Ableton - Can you help with Live 7 performance test?

Tarekith · Post by **Tarekith** » Fri Nov 30, 2007 3:13 pm

Is there anyway that someone at Ableton can devise a new performance test project, something that will take into account the fact that most laptops these days are multi-processor and how Live 6 and 7 deal with this? I know that you did not create the earlier performance test, but it's obviously not relevant anymore, yet people keep referring to it and then getting the results confused.

I would figure that you guys would be able to design the best test to spread processor load, based on how the software is coded. It doesn't even have to be musical or anything, just something that would give us more realistic results based on how newer versions of Live work.

I think this would be a great help!

zeropoint · Post by **zeropoint** » Fri Nov 30, 2007 4:40 pm

sparklepuff · Post by **sparklepuff** » Fri Nov 30, 2007 4:58 pm

globalgoon · Post by **globalgoon** » Fri Nov 30, 2007 5:50 pm

how about:

first 4 bars = 1 track enabled with simpler + 2 reverbs
next 4 bars = 2 tracks enabled each with simpler + 2 reverbs
next 4 bars = 3 tracks enabled each with simpler + 2 reverbs
next 4 bars = 4 tracks enabled each with simpler + 2 reverbs
.
.
next 4 bars = 8 tracks enabled each with simpler + 2 reverbs

then report the CPU for each 4 bar section

but save it in Live 6 so a 6 --> 7 comparison can be made

eg: http://globalgoon.com/test.rar

for my core2quad gives 11,12,12,13,19,20,22,23

Tarekith · Post by **Tarekith** » Fri Nov 30, 2007 5:57 pm

I personally don't care what it is, I'd just like Ableton to chime in and let us know FOR SURE the best way to set it up.

globalgoon · Post by **globalgoon** » Fri Nov 30, 2007 6:19 pm

Tarekith wrote:I personally don't care what it is, I'd just like Ableton to chime in and let us know FOR SURE the best way to set it up.

Robert Henke wrote:Each track including the returns are splitted amongst the cores.

Like:

track 1 core A
track 2 core B
track 3 core A
...
...
return A core A
return B core B
...

Robert

Tarekith · Post by **Tarekith** » Fri Nov 30, 2007 6:22 pm

Cool man, go nuts then

globalgoon · Post by **globalgoon** » Fri Nov 30, 2007 10:44 pm

UPDATED FIGURES, I had the audio buffers set differently in live 6 and live 7

core 2 quad PC XP sp2 v6 versus v7demo
v7: 8,9,9,9,16,17,17,18
v6: 8,8,8,9,16,16,17,18

djastroboy · Post by **djastroboy** » Fri Nov 30, 2007 10:56 pm

globalgoon wrote:core 2 quad PC XP sp2 v6 versus v7demo
Code: Select all
Tracks	V6	V7
01			11	08
02			12	09
03			12	09
04			13	09
05			19	16
06			20	16
07			22	17
08			23	18

Are you saying that the CPU usage in 7 is twice as efficient 6? That's amazing. Are you sure there's not something else set differently?

globalgoon · Post by **globalgoon** » Fri Nov 30, 2007 10:59 pm

djastroboy wrote:...
Are you saying that the CPU usage in 7 is twice as efficient 6? That's amazing. Are you sure there's not something else set differently?

i did something wrong in the first test, have updated the figures!

Tarekith · Post by **Tarekith** » Fri Nov 30, 2007 11:03 pm

Care to share your project file? I'll host it if you want.

globalgoon · Post by **globalgoon** » Fri Nov 30, 2007 11:03 pm

oops, the audio buffer was set to 64 samples for one test and 576 for another

I'll redo the test

globalgoon · Post by **globalgoon** » Fri Nov 30, 2007 11:09 pm

Tarekith wrote:Care to share your project file? I'll host it if you want.

yeah, it's here: http://globalgoon.com/test.rar
with the audio buffer set to 512 samples:
v7: 8,9,9,9,16,17,17,18
v6: 8,8,8,9,16,16,17,18

6 and 7 seem to be pretty much the same!

globalgoon · Post by **globalgoon** » Fri Nov 30, 2007 11:16 pm

comparing single processor mode to all 4cores working in live 6 I get

1 core: 10,18,26,33,42,50,58,66
4 core: 08,08,08,09,16,16,17,18

Tarekith · Post by **Tarekith** » Sat Dec 01, 2007 12:18 am

Cool, I forgot I don't have version 6 installed currently, doh.