?). I mean, really lfo rate not should drop to zero at this condition ?
Or by other words - the maximum value of Seq columns not in any case will have maximum impact to parameter with default value stage ?
Randy thanks, now works properly ! (sorry for delay reply, i just woke up).
(But no, problems was not just in installer, version number in UI corner was exactly 131 too, usually i always check this while testing new version, but no essence, thanks anyway for quick fix !).
Randy, all the same in 131. (win64)
Yes, i can confirm, in Win 64bit version the same behavior (about mod.inputs for Nonlin, Tone and Pitch).
Oh damn, unexpectedly, i even missed about betas (though it was 9dec), thanks Randy ! Good.
About Carrier Air factory preset (from Ed Ten Eyck section), just i noticed that it not produce any sound. I also tried change input\threshold levels for incoming audio, but all the same. Maybe it for some specific purpose (or maybe some issue)?
:D Oh, then all ok. (Just i thought some problem on my side).
Just tried Virta demo, very good. Some thought - would be good to add formants freeze button, i mean if one taken formants moment (while it work in vocoder mode) would be sound infinitely. (Somethin like spectral freeze)
Althought 45% it is quite a much indeed (for 1 voice ?), but Aalto never been a lightweight for cpu, in my system Aalto looks like that (in default state, GUI close):
At voices amount 1 - 12-13%, 2 - 20-22%, 3 - 29-30%, 4 - 37-38% (all results - for one of core of CPU (quad), ofcourse).
Windows 7, Studio One v2 64bit, buffer size 256
Added: But btw just now when make cpu test, noticed some strange behavior, when at 4 voices cpu show the same amount as at 3 - about 28% (Aalto just was inactive, in the backgrounds in daw), i could not understand what the reason, interesting. But such trick i could not repeat with 1 voice, unfortunately :D
It will be necessary then to explore it better.
Randy check please 121 installer distributive (Win 32bit vst versia), i uninstall 120 then installed 121 but version on GUI marked the same as 120, and the same cpu behavior.
Also as evidence the fact that the host does not read (not re-scan) the updated dll.
Probably in the installer still contains the previous dll version?
Hi, i noticed that some users have avatars, but i can't find such button in my account settings :D
Oh clear, thank.
Randy I sent email.
I tested 1.2 just now, unfortunately - i compared with 1.1, CPU loading increased significantly.
One of my patches which use 36-40% cpu in v1.1, load with crackles now about 100% of cpu.
After all, it should have been are optimized?
System: Win7 64bit, CPU intel, Kaivo VST 32bit; DAW Studio one 32bit
Randy Ok, you're right, look at the possible responses. In an extreme case, I think to completely reinstall the system.
Then I do not understand anything.
Basically, I really noticed for some time that some of the recent updated plug-ins (from different manufacturers) also began to cause sometimes CPU spikes or increased CPU load.
I do not exclude that the problem is really on my side. (Although usually i watch the cleanliness of the system).
But this case is just broke all tops.
Although .. maybe it's fortunately (if it really would be my problem) because this my Kaivo problem is clearly pronounced (as I have said - increasing load by more than two times), and the impact of any changes to it will be easier to monitor during the process of setting up the system.
(Maybe I'll even say thank you for that as a result :D )
But for now I'm in shock.
Well, I will try to understand what is the reason.
Maybe I'll send to email the video as proof.
"Your patch works OK here. try without the plugin window—maybe it's a graphics issue?"
Randy, the same and with closed plugin window. (for just in case - video GTX760 ,desktop).
Sounds good :)
Just checked now - o very good, incoming midi issue (for Mod CC) fixed now ! Thanks Randy.
BTW why dll began to weigh less? (about 2mb) :D
the same as whole installer (Win), about 5 mb.
I can't wait for Aalto update :D (with fix for early described issue with incoming midi).
Oh and new plugin it ofcourse very good news.
Just as idea. Implement to Aalto engine some eco \ lo-fi mode, wich can be more lightweight for CPU, but as optional mode ofcourse (by wish). And which would allowed the degradation of sound quality (for example sometimes I really find a cruel aliasing as a interesting FX).
Thus, i think it would be also interesting as some new timbre way (include for already existing presets, it would be fun to see to them with another angle).
And even better if such modes would be more than one :) (with different algorithms).
Randy thanks for reply. It would be really good. I do not know, in my opinion is a very important detail (because one of the most important Aalto aspects it use as a generator, without bind to the sequencer).
Also need mention that as i said in v1.5 it was work.
Yep,"good" optimization. No, nonetheless i hope it's a bug.