yeah i know i got his point. i was really trying to provoke a different explanation or further exploration into this mysteryfluffy_little_something wrote:solipsvs wrote:wait maybe jo intends to mask whats actually happening in the core of the audio engine by being ambiguous when providing an explanation or a working solution to your spec. ok in this case perhaps it is a decent excuse for inaccurate info. jo doesnt wanna be hacked and my guess is this is a window into how mulab operates. much like uh-e masked processor usage in the latest pro-one demo thingy recently. someone might be able to tell something about the algorithms used by closely watching the processor load. am i way off here?
Nah, there is nothing for Jo to deliberately hide. If anything there is some unknown issue...
U-he only did that with their synth in order to keep people from thinking that the most demanding filter model must also sound the best.
Mulab's CPU efficiency ...
- KVRist
- 329 posts since 6 Mar, 2015
ACKCHYUALLY
-
- KVRist
- 179 posts since 26 Feb, 2007
Easy guys. Just checking in here quickly, because when I was using Mulab I noticed a CPU inefficiency as well.
*HOWEVER*: Bear in mind that on OSX Mulab is still 32bit.
NI Polyplex for example uses nearly 15% more CPU in Mulab than in Logic Pro X, which is 64bit.
(Also there are sometimes differences between VST and AU on load in general.
Sometimes bigger, sometimes smaller, sometimes the same.)
Now the interesting part here is that when running Polyplex in Energy XT (32bit) its *the same thing*.
Also high load. When running in Tracktion 32bit also. Changes in 64bit tracktion, but also different changes when you use the AU or VST version of it.
So in my experience this has nothing to do with MuLab, but some plugins run way better on a 64bit host.
Some, not all of them. So I think honestly there is really not much that can be done about that.
Also strong point here:
Some time ago I tested Polyplex with various DAWS at that time, because its pretty heavy on CPU and
I wanted to know what the hell is going on with some programs.
And what was more interesting is the result I got on Mulab:
Even with the CPU load being much higher on Mulab, Mulab was still pretty smoothly running with nearly maxxed out CPU cores.
Thats highly unusual and speaks for good multicore utilization.
In some other DAWs you start to get dropouts and crackles sometimes already around 60-70% on load.
You can push Mulab pretty far, which gives you an extra "upside buffer" so to speak and it is still working smoothly unlike some other DAWs where the GUI tends to slow down as well or the DAW gets instable, and is prone to crashes, which I particularly hate with some other DAWs. So I think the Mulab code is f****n rocksolid.
(On a sidenote: The best multicore management I have seen by far has Renoise. By miles. Kicks the sh** even out of Reaper and Logic. Thats also one of my reasons why I am back in the tracker camp since quite a while now because they are by nature superefficient and lean and slick in many ways and I can work so friggin' fast with theese, but thats going offtopic and is not the point I am trying to make here. )
Ah yeah, by the way, someone asked for a realtime monitoring/measuring per core utility for OSX.
This is what I use: https://www.ragingmenace.com/software/menumeters/
Its free, highly configurable and very consistent.
Be nice to each other..
Best,
tL.
*HOWEVER*: Bear in mind that on OSX Mulab is still 32bit.
NI Polyplex for example uses nearly 15% more CPU in Mulab than in Logic Pro X, which is 64bit.
(Also there are sometimes differences between VST and AU on load in general.
Sometimes bigger, sometimes smaller, sometimes the same.)
Now the interesting part here is that when running Polyplex in Energy XT (32bit) its *the same thing*.
Also high load. When running in Tracktion 32bit also. Changes in 64bit tracktion, but also different changes when you use the AU or VST version of it.
So in my experience this has nothing to do with MuLab, but some plugins run way better on a 64bit host.
Some, not all of them. So I think honestly there is really not much that can be done about that.
Also strong point here:
Some time ago I tested Polyplex with various DAWS at that time, because its pretty heavy on CPU and
I wanted to know what the hell is going on with some programs.
And what was more interesting is the result I got on Mulab:
Even with the CPU load being much higher on Mulab, Mulab was still pretty smoothly running with nearly maxxed out CPU cores.
Thats highly unusual and speaks for good multicore utilization.
In some other DAWs you start to get dropouts and crackles sometimes already around 60-70% on load.
You can push Mulab pretty far, which gives you an extra "upside buffer" so to speak and it is still working smoothly unlike some other DAWs where the GUI tends to slow down as well or the DAW gets instable, and is prone to crashes, which I particularly hate with some other DAWs. So I think the Mulab code is f****n rocksolid.
(On a sidenote: The best multicore management I have seen by far has Renoise. By miles. Kicks the sh** even out of Reaper and Logic. Thats also one of my reasons why I am back in the tracker camp since quite a while now because they are by nature superefficient and lean and slick in many ways and I can work so friggin' fast with theese, but thats going offtopic and is not the point I am trying to make here. )
Ah yeah, by the way, someone asked for a realtime monitoring/measuring per core utility for OSX.
This is what I use: https://www.ragingmenace.com/software/menumeters/
Its free, highly configurable and very consistent.
Be nice to each other..
Best,
tL.
Last edited by LYTZ on Fri May 27, 2016 6:34 pm, edited 1 time in total.
Professional.
-
fluffy_little_something fluffy_little_something https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=281847
- Banned
- Topic Starter
- 12880 posts since 5 Jun, 2012
Yes, it seems that Mulab handles loads well, judging from what I hear, or don't hear for that matter
Still, I would like to have some additional indicator for real CPU load, not just my ears.
Is the CPU meter in Mulab linear? I get the impression that it easily goes to 40 to 60%, but after that the increase is slower, until finally the first red bars and audio issues appear.
Still, I would like to have some additional indicator for real CPU load, not just my ears.
Is the CPU meter in Mulab linear? I get the impression that it easily goes to 40 to 60%, but after that the increase is slower, until finally the first red bars and audio issues appear.
-
- KVRist
- 179 posts since 26 Feb, 2007
lookie here. An in-App measuring tool is not available AFAIK, so this of course is systemwide but extremely percise.:fluffy_little_something wrote: Still, I would like to have some additional indicator for real CPU load, not just my ears.
LYTZ wrote: Ah yeah, by the way, someone asked for a realtime monitoring/measuring per core utility for OSX.
This is what I use: https://www.ragingmenace.com/software/menumeters/
Its free, highly configurable and very consistent.
You do not have the required permissions to view the files attached to this post.
Professional.
-
fluffy_little_something fluffy_little_something https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=281847
- Banned
- Topic Starter
- 12880 posts since 5 Jun, 2012
W7 
- KVRAF
- 8582 posts since 29 Sep, 2010 from Maui
https://www.rainmeter.net/ is what I use under windows.
I'm sure you could find a skin to monitor whatever you want.
I've been using it for years. Mostly because Conky is not available
for windows. I switched to it because the old windows gadgets used
75 MB of ram to run while rainmeter used 10 MB. And was a lot more
interesting on top of that.
4 versions later it now uses 11 MB to run.
-Cheers
I'm sure you could find a skin to monitor whatever you want.
I've been using it for years. Mostly because Conky is not available
for windows. I switched to it because the old windows gadgets used
75 MB of ram to run while rainmeter used 10 MB. And was a lot more
interesting on top of that.
4 versions later it now uses 11 MB to run.
-Cheers
- KVRAF
- 2473 posts since 25 Sep, 2014 from Specific Northwest
I was just throwing my experience with CPU efficiency into the ring for others who may be trying to test for themselves.fluffy_little_something wrote:Yes, I am aware it is not an efficiency issue as such, we have already settled that. The problem is simply that I have no idea how much CPU power is actually being used under Mulab because unfortunately it handles these things differently compared to other DAW's, leading to bogus values on the Windows task manager core meters.
I started on Logic 5 with a PowerBook G4 550Mhz. I now have a MacBook Air M1 and it's ~165x faster! So, why is my music not proportionally better? 
- KVRAF
- 2473 posts since 25 Sep, 2014 from Specific Northwest
I live and die by these meters! My music rig is frozen at 10.9 so I can use ALL my VSTs, new and old.LYTZ wrote: Ah yeah, by the way, someone asked for a realtime monitoring/measuring per core utility for OSX.
This is what I use: https://www.ragingmenace.com/software/menumeters/
Its free, highly configurable and very consistent.
Be nice to each other..
Best,
tL.
Wish Menu Meters worked with 10.11, though... There's nothing else like it that I have found.
I started on Logic 5 with a PowerBook G4 550Mhz. I now have a MacBook Air M1 and it's ~165x faster! So, why is my music not proportionally better? 
- KVRAF
- 13865 posts since 24 Jun, 2008 from Europe
Note that, as explained already, MuLab's CPU meter and the task manager measure different things and so the values are not be compared.fluffy_little_something wrote:Tried one of those demo tunes.
Mulab meter values:
1 core: 27% avg, 45% max
6 cores: 15% avg, 23% max
Task Manager values:
1 core: CPU 8%, about 20% on one core, the rest 0 to 5%
6 cores: CPU 20%, 10-20% on all six cores
About the impact of using multiple cores on your CPU usage: It's depending on the project. For example Dream of the piano demo project has only 4 parallel racks and 2 effect racks and so it will maximum use 4 cores at the same time because the effect racks can't be processed before the intrument racks are fully processed. So in fact i should not have suggested to test things with one of the demo projects as they're all too small for a 6 core machine. The bigger the project ie. the more parallel racks, the more you'll take advantage of a multi-core setup.
-
fluffy_little_something fluffy_little_something https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=281847
- Banned
- Topic Starter
- 12880 posts since 5 Jun, 2012
Sorry, but I don't think you really understand the issue at hand, judging from your comments. I couldn't care less what Mulab's meter measures and says, the Task Manager is what matters to me as the not so proud owner of a weak CPU. And in that TM Mulab leads to wrong values for individual cores and the global CPU. The scaling based on the number of cores used is screwed up somehow. When I do similar tests in Mixcraft, i.e. using 1 vs 6 cores, the results shown in the TM are very different and they make sense to me.
And regarding your comment on project size, I am a minimalist, I don't use zillions of tracks.
And regarding your comment on project size, I am a minimalist, I don't use zillions of tracks.
- KVRAF
- 13865 posts since 24 Jun, 2008 from Europe
The values in TM are not 'wrong' nor is MuLab's multi-core engine 'wrong'.
This already has been explained in this topic.
And also on http://www.mutools.com/info/docs/mulab/audio-setup.html
This already has been explained in this topic.
And also on http://www.mutools.com/info/docs/mulab/audio-setup.html
-
- KVRist
- 179 posts since 26 Feb, 2007
MenuMeters El Capitan Port by some Japanese guy..syntonica wrote:
I live and die by these meters! My music rig is frozen at 10.9 so I can use ALL my VSTs, new and old.
Wish Menu Meters worked with 10.11, though... There's nothing else like it that I have found.
http://member.ipmu.jp/yuji.tachikawa/Me ... ElCapitan/
Professional.
-
fluffy_little_something fluffy_little_something https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=281847
- Banned
- Topic Starter
- 12880 posts since 5 Jun, 2012
I just think it makes more sense the way it is done in Mixcraft. There TM shows how much CPU is actually being used, there are no idle threads running and showing fake CPU loads on cores.mutools wrote:The values in TM are not 'wrong' nor is MuLab's multi-core engine 'wrong'.
This already has been explained in this topic.
And also on http://www.mutools.com/info/docs/mulab/audio-setup.html
As a user I don't want to be bugged with such nerdy decisions as the number of threads to be used. The DAW should pick the ideal settings by itself, based on the current project specs and needs.
Have you renamed that parameter by the way? I think it was number of cores instead of threads previously...
- KVRAF
- 13865 posts since 24 Jun, 2008 from Europe
No, not that i can remember. Anyway, "threads" is technically the right term as the user cannot choose the number of cores on a machine. (well, besides making such choice upon purchasing a computer)fluffy_little_something wrote:Have you renamed that parameter by the way? I think it was number of cores instead of threads previously...
