This can indeed be a dilemma. Here's my view on it. Small buffer sizes give you the benefit of low latency at a cost of CPU usage. Large buffers take less CPU, but creates longer latency. It's a good idea to have two different presets for buffer sizes: one with small sizes for monitoring DSP effects and one with larger buffer sizes for mixing. I found my best settings by starting with very large buffers, then lowering them step by step until things started to get unstable. Then, I increased the buffer size a bit and that became my DSP monitoring preset. Then, I picked a fairly arbitrary larger buffer setting to use as my mixing preset. In monitoring mode, I get 3 msec latency (at 44.1) and in mixing mode about 20. If I remember correctly, you're using Cubase/Nuendo. So am I, so if you'd like, I could tell you exactly how I set up mine.

If I remember correctly, you're using Cubase/Nuendo. So am I, so if you'd like, I could tell you exactly how I set up mine.

I would love to hear this please.

This is how I did it with Cubase SX 2.0 and an M-Audio Delta-66 soundcard, but it should be similar for other soundcards or for other programs.

Open Cubase or Nuendo’s ASIO Multimedia Setup utility (wpsetup.exe). Click ‘Advanced Options’.

Open your hardware’s configuration utility and set the DMA buffer size to its largest setting.

Start with your sync reference set to either ‘Sample Position – Input’ or ‘Sample Position – Output’. If you lose sync fairly early during the process, switch to the other one. If both are losing sync early, switch to ‘DMA Block’ as a last resort. FWIW, ‘Sample Position – Input’ is the one that worked for me.

I used five buffers in the ASIO Multimedia Setup because anything less caused premature loss of sync. This may be different for you, but I’d recommend starting with five buffers.

And now for the Algorithm:

1. Select an output port. Click ‘Detect Buffer Size’. When it returns a value, set all of the output ports to this same amount of samples.

2. Repeat step one for the input ports.

3. Click ‘Run Simulation’.

If the simulation passes, lower the DMA buffer size of your hardware and repeat steps 1-3. If the simulation fails, revert back to the values from the last iteration. This setting should be your lowest safe latency.

This algorithm has allowed me to achieve a latency of 3 msec with no loss of sync. I hope this works as well for all of you as it has for me.

Thanks David, this is some good info.

No prob, Dab :)

So did anyone use my advice? How is it working?

Nice Job David! 8-)

I use a similar approach but I like your way better. I'm making a few changes to my DAW but I'll try it as soon as she's back up.

I have'nt tried your suggested tune-up yet David (job has me traveling for the time being). But with what you stated, sounds like it will do the trick. Much appreciation dude.

I know this might be dumb but i dont know what DMA or DAW means and cant find the DMA to change it to the highest setting. Everything else though, I think I got and I want to try this out!

DAW = Digital Audio Workstation, i.e., your music and recording computer

DMA = Direct Memory Access

You should have some kind of a software mixer that came with your soundcard. Open it up and look around for a buffer size adjustment field.

Bringin up an older post here...

Hey David, when you say in your instructions under the algorithm part to select an output port (or input) and run simulation, do you mean you only have a check in the box next to the port you are testing, or that you are only highlighting one port and have all the ports checked off?

Just curious.