[5184] L8 uses 30% more CPU than L7

[Tone Deft]

is borked.

[Tone Deft]

[

what the hell are you talking about? can I be any more clear that I'm not a coder?

goddamn interwebs is broken!

[evon]

Hey Tone I respect you when you talk about Live and music stuff..but the ranting thing is not cool..dont beleive I am going to be scared about your 13,000+.

[Tone Deft]

I don't know if it's more fun to troll this guy into another tissy fit or just watch him run amuck on his own.

3 million sperm and he won??? wow.

[evon]

Anyway....my 2cents...the CPU thing is an issue and I beleive this would be one of the first things anyone beta testing would have noticed. Its like the speedometer on a car, every time one adds a track, an effect etc instintively you check the meter to see how its responding.
I think what Tone might be trying to get accross is that the Abes must have been aware of this. Also, keeping the cpu down, I guess they would have been doing a performance test every few lines of code.

[Friedemann]

Hi slirak,

we tried to recreate the test you did here but we did not see a substantial increase in the cpu consumption.
Could you check again if really all settings (like sample rate, buffer size) where the same in Live 7 and 8?

Best, Friedemann

[bulo]

Friedmann, many people did run the test have notice that there is a significant increase. Don't know for Slirak, but personnally, I've run the test with the exact same settings for every Live 6, 7 and beta 8. L7 is a little bit more hungry than L6, something around 5/7%, and L8 is something like 10 to 20% more than L7.

[glitchrock-buddha]

Hi slirak,

we tried to recreate the test you did here but we did not see a substantial increase in the cpu consumption.
Could you check again if really all settings (like sample rate, buffer size) where the same in Live 7 and 8?

Best, Friedemann

Hi Friedeman, thanks for stepping in.

The file some of us are testing with is a performance test Tarekith made a while back for Live 6 and 7. So since you couldn't see much difference between the two versions, it makes me think the cpu increase is only under certain circumstances. Which would possibly indicate a bug? This performance test uses only simplers and reverbs, so maybe it's a problem with those devices.

Here is the performance test which I modified slightly to almost max out on my machine:
http://www.mediafire.com/?sharekey=d9dc ... f6e8ebb871

There are 11 tracks with simpler playing one drum sample on each. Each track has 2 reverbs. There are also 4 return tracks with 2 reverbs each.

On my machine (macbook pro 2.2 Ghz, 3gb ram, OS 10.5.6), I get a 20% increase or more in cpu on Live 8:
Live 7: Peak at 77% for a second and then playback is steady at 70% (no stuttering or breaking up, all is smooth)
Live 8: Peak at 105% for a second and then playback is at 90% (completely broken up and stuttering, unplayable)

All settings above are equal between versions (Audio/Plug-in buffer etc).

Could you try the performance test on your machines? Perhaps you could narrow down the problem that way.

Regards

[FRGus723]

"goddamn interwebs is broken"

Nearly fell of my chair.

[Friedemann]

Hi glitchrock-buddha,

thanks for the test set. This looks definitely like a bug, we'll investigate it.

Best, Friedemann

[glitchrock-buddha]

Hi glitchrock-buddha,

thanks for the test set. This looks definitely like a bug, we'll investigate it.

Suck on that T-Deft!

I kid, I kid...

[Tone Deft]

lol-ing.

good work man, we all appreciate it.

[bulo]

Suck on that T-Deft!

I kid, I kid...

[slirak]

Hi slirak,

we tried to recreate the test you did here but we did not see a substantial increase in the cpu consumption.
Could you check again if really all settings (like sample rate, buffer size) where the same in Live 7 and 8?

Best, Friedemann

Hi Friedemann,

Sorry about the late answer, but I've been working and then off to a brilliant David Byrne concert and just got home - wow, that guy and his legacy is really something!

My results are 100% repeatable with my system at least. And yes, using identical settings.

Glitchrock is right about the test, but I've been using the original one found here: http://forum.ableton.com/viewtopic.php? ... mance+test

As I wrote originally, I get a 65% CPU activity (initially peaking at about 75%) with L8bxx, 49% (pretty steady, no initial peaks) with L7xx, which makes L8 about 30% more CPU hungry (in relative terms).

The difference is identical since at least L8b14 (that's when I got onboard the beta program). It's still the same in L8b20. I've tried trashing the prefs to no avail.

Real world projects exhibit similar differences.

Regards
/Jonas

[Nogi]

i compared the cpu performance with tarekith´s test, no more than 7 % increase of cpu power consumption

I suspect you are using a quad core? 7-8% per core => ~30%