Live is so slow with plug-ins-- stats tell the truth

[feyshay]

Let me put some figures up here.
I compared Live 5.02 (afraid to get 5.03 with what I've heard, waiting for 5.2) to Sound Forge 8--running high intensity plugs within them.
I used two devices--MOTU Traveler (Firewire) and a USB Mini-Dac--they had comparable speeds.
I have a Dual Processor Dell 3.2GHz with 2GB RAM. Fast computer.
The VST or DX (because you can also run DX within Sound Forge) I ran included PSP Neon HR (a new high resolution linear phase equalizer), MasterComp (another high intensity), and a RoomVerb M2 from Spin Audio.
I was using 44 kHz samples at 24bit. I had my sample on my soundcards down to in the thousands (2024?). 44kHz/24 Bit on my soundcards.

My resting CPU rate in Live 5 is 2-3%.
RoomVerb as plug-in within Sound Forge (3% CPU use).
Within Live--12%.
It gets worse.
PSP MasterComp on FAT in Sound Forge runs 5-7% range.
PSP MasterComp on FAT runs can run as high as 40% in Live 5.
It gets even worse.
PSP Neon HR--
Standard res in FAT Mode in Live 5 CPU consumption at 49% (on the Traveler) and 46% on the Apogee Mini-Dac USB
8% in Sound Forge
High Res without FAT Mode in Live 5= 47%
7% in Sound Forge
Maximum res without FAT Mode--Froze Live 5 (had to reboot--why not just crash the VST or something. Live is so bad I had to reboot every time I got to high on CPU consumption).
10% in Sound Forge
High res with FAT Mode--crashed in Live 5
10% in Sound Forge
Max Res with FAT Mode--crashed in Live 5
12% in Sound Forge

I was able to chain Neon in Max res/FAT mode, with MasterComp in FAT Mode, and RoomVerb all together and still was only at around 50% CPU consumption in SoundForge. Live 5 would have crashed just the Neon in High res/FAT Mode.
I made sure everything was matched. Stats don't lie.
Luckily, I do my mastering in Sound Forge...
What is wrong with Live? I don't have another DAW to compare it to, but it is really inefficient.

[Lunch Money]

I just can't understand how the host's code could affect it so drastically. In essence, each plug-in is just a unit of code that the host "allows" interaction with. There's no "logical" reason for one host to use more CPU with the same plug-in than another!

I'm quite confident that someone out there could explain it to me, but as an outside observer (ie. not a programmer), it doesn't make any sense at all.

I'm not questioning your discoveries-- on the contrary, I believe them. But I just don't know why the heck it should be the case.

Greg

[soulkraka]

cpu rate via task manager processes or app?

[jackson]

Yep, I've found the same versus running Logic 5.5 - try it one day, you'll be blown away at how many plugins you can use at once. I'm really struggling to come to terms with how few plugins I can use in Live and eXT.

And it's not just because of Logic's useful feature of dynamically turning off plugins that aren't procesing any audio.

[FaX]

Let me put some figures up here.
I compared Live 5.02 (afraid to get 5.03 with what I've heard, waiting for 5.2) to Sound Forge 8--running high intensity plugs within them.
I used two devices--MOTU Traveler (Firewire) and a USB Mini-Dac--they had comparable speeds.
I have a Dual Processor Dell 3.2GHz with 2GB RAM. Fast computer.
The VST or DX (because you can also run DX within Sound Forge) I ran included PSP Neon HR (a new high resolution linear phase equalizer), MasterComp (another high intensity), and a RoomVerb M2 from Spin Audio.
I was using 44 kHz samples at 24bit. I had my sample on my soundcards down to in the thousands (2024?). 44kHz/24 Bit on my soundcards.

My resting CPU rate in Live 5 is 2-3%.
RoomVerb as plug-in within Sound Forge (3% CPU use).
Within Live--12%.
It gets worse.
PSP MasterComp on FAT in Sound Forge runs 5-7% range.
PSP MasterComp on FAT runs can run as high as 40% in Live 5.
It gets even worse.
PSP Neon HR--
Standard res in FAT Mode in Live 5 CPU consumption at 49% (on the Traveler) and 46% on the Apogee Mini-Dac USB
8% in Sound Forge
High Res without FAT Mode in Live 5= 47%
7% in Sound Forge
Maximum res without FAT Mode--Froze Live 5 (had to reboot--why not just crash the VST or something. Live is so bad I had to reboot every time I got to high on CPU consumption).
10% in Sound Forge
High res with FAT Mode--crashed in Live 5
10% in Sound Forge
Max Res with FAT Mode--crashed in Live 5
12% in Sound Forge

I was able to chain Neon in Max res/FAT mode, with MasterComp in FAT Mode, and RoomVerb all together and still was only at around 50% CPU consumption in SoundForge. Live 5 would have crashed just the Neon in High res/FAT Mode.
I made sure everything was matched. Stats don't lie.
Luckily, I do my mastering in Sound Forge...
What is wrong with Live? I don't have another DAW to compare it to, but it is really inefficient.

Strange I have a P4 HT 3.2ghz P4 Toshiba P20 notebook.
Live ALWAYS idles at 0% rest rate.
Reads the same in Task Manager.
I find I can use more than enough plugins myself.
Infact on tracks where all midi tracks have been resampled and arranged as audio only I have had in excess of 32 and up too 45 plugin effects at once with no CPU over 55%.
I can easily sequence 16 Combinators (from within LIVE using Reason 3) and add effects + run further stereo tracks no problem.
Something really sounds wrong there.
I don't use those plugs but Live is more efficient than P5V2 on my system with the same plug ins in 24/48 (which is my prefered sample rate and bit depth).
What audio card and drivers are you using BTW.
I can generally work in 128-256 buffer sizes no problems also.
How many processes are open in your Task Manager when nothing else is running ie:straight on bootup.
I only have 15 processes active straight on bootup before I start LIVE.
Just curious too know as those figures seem way off.

[computermusic500]

Strange I have a P4 HT 3.2ghz P4 Toshiba P20 notebook.
Live ALWAYS idles at 0% rest rate.
Reads the same in Task Manager.
I find I can use more than enough plugins myself.
Infact on tracks where all midi tracks have been resampled and arranged as audio only I have had in excess of 32 and up too 45 plugin effects at once with no CPU over 55%.
I can easily sequence 16 Combinators (from within LIVE using Reason 3) and add effects + run further stereo tracks no problem.
Something really sounds wrong there.
I don't use those plugs but Live is more efficient than P5V2 on my system with the same plug ins in 24/48 (which is my prefered sample rate and bit depth).
What audio card and drivers are you using BTW.
I can generally work in 128-256 buffer sizes no problems also.
How many processes are open in your Task Manager when nothing else is running ie:straight on bootup.
I only have 15 processes active straight on bootup before I start LIVE.
Just curious too know as those figures seem way off.

Whoa...I am amazed at your efficiency with Live. Annnd also the fact that you only have 15 processes running because I currently
have 46!!! How can I decrease that to 15? I need to figure out what I can disable!

[jackson]

Here's a good start:
http://members.rushmore.com/~jsky/id01.html

I only have 15 processes running on a P4 3Ghz with a gig of RAM. I still find Live to be inefficient.

[computermusic500]

Here's a good start:
http://members.rushmore.com/~jsky/id01.html

I only have 15 processes running on a P4 3Ghz with a gig of RAM. I still find Live to be inefficient.

Thanks for the useful website, and yes, even after disabling a lot of processes Live is still hectic on the CPU.

[FaX]

Maybe because I only have Project 5 too compare it too .
I use an external 7800 RPM 8meg cache FW audio drive.
PCMCIA audio interface @ 24/48.
The notebook was configured by Vinn@AAVIMT (PlanetAwave).
Totally rock solid performance wise.
I find I get great use out of it for my needs.
I have a swag of Reaktor4/Session ensembles for example that only use 4-8% CPU per instance with R5.
String patches for solo cello / viola etc in DimensionPro only use 2% and a Piano with 6-7 note chord only 7-8% CPU in Live for example.
Most synth edit synths are mega resourceful for me also.
None of Ugo's stuff NEVER goes over 10% per instance.
I've got delays - eq's - gates - comps etc that use little more that 1-4% per instance at most also.
I also find the more I chuck at it the tighter the CPU consumption becomes.
Live will happily function at 80% CPU without a glitch if I wan't to really push it with Reaktor ensembles also.
Strange is all I will say.
As stated though I only have P5 V2 to compare it too.

[kara]

I just can't understand how the host's code could affect it so drastically. In essence, each plug-in is just a unit of code that the host "allows" interaction with. There's no "logical" reason for one host to use more CPU with the same plug-in than another!
Greg

There definitifly is ! I could make the same comparision table between Ext and Chainer, same fx,same routing,same PC, Ext uses more CPU then Chainer
And that's a fact

[opia]

keep up the good work
I'm sure lowering those cpu percentages will help people make better music

[fabi]

keep up the good work
I'm sure lowering those cpu percentages will help people make better music

keep up the good work
I'm sure lowering those cpu percentages will help people make music.

[opia]

I'm silenced by your wit

[jupiter8]

I have'nt made a direct comparison but i did'nt notice a difference between Cubase and Live. Maybe i'll look into it. But i doubt the differences will be that big for me.

[fabi]

I'm silenced by your wit

apparently not quite.


if eXT2 will be more optimized cpu-wise that'd be wonderful, but things being as they are now i can work just fine, already.

can't/don't want to afford live, so can't comment on its behavior.