LIVE FIVE PERFORMANCE TEST - results here

[Machinesworking]

So after talking to the guy in the store, and asking why the imac and macbook would be different, his best guess was the hard drive speed. Now I never thought that that had too much effect on cpu performance, but he says it definitely can and that would make sense here. The imac would have a 7200 rpm drive, and the macbook that ran this test had a 5400 rpm drive. I don't know what else could account for such a performance difference of two identical speed processors on the same operating system.

I'd be interested to see a test done on a macbook pro with a 7200 rpm drive, or even better, one of those new seagate perpendicular 160gb 5400rpm sata drives (which many computer folk think will be as fast as the 7200rpm drives but with the power consumption of a 4200rpm drive...)

still leaning towards the macbook here....

grb

I doubt it was the hard drive speed difference. There's only one audio file actually playing in the performance test and the other audio files are loaded into RAM.

Either way, the MacBook is DOPE!

Motherboards play into all of this as well. Funny because I thought the imac would be trounced by the macbook... Hopefully somebody springs for the 2.13 model and posts results...

I'm going to try and hold out for the merom macbook in the summer sometime, wish me luck!

[hoffman2k]

So after talking to the guy in the store, and asking why the imac and macbook would be different, his best guess was the hard drive speed. Now I never thought that that had too much effect on cpu performance, but he says it definitely can and that would make sense here. The imac would have a 7200 rpm drive, and the macbook that ran this test had a 5400 rpm drive. I don't know what else could account for such a performance difference of two identical speed processors on the same operating system.

I'd be interested to see a test done on a macbook pro with a 7200 rpm drive, or even better, one of those new seagate perpendicular 160gb 5400rpm sata drives (which many computer folk think will be as fast as the 7200rpm drives but with the power consumption of a 4200rpm drive...)

still leaning towards the macbook here....

grb

I doubt it was the hard drive speed difference. There's only one audio file actually playing in the performance test and the other audio files are loaded into RAM.

Either way, the MacBook is DOPE!

Motherboards play into all of this as well. Funny because I thought the imac would be trounced by the macbook... Hopefully somebody springs for the 2.13 model and posts results...

I'm going to try and hold out for the merom macbook in the summer sometime, wish me luck!

No luck needed dude! When in doubt... get a mini
I'm right there with ya. Holding out for merom.
They should be available around my bday
And then the joy of dual cpu support in Live 6..........

I'll probably walk around like this for the next 6 months:

[Groovecake]

Please explain the Merom issue. Thanks.

[Digi V]

merom is intels new chip architecture bringing 64bit computing (something amd has already) however it will be duel core (meaning two processors on one die)

intel claims these chips will run 30percent faster than the current high end desktop solutions AMD has. be reminded that 30percetn may not seem like a WOAH factor but these chips will be low power consuming chips for LAPTOP and MOBILE use.

so yeah that is a big deal .



the time frame given was summer 06, i dont think so though. if anything september AND BE REMINDEED that just because intel launches their new chips around this time DOES NOT MEAN you will get the new laptops then.


i estimate thanksgiving the earliest, january or feb the latest for macbook pros.

[AdamJay]

Are you sure that when you use the Macbook that you have the Energy Saver set to highest performance for Processor?

There is no highest performance processor mode.

and that is the problem.
when Ableton rented a shiney new HD G4 Powermac from Apple for the NAMM '06 Ableton booth, i noticed that the processor usage was exceptionally high when i was setting up the demos i made with Rahlo on it.

Apple's response was basically there old speed-stepping technology was too slow. When you needed more power, it didnt kick in IN time.
They have a new stepping technology as of the HD G4's and i presume now the Macbook Pro's. The plus to this technology is on "automatic" it kicks in when you need it MUCH faster, and is pretty much without flaws in that regard. You save battery life by not using processor you don't need. The negative is you cannot set the processor performance to "reduced, highest, automatic, etc."

So. unfortunately we will have to adapt and formulate a new benchmark test for determining the power of this new technology.

Logic Users are familiar with the "platinum reverb test."
We may need to do a Live Reverb test. That is, how many Live Reverbs can you run on a single stereo 16bit/44.1khz file before it craps out.

When i get my macbook pro (probably by the end of next week it should be delivered), i will devise a new test for this.

We can also start looking around or hoping for a haxie that allows us to overide this eternal "automatic cpu performance" feature of newer portable macs.

[bensuthers]

> That is, how many Live Reverbs can you run on a single stereo 16bit/
> 44.1khz file before it craps out.

it seems such a musical way of thinking doesn't it.

Have you heard my new piece for an ensemble of 96 Platinum Reverbs?

[AdamJay]

> That is, how many Live Reverbs can you run on a single stereo 16bit/
> 44.1khz file before it craps out.

it seems such a musical way of thinking doesn't it.

Have you heard my new piece for an ensemble of 96 Platinum Reverbs?

yea, it sucks. but thats adaptation for you.

there is probably some way to turn on highest cpu in Terminal... some way..

[glitchrock-buddha]

Are you sure that when you use the Macbook that you have the Energy Saver set to highest performance for Processor?

There is no highest performance processor mode.

and that is the problem.
when Ableton rented a shiney new HD G4 Powermac from Apple for the NAMM '06 Ableton booth, i noticed that the processor usage was exceptionally high when i was setting up the demos i made with Rahlo on it.

Apple's response was basically there old speed-stepping technology was too slow. When you needed more power, it didnt kick in IN time.
They have a new stepping technology as of the HD G4's and i presume now the Macbook Pro's. The plus to this technology is on "automatic" it kicks in when you need it MUCH faster, and is pretty much without flaws in that regard. You save battery life by not using processor you don't need. The negative is you cannot set the processor performance to "reduced, highest, automatic, etc."

So. unfortunately we will have to adapt and formulate a new benchmark test for determining the power of this new technology.

Logic Users are familiar with the "platinum reverb test."
We may need to do a Live Reverb test. That is, how many Live Reverbs can you run on a single stereo 16bit/44.1khz file before it craps out.

When i get my macbook pro (probably by the end of next week it should be delivered), i will devise a new test for this.

We can also start looking around or hoping for a haxie that allows us to overide this eternal "automatic cpu performance" feature of newer portable macs.

Ah yes, this makes total sense. There was a guy on this forum who had bought the latest rev of the 1.67ghz powerbook who was showing twice the cpu usage in reaktor ensembles that I was on my older rev 1.67ghz powerbook, and he said there was no power settings, whereas there is on mine. However he could still run as many plug-ins as I could because it became more efficient when more processor load was added. And I guess it's the same for the macbooks.

So that accounts for the difference between the macbooks and the imacs then. Makes me feel better about getting one, thanks.

grb

[AdamJay]

post deleted

[kenn michael]

great info AdamJay!

never thought about that. I'll have to do some Logic benchmarking to see if there is a speed discrepancy between the iMacs and the MacBooks there.

[DJSK]

Wow now I am starting to feel even better. Now I really can't wait for my Macbook to arrive. Still though I am wondering how this automatic processor setting will affect live performance. I sure hope it doesn't do more harm than good.

[Angstrom]

Are you sure that when you use the Macbook that you have the Energy Saver set to highest performance for Processor?

There is no highest performance processor mode.

and that is the problem.

Apple's response was basically there old speed-stepping technology was too slow. When you needed more power, it didnt kick in IN time.

This seems strange, I can understand the speedstep issue with something like photoshop - where a CPU intensive task would be hard to predict in time. But, for a DAW .. surely:
CPU thinks, hmm, I'm playing audio, should I step up? nah not yet, a moment later CPU thinks - still playing audio shall I step up...etc. Damn, if it hasn't stepped up after calculating DSP for over a minute, when the hell does this thing decide to step up to full speed?

Can you explain how the Logic reverb test overcomes this where the current one doesn't ?
It would be a shame to move to a 'reverb' test because I would imagine greater cross platform disparity in that test than the current one. By cross platform I mean between G series and Intel series computers of all marques. Essentiall what is being tested there is "can an intel chip calcualte reverbs better than a G series"
I guess Apple may relent and give back the manual power facility at some point.

[AdamJay]

This seems strange, I can understand the speedstep issue with something like photoshop - where a CPU intensive task would be hard to predict in time. But, for a DAW .. surely:
CPU thinks, hmm, I'm playing audio, should I step up? nah not yet, a moment later CPU thinks - still playing audio shall I step up...etc. Damn, if it hasn't stepped up after calculating DSP for over a minute, when the hell does this thing decide to step up to full speed?

Can you explain how the Logic reverb test overcomes this where the current one doesn't ?
It would be a shame to move to a 'reverb' test because I would imagine greater cross platform disparity in that test than the current one. By cross platform I mean between G series and Intel series computers of all marques. Essentiall what is being tested there is "can an intel chip calcualte reverbs better than a G series"
I guess Apple may relent and give back the manual power facility at some point.

It probably steps up once it gets to a certain % of processor used. As it is now, its not breaking 40% when stepped down, so why step up?
Perhaps it begins this stepping, slowly at 55% or 65%, etc.

The Logic reverb test, isn't one of my favorites, admittedly
But it is a Stress Test, and it still gives us a comparitive benchmark when comparing one machine to another machine to another machine.
If a MacBook Pro dual 2ghz can run 22 Reverbs, and an HP dual 2ghz can run 22 reverbs, and a G4 1.67ghz can run 11 reverbs then we know that the G4 is half as fast as the Intel machines.

Remember, my tests aren't about testing how well a given machine runs Live (everyone uses Live differently), but rather how much better or worse a given machine runs Live when compared to a 2nd and 3rd given machine.

[Angstrom]

what I mean to say is - perhaps due to the calculation differences between IBM G series and Intel Merom we may not see a true comparison.

Hypothetical example:
Intel Merom in Apple runs 220 reverbs at 100% CPU
Intel Merom in Apple runs 223 "complex warped" clips at 100% CPU

IBM G5 quad runs 250 reverbs at 100% CPU
IBM G5 quad runs 173 "complex warps" at 100% CPU

So different process may calculate better or worse on each platform, the true performance is not that the G5 runs more reverbs but the average of several key values, which is what we currently have.
(in the imaginary example above Intel total = 443, G5 total = 423)

Perhaps a modification to the current test might be better (for 5.2 and above) ?
keep adding reverbs on the master output of the curent test, report the base CPU then also the amount of default reverbs that are need to be added before 100% is reached?
it gives the current test's benefits (the broad scope) and adds the ability to scale up to reach the CPU ceiling?

[AdamJay]

good points Angstrom.

would you be interested in formulating such a test, yourself?