LIVE 6 & 7 PERFORMANCE TEST - Discussions Here

guanche · Post by **guanche** » Sat Dec 08, 2007 10:46 am

I'm running a p4 3 ghz and i got much more latency with live 7 than with 6!!!

Does anybody know the reason?

...and other thing... i like the concept of drum rack but it's not possible to make sends for individual tracks!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

I don't understand it!!

zeropoint · Post by **zeropoint** » Sat Dec 08, 2007 12:39 pm

sparklepuff wrote:Yup, Coolbook dropped the CPU usage in Live down to 56% for me. That's a 13% improvement. I'm sure I could tweak more to get better performance, I just installed it.

I am curious what it would do for your already low 56%, Tarekith. Maybe even knock it down to the low 40's, perhaps?

Can I ask....does the unregistered version of Coolbook measure the change in clock speed as you put load on the processor? I have tested it with the Live Performance test and the clock speed stays at 800mhz. on my MBP 2.4

Thanks

earsmack · Post by **earsmack** » Sat Dec 08, 2007 12:53 pm

Thanks for the corroboration - I am again looking forward to the 2.6GHz MBP

Lux Libra · Post by **Lux Libra** » Sat Dec 08, 2007 1:00 pm

guanche wrote: ...and other thing... i like the concept of drum rack but it's not possible to make sends for individual tracks!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

I don't understand it!!

open the chain editor, click on the R in the down left corner, put your send effects here, click on the S to adjust the send levels for each drumpad, voila

guanche · Post by **guanche** » Sat Dec 08, 2007 1:16 pm

m:o wrote:
guanche wrote: ...and other thing... i like the concept of drum rack but it's not possible to make sends for individual tracks!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

I don't understand it!!

open the chain editor, click on the R in the down left corner, put your send effects here, click on the S to adjust the send levels for each drumpad, voila

Ok!!!! merci man!! I'm gonna try.

Nokatus · Post by **Nokatus** » Sat Dec 08, 2007 1:26 pm

pribeh_tom wrote:I just installed it (and applied default settings with a "high throttle") on my Macbook Pro and I scored 49% instead of 67-9. That's a 20% difference

Doesn't that just mean that Coolbook throttles the CPU to its maximum setting, and you're seeing better percentages because of that

?

In other words, it seemingly makes the score dramatically better, as a load of 67% isn't necessarily high enough for the CPU to throttle to a higher speed according to the default settings. However, if you were to hit a higher load, the speedup would occur eventually in any case.

"Prematurely" throttling to a higher setting allows you to measure the potential of the system, as far as the load percentage goes, but it doesn't actually conjure 20% more performance out of the system. There is an actual benefit only if the default throttling is "lazy" enough to induce dropouts before speeding up.

pribeh_tom · Post by **pribeh_tom** » Sat Dec 08, 2007 5:04 pm

Nokatus,

I've suspected this all along but only now thought of testing it. I tested an iMac with almost identical specs and the test score showed up as 44 (so this leads me to believe that T's test is sufficient). But in order to ensure that what you're saying isn't the case all I did was was duplicate the tracks on my mpb, so 12 instead of 8. Prior to using coolbook the test broke my mbp, 90-110% (rendering it unusable). With coolbook however, I can run 12 at 64%.

pribeh_tom · Post by **pribeh_tom** » Sat Dec 08, 2007 5:05 pm

double. sorry.

R.J.Dubya · Post by **R.J.Dubya** » Sat Dec 08, 2007 5:07 pm

Nokatus wrote:
pribeh_tom wrote:I just installed it (and applied default settings with a "high throttle") on my Macbook Pro and I scored 49% instead of 67-9. That's a 20% difference
Doesn't that just mean that Coolbook throttles the CPU to its maximum setting, and you're seeing better percentages because of that ?

In other words, it seemingly makes the score dramatically better, as a load of 67% isn't necessarily high enough for the CPU to throttle to a higher speed according to the default settings. However, if you were to hit a higher load, the speedup would occur eventually in any case.

"Prematurely" throttling to a higher setting allows you to measure the potential of the system, as far as the load percentage goes, but it doesn't actually conjure 20% more performance out of the system. There is an actual benefit only if the default throttling is "lazy" enough to induce dropouts before speeding up.

Which it is. Because Ableton stutters at pretty low cpu quite often (lke 65%), it is totally worth it to get that extra throttling from the processor. The computer doesn't think it needs to be at full speed when live is around 60-70%, but at the same time that's really Live's maximum useable percentage, unfortunately. So you end up maxing out live before the computer even decides to go to full speed. I personally think it stinks both on the part of apple for it's crappy speed-steeping implementation, and ableton for either it's poor stability at moderate cpu use or improper cpu reporting or something.

Think about it. Live says "I'm using 70% of the cpu". The macbook sees that much of it needed, and stays at a lower cpu rate, yet ableton is crapping out and needs more power. I'm no hardware wiz, but it seems like they need to communicate better. It bugs me to have to get a thrid-party app to use the power that's already there. And it also bugs me that Live says it's at 65% when really it's geting near maxed. It should say 95%.

pribeh_tom · Post by **pribeh_tom** » Sat Dec 08, 2007 5:07 pm

Mammalux wrote:Can I ask....does the unregistered version of Coolbook measure the change in clock speed as you put load on the processor? I have tested it with the Live Performance test and the clock speed stays at 800mhz. on my MBP 2.4

It should show you which speeds its jumping around from. In my case it shows that it jumps from 2171 to 2338. Buy it though, and try it out. It's only ten bucks. You can't go wrong. And test it then.

pribeh_tom · Post by **pribeh_tom** » Sat Dec 08, 2007 5:16 pm

R.J.,

If you use coolbook to monitor the frequency (speed) of your processor without modifying it (you download and do this for free) you'll notice that the speed is jumping constantly back and forth between frequencies an awful lot while playing the T's test. Once I started just throttling I saw 49% come up.

Another thing, if I only throttled (on high, with default settings) in CoolBook I would get 120% spike at first while running 12 tracks (of T's) test. Once I used the highest processor frequency (by setting it as the only option for the MBP) I could play 12 tracks (of T's) test without any high spiking (120%) while getting a smooth 64%.

Tarekith · Post by **Tarekith** » Sat Dec 08, 2007 5:22 pm

sparklepuff wrote:I am curious what it would do for your already low 56%, Tarekith. Maybe even knock it down to the low 40's, perhaps?

Again (no offense), but I really don't care enough to try it.

If I can get 56% running a project that uses way more plug ins than I'll ever use in real life, why mess with it? That's still way more than enough juice for me finish a song, so it's just not something I want to even mess with.

One question though, wouldn't doing this kind of work cause the fans to kick on more frequenctly? I'd imagine using more CPU power requires more mains power, which means more heat, no?

capo-wear-i · Post by **capo-wear-i** » Sat Dec 08, 2007 7:17 pm

Tarekith wrote:
sparklepuff wrote:I am curious what it would do for your already low 56%, Tarekith. Maybe even knock it down to the low 40's, perhaps?
Again (no offense), but I really don't care enough to try it. If I can get 56% running a project that uses way more plug ins than I'll ever use in real life, why mess with it? That's still way more than enough juice for me finish a song, so it's just not something I want to even mess with.

One question though, wouldn't doing this kind of work cause the fans to kick on more frequenctly? I'd imagine using more CPU power requires more mains power, which means more heat, no?

By definitition, i'd say definitely yes. Certainly setting my PC laptop to 'always on' produces more heat when runinng ableton. I recommend the Zalman NC1000 cooler - miles quiter than your laptop fans on full tilt. That way you get all the power without the racket.

donnydonny · Post by **donnydonny** » Sat Dec 08, 2007 8:21 pm

R.J.Dubya wrote:
garyboozy wrote:ok after my higher than expected cpu hit with the same mac as tarekith, i've updated to leopard. tried the test again with the built-in audio and got the same hit of 67%, which is what i had with tiger using both built-in and using the saffire LE. was a bit gutted. so, plugged in the saffire and i now get 54% cpu! seems like leopard has been seriously tweaked for firewire audio use. very nice!
Wow, that's really good to know. A 15% drop because of a firewire interface? I'll have to check this with my firepod. So just having your outputs set to use your saffire brought on the improved cpu useage? I wonder how USB2 interfaces fare...

With my Mbox 2, I too, got 68% cpu usage on my Macbook Pro. After reading the posts in this thread, I did the test with my Ultralite and wuddya know... 55%. Amazing.

pribeh_tom · Post by **pribeh_tom** » Sat Dec 08, 2007 9:23 pm

donnydonny,

Thanks for confirming this. It's clear now, that in order to get maximum performance out of your MBP with Live you should:

- use a firewire audio interface (or expresscard if you can afford a symphony system).

- use CoolBook to throttle the processor or force OS X to run the processor at full speed.

- And I'd add a third: and that's to use an external firewire hard-disk to record and store your audio and all other project files.