Mulab's CPU efficiency ...

[samardac]

I made quick test too, without any DAW tuning just as it is from the box, because a lot of users never touch all that stuff.
I loaded PG-8X (https://www.facebook.com/MLVST) free VST and play one chord in Reaper and in MuTools.
In Reaper it shows CPU usage about 0,7%
In MuTools it shows CPU usage about 3,4%
About 5 times difference...

[MuTools]

Which display did you read out?
Note that MuLab's own cpu display shows the percentage of cpu consumption compared to what can be done in 1 audio block.
Windows' cpu display is measuring overall and thus will show lower values. I don't know what Reaper's display measures. There is a real chance that the 0.7 and 3.4 cannot be compared because they are simply different types of values.

[samardac]

I used standard Task Manager from Windows 10 for both apps.

[MuTools]

When cpu measurement is done 'overall' then block size won't influence the measured value a lot. But when cpu measurement is done 'per audio block' (as MuLab does, cause that's the most relevant value) then if you measure for say blocksize 512 samples and then measure the same project for blocksize 256 samples the measured value will at least be double. When cpu measurement is done 'overall' the measured value in the second case may only be a little bit higher, maybe even not noticable. It are really 2 totally different types of measurement.

[Zexila]

The way I did testing before (not for MuLab unfortunately) is to set same buffer settings and do same stress test and see which one will have highest track count before we hit the limit, if there is difference, one will hit first with lower number of instances, easy as that, nothing scientific about it, we are making music and number of instances we can run is all that matters to us, right, so do stress test and see what really gives you more.

[MuTools]

I used standard Task Manager from Windows 10 for both apps.

Well i'm sure that there must be other differences then. I did a quick double-check test myself yesterday compared M7 to Ext 2.7 and FLS 12.0 playing a 3 note chord to the same VSTi and it was all quite equal. We all get very different results it seems. I know Reaper is very efficient indeed may be one of the bests in that regard, partially because it uses some trick that it precalculates things which has a penalty on realtime aspects, as you can hear when doing certain things. MuLab is simple and straightforward: Everything is calculated in realtime, no delays.

[fluffy_little_something]

Sorry for dumb&OT post, but how do one even change buffer size for on-board card on OS X in MuLab?

I have the same problem with that explanation given before. I set every parameter I have access to the same way. I set the sampling rate to 96, and in the ASIO control panel of my sound card 16bit and a latency of 8ms, which automatically results in a buffer size of 768 in Mixcraft, a value I can't change, except if I pick another latency in ASIO.
I don't know what buffer size Mulab determines based on my ASIO settings.

Either way, no matter what the reasons for the differences are, somehow Mixcraft and maybe Reaper etc. as well seem to do things differently, more efficiently than Mulab.

Does Mulab use higher accuracy inside? I think I remember reading that some DAW developers have claimed such things in the past.

[fluffy_little_something]

I used standard Task Manager from Windows 10 for both apps.

Well i'm sure that there must be other differences then. I did a quick double-check test myself yesterday compared M7 to Ext 2.7 and FLS 12.0 playing a 3 note chord to the same VSTi and it was all quite equal. We all get very different results it seems. I know Reaper is very efficient indeed may be one of the bests in that regard, partially because it uses some trick that it precalculates things which has a penalty on realtime aspects, as you can hear when doing certain things. MuLab is simple and straightforward: Everything is calculated in realtime, no delays.

Which VST did you use? It should be something demanding so that any differences become more obvious. When you use a very economic VST to begin with, the differences will be so small that you can't even rule out that they result from some Windows processes or whatever. After all, the Windows resource manager values also slightly fluctuate over time...

[MuTools]

Sorry for dumb&OT post, but how do one even change buffer size for on-board card on OS X in MuLab?

In the audio device's preference panel.

[Zexila]

Sorry for dumb&OT post, but how do one even change buffer size for on-board card on OS X in MuLab?

In the audio device's preference panel.

One in OS X and one in MuLab...

[samardac]

I used standard Task Manager from Windows 10 for both apps.

Well i'm sure that there must be other differences then. I did a quick double-check test myself yesterday compared M7 to Ext 2.7 and FLS 12.0 playing a 3 note chord to the same VSTi and it was all quite equal. We all get very different results it seems. I know Reaper is very efficient indeed may be one of the bests in that regard, partially because it uses some trick that it precalculates things which has a penalty on realtime aspects, as you can hear when doing certain things. MuLab is simple and straightforward: Everything is calculated in realtime, no delays.

Which VST did you use? It should be something demanding so that any differences become more obvious. When you use a very economic VST to begin with, the differences will be so small that you can't even rule out that they result from some Windows processes or whatever. After all, the Windows resource manager values also slightly fluctuate over time...

I wrote in my post, https://www.facebook.com/MLVST

[fluffy_little_something]

Here's my spec. First, with no DAW loaded:

Code: Select all

	AMD FX(tm)-8350 Eight-Core Processor 

	Maximum speed:	4.10 GHz
	Sockets:	1
	Cores:	4
	Logical processors:	8
	Virtualisation:	Enabled
	L1 cache:	384 KB
	L2 cache:	8.0 MB
	L3 cache:	8.0 MB

	Utilisation	2%
	Speed	2.13 GHz
	Up time	0:00:33:32
	Processes	66
	Threads	1002
	Handles	25995

Then with Reaper loaded, with two tracks (input monitoring, etc, enabled) each with mda ePiano. (Edit: and I went back and checked it made no difference if only one was monitoring input, too.)

Code: Select all

	Utilisation	4%

Next, MuLab 6 with mda ePiano.

Code: Select all

	Utilisation	4%

Finally, MuLab 7.0.28 with mda ePiano.

Code: Select all

	Utilisation	4%

OT, but I have seen that you use the AMD FX 8350 processor, how does that work for really demanding plugins? I am asking because the 8370 is now on sale here and includes the new silent stock cooler, so I am thinking of getting that

[fluffy_little_something]

I used standard Task Manager from Windows 10 for both apps.

Well i'm sure that there must be other differences then. I did a quick double-check test myself yesterday compared M7 to Ext 2.7 and FLS 12.0 playing a 3 note chord to the same VSTi and it was all quite equal. We all get very different results it seems. I know Reaper is very efficient indeed may be one of the bests in that regard, partially because it uses some trick that it precalculates things which has a penalty on realtime aspects, as you can hear when doing certain things. MuLab is simple and straightforward: Everything is calculated in realtime, no delays.

Which VST did you use? It should be something demanding so that any differences become more obvious. When you use a very economic VST to begin with, the differences will be so small that you can't even rule out that they result from some Windows processes or whatever. After all, the Windows resource manager values also slightly fluctuate over time...

I wrote in my post, https://www.facebook.com/MLVST

I was asking MuTools

[MuTools]

Sorry for dumb&OT post, but how do one even change buffer size for on-board card on OS X in MuLab?

I have the same problem with that explanation given before.

Youcan change buffer size in the ASIO control panel.

I don't know what buffer size Mulab determines based on my ASIO settings.

MuLab uses the ASIO buffer size set in the ASIO control panel. Simple, straight.

Either way, no matter what the reasons for the differences are

We're still discussing how to measure, so we can't conclude about differences yet.

somehow Mixcraft and maybe Reaper etc. as well seem to do things differently, more efficiently than Mulab.

"Seems" is the right word indeed, imho.

Hey i'm not saying MuLab will be the topmost efficient of all DAWs, may be not. I do care about efficiency but it's not the topmost priority for me. Sound quality, stability and creative flow are top prios. Efficiency is important too and i invest important r&d time in cpu optimizations. What i don't believe is the figures some of you guys are posting here which tell that MuLab would use double even triple the cpu Reaper or Mixcraft are using. That's absurd, that can't be possible. There must be some reasons that explain those differences. I already replied several things that could explain differences.

Does Mulab use higher accuracy inside?

Speaking about MUX Modular: Yes i always choose for no compromise wrt sound quality!! I'd rather spent more cpu cycles, but sound quality is top prio.
But that's not relevant here as we're talking about plugin in a VSTi. And that's a black box for the DAW.

[samardac]

I was asking MuTools

Aaa, ok