Is it really sick to have 20 synthesizers?
- KVRAF
- 3889 posts since 28 Jan, 2011 from MEXICO
I am jaded of trying, installing, uninstalling and learning software synths.
I just stick to a couple now.
I just stick to a couple now.
dedication to flying
- GRRRRRRR!
- 15844 posts since 14 Jun, 2001 from Somewhere else, on principle
How much actual learning does it take? At the end of the day they are all synths and the basic paradigm is mostly very familiar. I find I only need to learn a few of the more in-depth things that vary from one to another but I can mostly install something and start using it immediately, even something as out there as TRK-01 or Cyclop.
For a start, I don't use a "DAW", I use a virtual studio, and checking Orion's CPU meter with Task Manager, it actually reads high. The fact that combining the patterns into a single instance only reduces CPU use by 5% compared to two instances would tend to suggest it has little to do with multi-core performance, wouldn't you say?machinesworking wrote: ↑Thu Mar 21, 2019 9:09 pmAlmost every DAW I know of reads the cores ... uhm.. weirdly, i.e. it takes the highest core as the percentage, not the percentage of a combination of your cores. So a single instance of a plug in ramps to 45%, which if you have a four core machine isn't a real number, since it's 45% of one core.
I don't need to "test" it, I have seen it in production situations. e.g. I add Reaktor (TRK-01) to a song that's already registering 40% CPU, and would presumably be spreading the load across both cores, and it goes up to 65%. Then I add a second instance and it only goes up to 70%. I put that down to the fact that once the Reaktor engine is up and running in RAM, it can handle multiple instances efficiently, such that the CPU increase is in line with the increased polyphony that goes with it.I don't care much for anecdotes, so I spent a lot of time just actually doing tests, and this is all very testable. So don't take my word for it, test it yourself, it's not hard, and it dispels magical thinking.
NOVAkILL : Asus RoG Flow Z13, Core i9, 16GB RAM, Win11 | EVO 16 | Studio One | bx_oberhausen, GR-8, JP6K, Union, Hexeract, Olga, TRK-01, SEM, BA-1, Thorn, Prestige, Spire, Legend-HZ, ANA-2, VG Iron 2 | Uno Pro, Rocket.
-
machinesworking machinesworking https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=8505
- KVRAF
- 6126 posts since 15 Aug, 2003 from seattle
No, I wouldn't say, but if you want to believe you're saving CPU by using the same synth by all means go ahead. Every test I've ever done has looked the same as what you're describing at lower CPU levels, but does not hold up when pushing the CPU, it points to what I'm saying to be true, the way CPU is read out is not a linear formation, you don't get a true 200% CPU use out of a two core machine, and the percentages are measures of how close a CPU is to failure, not a percentage of a possible 200%.BONES wrote: ↑Thu Mar 21, 2019 11:40 pmFor a start, I don't use a "DAW", I use a virtual studio, and checking Orion's CPU meter with Task Manager, it actually reads high. The fact that combining the patterns into a single instance only reduces CPU use by 5% compared to two instances would tend to suggest it has little to do with multi-core performance, wouldn't you say?machinesworking wrote: ↑Thu Mar 21, 2019 9:09 pmAlmost every DAW I know of reads the cores ... uhm.. weirdly, i.e. it takes the highest core as the percentage, not the percentage of a combination of your cores. So a single instance of a plug in ramps to 45%, which if you have a four core machine isn't a real number, since it's 45% of one core.
Unless you do tests to failure, it's just anecdotal, that's it, period, end of story. You can think whatever you want to think, but like I said in my experience CPU meters are odd things, they're basically estimates.I don't need to "test" it, I have seen it in production situations. e.g. I add Reaktor (TRK-01) to a song that's already registering 40% CPU, and would presumably be spreading the load across both cores, and it goes up to 65%. Then I add a second instance and it only goes up to 70%. I put that down to the fact that once the Reaktor engine is up and running in RAM, it can handle multiple instances efficiently, such that the CPU increase is in line with the increased polyphony that goes with it.I don't care much for anecdotes, so I spent a lot of time just actually doing tests, and this is all very testable. So don't take my word for it, test it yourself, it's not hard, and it dispels magical thinking.
Again, I hate anecdotal. Taking your idea to task here:
A typical way to do a to failure test is to use a generic arpeggio run and copy the exact same instance.
If what you say is true then this should absolutely be the most efficient way, but it's not.
In a 12 core machine running Diva I get 12 instances with a single note run before the audio starts crackling, I can push 13 but barely. On that same machine adding in a ten note chord run of the same 1/4 note length same speed etc. for each instance I can run five instances of Diva.
That's an increase in polyphony of 37 notes.
Even running the exact same patch in Diva, there's no improvement at all in running multiple instances of the same plug in, in fact there's a significant performance hit.
To add to this, at one point during this testing when I attempted to run ten instances of Diva with a ten note chord run the Activity monitor read 1850% a not even possible number on a 12 core machine. Again, CPU meters are estimates of CPU use, not an exact measure by any means.
{edited to correct the math... }
- GRRRRRRR!
- 15844 posts since 14 Jun, 2001 from Somewhere else, on principle
What an idiotic comment. It's not about what I believe, it is about the ability to use multiple instances of an instrument and the reality is that using two instances of the same instrument absolutely does not take up double the CPU overhead. That's all that matters - whether I can keep adding instances or not. I don't know how that's not completely obvious to you.machinesworking wrote: ↑Fri Mar 22, 2019 2:21 amNo, I wouldn't say, but if you want to believe you're saving CPU by using the same synth by all means go ahead.
Therein lies the difference - I actually use my equipment for its intended purpose, not for running tests.Every test I've ever done has looked the same as what you're describing
Pal, I use a machine with a Core i5 processor, it gets tested to failure pretty much every day. That's why knowing that I can add two instances of one synth more efficiently than one instance of two different synths is vital information for me, not something of passing interest.Unless you do tests to failure, it's just anecdotal, that's it, period, end of story.
Probably a Mac thing, poorly optimised ports from Windows. I dunno why anyone puts up with that krap.Even running the exact same patch in Diva, there's no improvement at all in running multiple instances of the same plug in, in fact there's a significant performance hit.
NOVAkILL : Asus RoG Flow Z13, Core i9, 16GB RAM, Win11 | EVO 16 | Studio One | bx_oberhausen, GR-8, JP6K, Union, Hexeract, Olga, TRK-01, SEM, BA-1, Thorn, Prestige, Spire, Legend-HZ, ANA-2, VG Iron 2 | Uno Pro, Rocket.
-
machinesworking machinesworking https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=8505
- KVRAF
- 6126 posts since 15 Aug, 2003 from seattle
You're literally using the definition of the argument from ignorance.BONES wrote: ↑Fri Mar 22, 2019 5:35 amWhat an idiotic comment. It's not about what I believe, it is about the ability to use multiple instances of an instrument and the reality is that using two instances of the same instrument absolutely does not take up double the CPU overhead. That's all that matters - whether I can keep adding instances or not. I don't know how that's not completely obvious to you.machinesworking wrote: ↑Fri Mar 22, 2019 2:21 amNo, I wouldn't say, but if you want to believe you're saving CPU by using the same synth by all means go ahead.Therein lies the difference - I actually use my equipment for its intended purpose, not for running tests.Every test I've ever done has looked the same as what you're describingPal, I use a machine with a Core i5 processor, it gets tested to failure pretty much every day. That's why knowing that I can add two instances of one synth more efficiently than one instance of two different synths is vital information for me, not something of passing interest.Unless you do tests to failure, it's just anecdotal, that's it, period, end of story.Probably a Mac thing, poorly optimised ports from Windows. I dunno why anyone puts up with that krap.Even running the exact same patch in Diva, there's no improvement at all in running multiple instances of the same plug in, in fact there's a significant performance hit.
I did the tests because I want to know my machine, you're using faulty logic and coming to half baked conclusions, the same sort of shit you give people crap about. You're literally stuck on one conclusion, and it's dead simple to test if it's true or not. I've outlined it, I think any fool that's bored enough to sort through this thread can tell who put in the work to come to a conclusion and who is acting a fool.
BTW, U-He is the last company in the world that would poorly port a windows plug in to OSX, he started as an AU only developer.
- Beware the Quoth
- 33109 posts since 4 Sep, 2001 from R'lyeh Oceanic Amusement Park and Funfair
Dont bring facts to an argument with BONES, he's literally incapable of accepting anything that doesnt reinforce confirmation bias.machinesworking wrote: ↑Fri Mar 22, 2019 6:25 amBTW, U-He is the last company in the world that would poorly port a windows plug in to OSX, he started as an AU only developer.
my other modular synth is a bugbrand
-
- addled muppet weed
- 105553 posts since 26 Jan, 2003 from through the looking glass
-
el-bo (formerly ebow) el-bo (formerly ebow) https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=208007
- KVRAF
- 16368 posts since 24 May, 2009 from A galaxy, far far away
-
el-bo (formerly ebow) el-bo (formerly ebow) https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=208007
- KVRAF
- 16368 posts since 24 May, 2009 from A galaxy, far far away
-
machinesworking machinesworking https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=8505
- KVRAF
- 6126 posts since 15 Aug, 2003 from seattle
-
grandmasterbird grandmasterbird https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=134988
- KVRist
- 363 posts since 7 Jan, 2007
Is it sick to have more than 20 synths? No.
Is it pointless?
Probably
Is it pointless?
Probably