The Bitwig evaluation thread

[Da hand]

OK this is pretty big!
As stated in another thread I was interested in doing a CPU test.
The Machine is a Mac Pro 8 core 2.66 Xeon with 24mb RAM, RME Fireface interface.

[[[ The caveats- Live 8 32bit VS Bitwig 64bit and DP8 64bit.
I'll run the same test with Live 8 64 bit later when I download and instal it. Ableton claim no difference in CPU but I seem to recall a 10% bump? Also since DP beat the crap out of the other two so badly I didn't bother raising it's buffer to 256. ]]]

Massive running the patch 1991 single bass not, 8 bars with an audio loop of a guitar note playing as well.

Live: 15 instances.
256 sample buffer.
Activity Monitor showing 312% CPU

Bitwig: 38 instances
256 sample buffer.
Activity Monitor in OSX shows around 750% CPU usage on the three threads Bitwig has there.


Digital Performer: 60 instances
This is at a 64 sample buffer!
Showing 1,499% CPU. DP seems to be handling multithreading/virtual cores very well.

So Bitwig beats the shit out of Live CPU use wise, actually kind of surprising.

I think you are missing some key info in your test results. All I see at the moment is that an increase in Massive instances increases CPU usage in a different program. What happens when you run 60 instances in Live?

[stringtapper]

OK this is pretty big!
As stated in another thread I was interested in doing a CPU test.
The Machine is a Mac Pro 8 core 2.66 Xeon with 24mb RAM, RME Fireface interface.

[[[ The caveats- Live 8 32bit VS Bitwig 64bit and DP8 64bit.
I'll run the same test with Live 8 64 bit later when I download and instal it. Ableton claim no difference in CPU but I seem to recall a 10% bump? Also since DP beat the crap out of the other two so badly I didn't bother raising it's buffer to 256. ]]]

Massive running the patch 1991 single bass not, 8 bars with an audio loop of a guitar note playing as well.

Live: 15 instances.
256 sample buffer.
Activity Monitor showing 312% CPU

Bitwig: 38 instances
256 sample buffer.
Activity Monitor in OSX shows around 750% CPU usage on the three threads Bitwig has there.


Digital Performer: 60 instances
This is at a 64 sample buffer!
Showing 1,499% CPU. DP seems to be handling multithreading/virtual cores very well.

So Bitwig beats the shit out of Live CPU use wise, actually kind of surprising.

I'm not following here. Why the difference in number of instances or difference in CPU percentage threshold?

Wouldn't you want to have a benchmark number of instances or a benchmark percentage and then evaluate when each DAW hits those benchmarks and compare?

[Machinesworking]

Sorry I forgot to mention this is a "to Failure" test.
It's a pretty standard way to test the CPU performance of a DAW.

Use exactly the same plug in with the same patch in it in every DAW you test.
Add instances of the plug in until audio crackling back off until the crackling is gone and you have an idea of what the DAW itself can handle CPU wise.

I added an audio track just to make it closer to a real world test, it's unimportant though, all DAWs these days can handle massive amounts of streaming from disk.

Live is clearly not using anywhere near the 8 cores of the Mac Pro, and DP seems to be using the virtual "16" that manufacturers have bragged about.

[Machinesworking]

I'm not following here. Why the difference in number of instances or difference in CPU percentage threshold?

Wouldn't you want to have a benchmark number of instances or a benchmark percentage and then evaluate when each DAW hits those benchmarks and compare?

Actually no, not at all. Every DAW reports CPU in a slightly different way. For instance DP acts like it's used all it's power after the first two cores are full, then continues on to use all 8 cores without compromising audio quality. The "Live Performance Test" is severely flawed, I don't go on about it because the people who did it had good intentions, but it's not at all a representation of what your computer can handle, it's just a representation of what Live 'thinks' your computer can handle.

[stringtapper]

That's what I figured after thinking about it.

But if you look at the numbers the number of instances and their corresponding percentages all seem more or less proportional.

15 instances = 312%

30 instances = 750%

60 instances = 1499%

So is it more about how much of the CPU Live is optimized to use rather than how much actual CPU it takes to run a comparable amount of instances?

[deva]

What happens if you take out the audio file? Are the results comparable?

Also, did you have Bitwig configured to use a separate process per plugin?

[Angstrom]

Never trust that CPU reporting for anything other than a contextual guide. When I first got StudioOne I freaked out that the resource usage seemed so much higher than Live (both using in-app meters). Until I checked using actual performance tests, my real world usage, they were roughly equivalent. They could achieve a similar amount of work.

Crackles and drop outs are the only real guide.

[Machinesworking]

That's what I figured after thinking about it.

But if you look at the numbers the number of instances and their corresponding percentages all seem more or less proportional.

15 instances = 312%

30 instances = 750%

60 instances = 1499%

So is it more about how much of the CPU Live is optimized to use rather than how much actual CPU it takes to run a comparable amount of instances?

Mostly it seems to me Live can only use two cores for audio performance.
Bitwig seems to be able to use all 8 cores, and DP can use the 16 virtual cores of the Xeon chips.
Honestly even 15 is taxing for Live. This has been an issue with Live forever on machines with more than two cores.

[Machinesworking]

Never trust that CPU reporting for anything other than a contextual guide. When I first got StudioOne I freaked out that the resource usage seemed so much higher than Live (both using in-app meters). Until I checked using actual performance tests, my real world usage, they were roughly equivalent. They could achieve a similar amount of work.

Crackles and drop outs are the only real guide.

bingo!
This makes Bitwig that much more appealing to me.
I don't use ReWire much with Live, but the lack with Bitwig is a big negative though.

[yur2die4]

Why is this test being done on Live 8? I'm confused about it all.

[stringtapper]

Because that's what he has. And yeah a proper one would need to use 9.

[One Reason]

I'm afraid that's STS.

Silly testing System

[Machinesworking]

Because that's what he has. And yeah a proper one would need to use 9.

Sure, but Live 9 certainly hasn't reported less CPU use!
In fact most people are reporting the opposite.

[Machinesworking]

I'm afraid that's STS.

Silly testing System

You don't have a clue about what you're talking about.
Not harsh, just a basic fact. This is as close as you can get to a level playing field.

[One Reason]

I'm afraid that's STS.

Silly testing System

You don't have a clue about what you're talking about.
Not harsh, just a basic fact. This is as close as you can get to a level playing field.

Bitwig has been spiking my CPU all day, I know that for sure..so.. as far as knowing what I'm talking about.. that.. I do know.

By the way...Empty project, open for two mins... Live 9 - 188K Mem. Butwang - 741K Mem.