Under-the-hood optimizations in Live 10?

[Grill Pheiss]

9 crackes under a certain amount of stress/load when sessions get
heavy, it just does.

Has anyone noticed any improvement with 10 in that department?

[Tarekith]

CPU use is almost identical in Live 10 as it was in Live 9.

[Stromkraft]

CPU use is almost identical in Live 10 as it was in Live 9.

Any clues to why this is not a priority for development?

[DrTone]

As posted elsewhere in this forum, I have a different experience:

Basic Code seems to be more efficient. Eg: I was working on a song in L9 (MacBook Pro, End 2013, 2,3 GHz i7, 16GB) which took something like 40-44%CPU-load. I loads longer into L10beta but once its loaded, it hits only @ 28-30%. That's roughly 30% increase in efficiency! Quite impressive.

That Set contains about every type of track, audio, vst-instruments, groups, ;MfL-Devices etc. and quite a few plug-ins - total of 38tracks. So, I don't really know, where exactly the improvements happen (and don't have the time to research it), but they seem to be there. (beta-version is 10.0b144).

[Tarekith]

As posted elsewhere in this forum, I have a different experience:

Basic Code seems to be more efficient. Eg: I was working on a song in L9 (MacBook Pro, End 2013, 2,3 GHz i7, 16GB) which took something like 40-44%CPU-load. I loads longer into L10beta but once its loaded, it hits only @ 28-30%. That's roughly 30% increase in efficiency! Quite impressive.

That Set contains about every type of track, audio, vst-instruments, groups, ;MfL-Devices etc. and quite a few plug-ins - total of 38tracks. So, I don't really know, where exactly the improvements happen (and don't have the time to research it), but they seem to be there. (beta-version is 10.0b144).

Interesting, this is the first I’ve heard of anyone seeing a difference compare to Live 9. You’re sure the latency was set to the same with both versions?

[Angstrom]

From my testing there's no difference in CPU utilisation (at least using native devices and M4L synths/effects) - the differences are well within error margin.

I agree with this. So far the differences are negligible, and I've not seen any notable improvements.

Its certainly possible that they improved resource usage, but I'm not seeing it, for whatever reason (Eg: they made improvements but then used those resource gains elsewhere, so the net result was -0%)

[DrTone]

Interesting, this is the first I’ve heard of anyone seeing a difference compare to Live 9. You’re sure the latency was set to the same with both versions?

Ha! You caught me. True. I was mixing only in L10, so the buffer was still @256samples. If I set it to 64samples I am back @42%. Doh! Okay, no difference here, too.

[Grill Pheiss]

I'm more conerned about the behaviour of the engine

example : a heavy session exhibiting audible crackels because of cpu load (I mean lives crackels, even with the cpu hovering at 69%...)
same session loaded in Live 10... same crackles?

[Grill Pheiss]

Well I'll answer my own question since I've been playing around with 10 for an hour with a 9-created project

Slightly worse!

Slighlty more crackles

Slightly more CPU load, about 7% for this heavy project I'm playing around with

So there we have it

No optimization in that department so far.