Latency Tests Report (based on iSchwartz posts)

Question

Hi all,
With an hour to kill at home while my family gets here,I set to discover what some latency figures are on my system.Please look at my signature,the equipment list is current.
Specifically I wanted to see if the latency figures were consistent,in order to insert a record Delay best suited to each IO buffer size for my particular equipment.also,I wanted to explore the possible default latencies of some of my favorite Logic plugins,plus one Nomad Factory EQ.
This is what I found:
Latency Test:
Digi002Rack,
Logic Pro 7.2.3
CoreAudio Driver 7.3cs1
44.1kHz 24bit Logic session
Physical outputs 1+2 and back into channel inputs 1+2,
with 6" balanced TRS audio cables.
Plugin Delay Compensation set to =ALL
Sample Accurate Automation set to=volume,Pan,Sends only.
Pass Keyboard Events to Plugins set to=all.
Audio Merge Defaults: crossfade time 20mS curve 0
Core Audio: Enabled
System Memory requirements:114.0 MB
Driver:Digidesign HW( 002)
I/O buffer Size = (see chart below)
Recording Delay = (see chart below)
Max Number of Audio Tracks: 80
64 Busses option set to: OFF (unchecked)
Universal Track Mode: ON (checked)
Larger Disk buffer: ON
24bit Recording:ON
Software Monitoring: ON
Process Buffer Range:LARGE
ReWire Behavior:Playback Mode (Less CPU Load)
Maximum scrub Speed:Normal
Scrub response:Normal
WITH NO PLUGINS,and 0 as the Recording Delay:
IO buffer size = 32 274 samples
IO buffer size = 64 274 samples
IO buffer size = 128 350 samples
IO buffer size = 256 606 samples
IO buffer size = 512 1117 samples
IO buffer size = 1024 2140 samples
By setting equal but negative values for the sample delay,I compensated the throughput of the Digi002R.
WITH NO PLUGINS,
Recording Delay set negative to sample delay above:
IO buffer size = 32 0 samples
IO buffer size = 64 0 samples
IO buffer size = 128 0 samples
IO buffer size = 256 0 samples
IO buffer size = 512 0 samples
IO buffer size = 1024 0 samples
With 1 Channel EQ on click track:
IO buffer size = 32 0 samples
IO buffer size = 64 0 samples
IO buffer size = 128 0 samples
IO buffer size = 256 0 samples
IO buffer size = 512 0 samples
IO buffer size = 1024 0 samples
With 1 GuitarAmpPro on click track[default setting]:
IO buffer size = 32 14 samples
IO buffer size = 64 14 samples
IO buffer size = 128 14 samples
IO buffer size = 256 14 samples
IO buffer size = 512 14 samples
IO buffer size = 1024 14 samples
With 1 SpaceDesigner on click track[default setting]:
IO buffer size = 32 -128 samples
IO buffer size = 64 -128 samples
IO buffer size = 128 -128 samples
IO buffer size = 256 -128 samples
IO buffer size = 512 -128 samples
IO buffer size = 1024 -128 samples
With 1 LinearPhaseEQ on click track[default setting]:
IO buffer size = 32 0 samples
IO buffer size = 64 0 samples
IO buffer size = 128 0 samples
IO buffer size = 256 0 samples
IO buffer size = 512 0 samples
IO buffer size = 1024 0 samples
With 1 Nomad Factory PEQ5B EQ on click track[default setting]:
IO buffer size = 32 0 samples
IO buffer size = 64 0 samples
IO buffer size = 128 0 samples
IO buffer size = 256 0 samples
IO buffer size = 512 0 samples
IO buffer size = 1024 0 samples
I hope this is informative to you out there wondering all about latency,I/O buffer settings,et al.
Cheers

Answer 1

I posted this to the logic pro forum, but in case it helps anyone else here it is. Since it is slightly different from iS's method, whereby he drags the file around in the sample editor, I use the sample delay plug, I figured someone might benefit from this, and this thread has the right title for it.
Well I don't know about firewire latency issue specifically, but here is the hardware I tested and results:
Apogee Symphony
Apogee Rosetta 800
MacPro
Logic 7.2.3
OS X 10.4.9
Ran the test which went like this for anyone who wants to try it.
1- At your audio interface, take output 8 and connect it to input 8 (or which ever in out combo you want, create a loop back connection)
2- In logic, setup the input of Master Test Track to the output of your click track so you can record your click track to an audio signal. This will give you a nice transient to work with and a simple wave form for you to look at.
3- After you record Master Test Track setup the I/O out of the track to send to track 7-8 (if you used out 8 on your interface) and pan it hard left so it will send on out 8.
4- Create new audio track on track 2 (or available open track) and set its input to input 8.
5- Record the signal that you just sent from Master Test Track to track 2
6- Before you play back, insert in Master Test Track and add sample delay, which is in the logic>delay section of plugins. Be sure it's set to 0.
7- Insert on Track 2 the gain plugin and reverse the phase by clicking the reverse phase button.
8- Be sure that both tracks I/O out at this point go to your monitors. Both tracks dead center.
9- Start playback. I setup a cycle loop over a certain section of the recording so it will loop as I look for the sample diff.
10- Go to Master Test Track and open the sample delay plug and if you click and hold where the number is, you will see a little - show up. This will let you increase the number slowly.
11- As it's playing and you increase the number, thereby delaying the master track's playback, you will hear the sound change and start to fade. At one point it will either become inaudible or barely audible. if you go one extra sample after that, it will start getting louder. The number before it gets louder is your record latency. So if 83 is softer than 84 or 82 than this is your number.
12- To be absolutely sure this is right. Reset your test by deleting the recorded audio on Track 2. Bypass the Master Test Track's sample delay. Now go to the Audio and Hardware Drivers settings and set your Recording delay to the NEGATIVE value of your result UNLESS your sample delay is negative. So in my case it was 83, I set it to -83. Repeat the steps to record the audio through your interface. What should happen this time is that when you play back with the Master Test Track sample delay BYPASSED you should either not hear the click or it should be incredibly faint and sound exactly like your test. This means your test worked, your numbers are right, and your tracks will line up nicely from now on.
What my tests showed...
It seems that using the above setup I have a sample shift of 83 samples. That number is consistent through reboots and hardware shut down. I tried it 3 times and each time the right number was 83.
I don't know if this behavior will be different with firewire audio interfaces since I use an audio card. However, for the hardware I use, it is absolutely consistent.
I will keep my eye on this number for a while to make sure it doesn't shift.
If I misunderstood something about the results, fire off.
R