Latency Questions for MOTU 896HD

I am trying to learn the relationships between samples per buffer with input and output latency. Currently, my input latency is 5.351 ms and my output latency is 8.844 ms. My samples per buffer is set at 192 from within the MOTU control panel. Naturally, as I decreased the samples per buffer my latency can drop significantly. When I first started using this interface my samples per buffer was 1024. I had been quite successful at 256 and still notice no change at 192.
I am aware that as latency increases the delay on monitor is greater and have heard as much as a quarter second as it would appear in the delay so I recognize this is not acceptable, but I’m looking for other areas of symptomatic problems that may develop as I go above or below what could be considered normal. Is there a preferred standard to stay with.

What are the symptoms with problems developing as the buffer sizes increase or decrease? And is there a brief explanation as to why it may occur? I am using the MOTU Audio ASIO Driver. and NO Asio Guard. Sample rate set to 44.1kHz. Thanks in advance for your thoughts

Hey there Knuckle 47,

Here is my understanding. Alot of your latency performance is related to your computer’s processing power. As you decrease the buffer size (and increase the sample rate) the computer has less time to properly stream the audio, and consequently, that is when you hear the common pops/clicks so forth.

It really isnt uncommon to set you buffer size as low as 64 samples for DAW playback on superior systems and audio cards with decent drivers. MOTU’s drivers are very decent for the price range, and much better than say M-audios or Presonus’s.

Now when setting the latency too low on my system when controlling say a VSTi with a keyboard, many times you can get this weird chorus effect. I think the cause is a bit technical as I recall, but normally you want to allow for more buffer when you are monitoring MIDI and obviously use your zero latency DSP software when recording audio. Since playback audio only encounters the ASIO output buffer, it can most likely take more punishment at lower buffer sizes than a recording setup regardless of your system.

Hello Bane,

Thank you for your explanation. I can see I will need to do some more testing on these settings. I did not realize, if I understand correctly, that I might use one setting for recording and a second setting on playback. I also need to check my MIDI configurations under record conditions and see which direction may stress the system using the outline you have described.

My question then is, if I cannot hear the pops, clicks or other elements of the wrong settings, will the performance metering be a good indicator of which settings will prove to be the most acceptable. My thinking on that says, IF…the computer were so capable, is there a window of Cubase operations that will fall well WITHIN the window and never show signs of performance issues?

Thanks again. These “thoughts and questions” arise from my reading the book after all these years and getting much deeper into the software s’ use


No, I am not referring to settings but rather the difference between ASIO input/output buffers provided by the driver. I’m not an expert on these things either, but all logical indications point to that outgoing audio will only encounter the output buffer and not the input, as an incoming audio signal would. This is a big deal, cause the input buffer accounts for approximately half of your ASIO latency.

Well in this case there are several different external factors on the hardware side especially, for instance the internal buffers and related safeguards that the interface has. As such, Cubase could obviously not account for these factors, but only for the strain being placed on your systems resources. Moreover, certain interfaces will allow you to push things a bit farther than others, though depending on your setup, this can go too far…

All in all, it defintely wouldnt hurt for you to try out different latency settings and see how far you can push it. I would guess you could get pretty darn low on a system like yours, with MOTU’s driver stability and performance also being above average, maybe you can achieve stability at an exceptionally low buffer size. Please let me know how it goes for you. :wink:

Hey Steve,

Nice link, that confirms my thoughts about the ASIO buffer signal flow. I believe that Presonus has a similar article concerning latency and signal flow, just maybe a little simpler. Its you look under their Audiobox VSL interfaces.

My system is very similar to what you were first describing there, except I’m on Vista. I’m having a helluva time trying to run sample based projects in Cubase, and though I maxed out my RAM slots with 2x2GB DDR2’s the 32bit OS is killing me by not letting me take full advantage of it.

Another thing to check out Knuckle is your computer’s power modes. Alot of times running on a powersaver/eco type mode will also affect latency performance. On my system, if the settings are switched from High Performance to Balanced, there is a very exact, reproducible overload of clicks/pops, etc.

Like I said earlier, MOTU’s drivers are very respectable and can achieve nice latency. I just upgraded my Steinberg CI1 to a Focusrite 8i6, and though the latter offers DSP software (which is helpful for recording monitoring), the playback latencies have to be set very high. If you go to DAWbench, you’ll see that their Saffire 6 interface was rated worst. In another thread on Gearslutz (which I found after purchasing the Focusrite) the same user found a 2i2 (in the same league as the 8i6) said its performance was so bad it was unratable!

Well I only just saw Steve’s link and will check it out shortly. Since I was just checking a few setting on the buffers I thought I would run here and list my findings. Quite possibly Steve’s link will point out everything I’ve just done is incorrect so bear with me.

I still may be way off track in the process but here’s what I did. I reset the MOTU buffers to 64. I then went and recorded 6 audio tracks and 8 midi tracks. Simple stuff and then went in and changed a few additional Cubase VSTi’s to the midi data which were input by the Casio PX 400 keyboard. After recording and playback I reset the buffers to 512. Again recorded 6 audio tracks and 8 midi tracks. Played both back also resetting the buffers to 64 and again to 512. The only indication of additional stress on the system was through the performance meter real-time gauge. In 512 it barely moved , under the 64 setting it moved approximately 25% of the length of the line with peaks to about 35%.

The playback for each of them was uneventful and sounded perfect regarding the quality of the reproduction…not the talent :smiley: so either I still am not pushing the right buttons or this is not enough recording input to throw things into a spiral.

I am only posting this question here because it has to do with studio hardware and set up.

Hey Steve : since I moved this studio into a larger room in our basement I also had to increase wiring outlets to allow everything to get its own plug. I have noticed that even though I’ve run my powered monitors (Roland D-90) through a power conditioner, there is a small amount of white noise that I get through the speakers particularly, the mouse wheel, makes me a little crazy because it actually shows up on the VU meters in Cubase as well.

Since it is a rainy day and I got out of work early I did nothing more than run a 12gauge extension cord 50’ to the power strip that will now run these two monitors and plugged them into a completely different outlet on a totally different circuit and they are clear as a bell… might you have a suggestion as to what is happening besides ground which I know is good.

ADDED: Received a 46 track midi file from a friend to test. Mostly orchestra stuff but did sound beautiful . at the 64 setting it pushed the performance meter to about 75% real time. Disk doesn’t even show up and the average load is under 15% ( all based on extrapolation of the grid lines on the bar being it appears divided into 25% increments )

Thanks… I think you might have solved it but I will solidly check tomorrow. I have a dimmer ! It is always set to low as these two overhead lamps are just over my work space desk and while 75W bulbs incandescent, I probably have them at 25W? It keeps the glare down and adds a softer feel to the “studio” We have 200 amp service and the breakers are as described. I never thought that the dimmer would do that but you’re right. Completely skipped my thought process and you are 3000 miles away and probably hit it first try!

I have DISCOVERED…that one of the complaints I have read about regarding Cubase, that I never saw or fully understood, is NOW starting to show up on occasion in my system as I begin to really add many more tracks and VST’s than ever before. Nothing major and as intolerable as some have said in their posts. My mixer never had any issues. Tonight the font was so small on the bottom half, there was no way to make it out. Like the warning label on an aspirin bottle. A simple toggle restored it though.

OK, so why would the latency change by going to 96kHz?

The data is clocked through faster, same buffer size, faster clocking… lower latency!

all sounds pretty good but with all of the give and take components of recording and computers, what becomes the down Side of this action?

More stress on the system.

I took the dimmer out this morning and replaced it with a regular decor switch and two 25W bulbs… Still noisy so I guess the problem in some where else but I’m checking everything

Problem is, now that I actually have a small handle on what I can do with midi and Cubase and my own desire to record, I have a greater interest in getting things set up properly instead of just plugged in. It looks better too.

I have a 2tb hard drive installed and 2 - 500 gb drives as well and there is plenty of extra space available… I am going to start experimenting with 96kHz

Right now …nothing is a very important recording. My CPU load tests were me reciting a rhythmic Mary had a little lamb for the vocals over midi tracks and audio tracks even instrument tracks over my jazz rendition of How High the Moon…I am not a singer. What makes it terrible is hearing your own voice on playback . That makes me think its even worse

Hey guys,

I don’t ever go above 44.1kHz here. On my first CD, I tracked at 48K just cause my CI could, but really never could tell a difference, save for the obviously larger file size. Resampling all your rough mixdowns in Wavelab or wherever is a bigtime p.i.t.a., especially if you don’t notice a difference. Now that I’m running on a Focusrite that can run as high as 96kHz, I’ll still never use it. As others say, the only advantage in higher sample rates is for the manufactures to market their interface. A low sample rate will never be why your mix sucks!

I’ve admitted this before . My Cubase skills have stagnated for years and only in the last few months have I had opportunity and abilities to sit, read and experiment. This has significantly enlightened my approach to the multi-tracking I initially planned in 2004. The midi stuff alone has increased the depth of my recordings 3-5 times over what I was doing. Adding ALL of the other flexibilities of Cubase have really opened so many many more options…I’m just on the edge of getting lost again only armed with a lot more information.

I am willing to experiment with this 96kHz recording since most of the tracks I am now recording are dupes of some of my previous stuff and my master stuff is preserved. Not mixed , but all of the raw tracks are there. As I have discovered in the last 3weeks, if I don’t ever try it, I’ll never know what it does. So I am anxious to do a few new things just to see outcomes.

I’ve taken a few audio guitar tracks and added a few process to them that changed them a lot. Some were good, some not. I think I may have gotten to a point in midi and VST’s where I think I need more versatility in the sounds. Out of the box stock collection of sounds in c7 and my VST instrument collection from 2003? is what i am using. I noticed tonight that I don’t care for the selection of brass instrument sounds I have in some of the pieces I have added them to. I did buy a sound set from Sequel 2 …rock, I think. I’ve read about people using different drums, orchestra etc. I see that I may need to look into these.

I believe my best discovery a few days ago is knowing that my computer equipment is running well and can handle the CPU loads and throughput . My drives are big and empty so if an experiment turns out nicely , fine but I am not against wiping it clean. At least my foundation is pretty solid, I can move on from here not feeling that my limits are causing issues.

In the long term, I’ve always just wanted to play with another guitar player who knows my quirks faults and style. By laying down multiple tracks, how lucky could I have gotten to find a guy like that on track 1 and 2 :smiley: now that I can add a few wisps of Hammond b-3 and a line or two of brass, bass, strings or pan flute…gee, it sounds great.

I would love to go into a recording studio and just watch the process the pros use to create their magic.
For me, it has been a great few weeks of of re-focus. All of these options have great potential. Cubase 5 had tutorials on how to start recording and go on from there. I would like to see a tutorial on the things we have discussed here and in other areas of the forum. I prefer that to reading it any day

Thanks to everyone who contributed to this… I have been gaining some rather useful tips knowing these things