Mulab's CPU efficiency

Official support for: mutools.com
chk071
KVRAF
17230 posts since 11 Apr, 2010 from Germany

Post Mon Jan 01, 2018 5:28 pm

ASIO4ALL is not nearly as efficient as a decent ASIO driver from a audio interface. I could use 448 samples, anything below that always caused crackles here, regardless of the sound card. And, unless you play in a piano piece, which needs extremely precise timing, it's kind of pointless to go that low also, because you will tax your CPU more, without much gain, because, when you just play a few cords or notes on a synth, 10 ms or higher will be more than sufficient.

SparkySpark
KVRian
1305 posts since 30 Aug, 2004 from Skövde, Sweden

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 4:22 am

Just a clarification because of the name of this thread: MuLab and MUX really impress me being so easy on the CPU.

I think Jo has done an amazing job on CPU efficiency.
:clap: :clap: :clap:
Making music is a nine-to-five job:
From 9 PM to 5 AM.
Go MuLab!

User avatar
EvilDragon
KVRAF
17062 posts since 7 Jan, 2009 from Croatia

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 4:26 am

Does MuLAB ever do some sort of render-ahead (anticipative processing)? If not, that's the main difference between it and Reaper. Anticipative processing gives Reaper the edge in being more CPU efficient than other DAWs.

User avatar
fluffy_little_something
KVRAF
12230 posts since 5 Jun, 2012 from Portugal

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 4:30 am

Michael L wrote:There is no way we can compare your MuLab and Reaper preferences, but you might want to review them if you are thinking about buying a high-cpu synth, and the u-he sales department may also help.
I just built a new computer, that will have to last at least until 2025.
Again, you people have a weird logic. You don't seem to get the problem:
I have one and the same system of hardware and software using the same settings, the only thing that is different is the DAW, namely Mulab vs Reaper. Only the latter performs without glitches when playing the same chords of the same demanding synth.
So, obviously the issue is the DAW.
I don't know about other DAW's, they are too big to download and install (I read one should not install big files on a SSD unless it is really necessary as SSD's have a certain limit in terms of write cycles).

User avatar
fluffy_little_something
KVRAF
12230 posts since 5 Jun, 2012 from Portugal

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 4:33 am

chk071 wrote:ASIO4ALL is not nearly as efficient as a decent ASIO driver from a audio interface. I could use 448 samples, anything below that always caused crackles here, regardless of the sound card. And, unless you play in a piano piece, which needs extremely precise timing, it's kind of pointless to go that low also, because you will tax your CPU more, without much gain, because, when you just play a few cords or notes on a synth, 10 ms or higher will be more than sufficient.
There are no native ASIO drivers for my audio solution, and ASIO4ALL is a fine piece of software, it works flawlessly with Reaper.
Even on my old computer it worked better than the native ASIO driver of my quality Asus sound card.

128 samples causes no problems whatsoever with Reaper, so obviously it is not a hardware or driver problem.

User avatar
fluffy_little_something
KVRAF
12230 posts since 5 Jun, 2012 from Portugal

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 4:36 am

There is another difference:
When I load Sylenth1, it has that ugly default patch. Be that as it may, that patch distorts in Mulab, but sounds clean in Reaper. Seems the two DAW's handle dynamics differently.

sfxsound3
Banned
453 posts since 30 Mar, 2016

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 4:39 am

On the contrary, your buffer is too low and ASIO4ALL can't cope, THIS is causing the dropouts. Reaper and DP have anticipative, or render ahead processing, which is why they are more effective by A LOT. You could try DP too and see for yourself. MuLab obviously doesn't have such things - you can't blame it, though, most apps don't have 'em.

chk071
KVRAF
17230 posts since 11 Apr, 2010 from Germany

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 4:45 am

Wie gesagt, so eine Latenz ist eigentlich völlig unnötig. Weil ASIO4ALL mit allen WDM-Karten kompatibel ist, ist das einfach eine Kompromisslösung, und da kann man nicht eine Leistung wie von speziell für die Karte geschriebenen ASIO-Treibern erwarten. Mit 128 Samples wirst du spätestens nach ein paar Spuren in deinem Projekt an die Leistung deiner CPU stoßen. Höhere Latenzzeiten geben dir immer noch genügend Timing zum Spielen, 3-8 ms ist was für Profis, die ein professionelles Pianostück einspielen wollen. Eine Verzögerung von 10-20 ms merke ich hier überhaupt nicht, und es belastet auch nicht das System wie wild. Ich würde wirklich versuchen, die Blockgröße raufzusetzen, an deiner Stelle.

mutools
KVRAF
9561 posts since 24 Jun, 2008 from Europe

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 5:55 am

Preferrably english please. Thx.

Here is google's translation:

"As I said, such a latency is actually completely unnecessary. Because ASIO4ALL is compatible with all WDM cards, it's simply a compromise solution, and you can not expect performance like ASIO drivers written specifically for the card. With 128 samples you will come across the performance of your CPU at the latest after a few tracks in your project. Higher latencies still give you enough timing to play, 3-8ms is for pros who want to play a professional piano piece. A delay of 10-20 ms I do not notice here at all, and it also does not burden the system like crazy. I would really try to raise the block size, in your place."

chk071
KVRAF
17230 posts since 11 Apr, 2010 from Germany

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 6:01 am

Sorry about that. The translation is spot on. :)

User avatar
EvilDragon
KVRAF
17062 posts since 7 Jan, 2009 from Croatia

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 6:14 am

The thing is that the fluffy doesn't have a "pro" or even a semi-pro audio interface, he has a gamer-home user Asus card. Those aren't usually known for their awesome ASIO drivers, because ASIO is not a priority for such devices.

Fluffy, you might wanna consider to upgrade your audio interface at some point, and get one that is not intended for gamers :)

User avatar
VariKusBrainZ
KVRAF
8379 posts since 16 Dec, 2002

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 6:19 am

sfxsound3 wrote:. Reaper and DP have anticipative, or render ahead processing, which is why they are more effective by A LOT. You could try DP too and see for yourself. MuLab obviously doesn't have such things - you can't blame it, though, most apps don't have 'em.
Surely this only works for notes programmed into the piano roll, no Daw can predict what will be played in realtime via a controller lol

Fluffy - can you confim if youre playing back pre programmed chords from the piano roll or via a controller in realtime and can you confirm both hosts are definately using the same Asio settings.

Is there any difference if you have the synth Gui visible or not?

User avatar
EvilDragon
KVRAF
17062 posts since 7 Jan, 2009 from Croatia

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 6:21 am

It doesn't matter if notes are programmed or not. It's a matter of taking a chunk of data in advance and rendering it ahead of time. So it will work with audio data and FX plugins on the track, for example, not just notes.

User avatar
toonertik
KVRist
375 posts since 15 Feb, 2017

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 6:42 am

EvilDragon wrote:The thing is that the fluffy doesn't have a "pro" or even a semi-pro audio interface, he has a gamer-home user Asus card. Those aren't usually known for their awesome ASIO drivers, because ASIO is not a priority for such devices.

Fluffy, you might wanna consider to upgrade your audio interface at some point, and get one that is not intended for gamers :)
Yes... awesome drivers needed...
BUT here is an interesting observation.
I also have a new build AMD computer. I have not yet installed my old EMU_1212M card (still on old DAW)
I am using the RealTek (or is it ReaTek.. which ever, it sounds like something from the Reaper folder ;) ) in built audio stuff with Asio4All!!!
I recently participated in the Thorn_OSC105, thorn is quiet CPU intensive.
Another participant posted his Mulab entry file. I downloaded it and grabbed the free Mulab.
This song goes up to 101% and .. well you will know.
However, I also have Samplitude, which comes with it's own magixLowLatency drivers.
If I set Mulab to use these drivers, it peaks at about 76%
The interesting thing is that the magix drivers also load up Asio4All.
I guess we can not expect Asio4All to be as effecient.. It is more like a "get out of jail free card" and one should really have the proper gear to do the job>> as ED mentioned!
Happy New Year all.

User avatar
VariKusBrainZ
KVRAF
8379 posts since 16 Dec, 2002

Re: Mulab's CPU efficiency

Post Tue Jan 02, 2018 6:47 am

EvilDragon wrote:It doesn't matter if notes are programmed or not. It's a matter of taking a chunk of data in advance and rendering it ahead of time. So it will work with audio data and FX plugins on the track, for example, not just notes.
How can a note I havent yet played on my keyboard be pre rendered?

Return to “MUTOOLS”