Latency Question - What is a real key to small latency...

So this is for all the super smart DAW techs (this forum has lots (and thank you))....

Read the Cakewalk discussion about latency, drivers, etc..... and my question comes down to this.....

For the best (shortest) latency, what is the main driving factor (or is there one?) for reducing latency:
Using a 'fast' Sound card?
Using WDM or ASIO drivers?
Fast Computer Processor...

Sorry, this question is beginning to seem dumb as I type it....

berlymahn


So this is for all the super smart DAW techs (this forum has lots (and thank you))....

Read the Cakewalk discussion about latency, drivers, etc..... and my question comes down to this.....

For the best (shortest) latency, what is the main driving factor (or is there one?) for reducing latency:
Using a 'fast' Sound card?
Using WDM or ASIO drivers?
Fast Computer Processor...

Sorry, this question is beginning to seem dumb as I type it....

All of the above.

berlymahn


So this is for all the super smart DAW techs (this forum has lots (and thank you))....

Read the Cakewalk discussion about latency, drivers, etc..... and my question comes down to this.....

For the best (shortest) latency, what is the main driving factor (or is there one?) for reducing latency:
Using a 'fast' Sound card?
Using WDM or ASIO drivers?
Fast Computer Processor...

Sorry, this question is beginning to seem dumb as I type it....
 

A well-written driver (asio or wdm) for your audio interface is obviously important, but a fast processor comes into play a lot when you're input monitoring with effects or playing softsynths, especially if you don't want to constantly bounce down. I'm still slightly comfortable with my Core2 believe it or not, and I'm the kind of person who normally builds a new machine every two years at most.

berlymahn


Wow.

Great responses...

Interesting.... I know about audio latency (and have found a good site for audio cards/interfaces, but the main killer I am dealing with is midi-latency.....thinking I need to push my midi through a faster processor.....  My machine downstairs is sluggishly slow on a rebuilt P4 motherboard....  Time to upgrade...

You've answered your own question.

Wow.

Great responses...

Interesting.... I know about audio latency (and have found a good site for audio cards/interfaces, but the main killer I am dealing with is midi-latency.....thinking I need to push my midi through a faster processor.....  My machine downstairs is sluggishly slow on a rebuilt P4 motherboard....  Time to upgrade...

I believe MIDI latency is helped with a faster processor. That is, since the MIDI latency is affected by the size of the hardware samples per buffer size. It's sort of commonly known that the higher the hardware buffer size, the greater the MIDI latency, and the lower the hardware buffer size, the lower the MIDI latency. And in my experience, a faster processor speed has helped me to obtain stable ASIO performance at lower latency settings. But like I said this is also dependent on good drivers for the audio interface, and in my opinion, also to some extent dependent on the other things I mentioned in my previous post in this thread (motherboard and RAM speed and quality.)

Anyway, that's how I see it. :) In other words, stable performance at the lowest latency settings depends on how low and stable the system's DPC latency is, and I think the things I mentioned in my previous post plays some role in the state of the DPC latency of a system.

As a general rule, PCI interface cards and Firewire will offer much better latency for you than USB interfaces.
This is especially true with higher resolutions (ie ... 24/96)

In fact, you may find most USB interfaces choking once you try to move too far beyond 16/44.1

Be willing to get the best interface you can as this really becomes the heart of your system in many aspects.

And as mentioned above, ASIO with good drivers.

mikespitzer


As a general rule, PCI interface cards and Firewire will offer much better latency for you than USB interfaces.
This is especially true with higher resolutions (ie ... 24/96)

In fact, you may find most USB interfaces choking once you try to move too far beyond 16/44.1

Be willing to get the best interface you can as this really becomes the heart of your system in many aspects.

And as mentioned above, ASIO with good drivers.

Hi Mike. Did you mean PCI or PCIe? Reason I ask is, I have the PCI version of the EMU 0404 and have not had very good luck with latency. I even tried to put it in my i5 rig with the new Windows 7 x64 drivers and I wasn't impressed. I'm guessing it's just because it's so old?

I'm able to run my M-Audio Fast Track Ultra at 96/24 at a software reported round trip latency of 6.2ms @ 128 samples. I'm very happy with that. If things start to bog down I just freeze some tracks and all is ok.

Bub

Jim Roseberry


At the risk of hi-jacking this thread...
 
Bub, did you try the clean Win7x64 install (to see if X1 worked better)?

Hi Jim,

Yes I did. I didn't want to post a reply in the other thread until I tested it out thoroughly. I will say, some of the oddball things like the metronome playback and record buttons on the transport not working and my audio clips becoming corrupted seem to be fixed now. Also my track interleave settings stay where I set them now. They would keep reverting back to mono, even if I set them to stereo and saved the project before I did the clean install.

Other things have cropped up, but I think they will be fixed with the X1b patch. I also found several bugs in the Loop Construction View and reported them. Tech support was able to verify them and sent it on to be fixed. Can't say for sure if that will be in the X1b patch though.

In a nutshell, I would have to say the clean install fixed the problems that seemed to be unique to my PC and I'm glad I did it.

Thanks,

Bub.

Ok, slartabartfast, I can accept that.

However, most common folk are not geeks concerning this MIDI thing. And most common folk are referring to the latency they hear when trying to play a softsynth in real time with a MIDI keyboard controller, when they say they want low MIDI latency. And the SONAR manual references this understanding of the common folk when it touches on this MIDI latency matter, as demonstrated below in a excerpt from the SONAR Help file:

"Note: WDM or ASIO drivers do not improve performance when you play back recorded MIDI data—the improvement comes only when you play a soft synth in real time from an external MIDI controller or keyboard.
 
You can play the soft synth in real time from a MIDI controller or keyboard. To avoid excessive latency, your sound card must be using a WDM or ASIO driver. Also, you must set mixing latency to the lowest achievable level (probably less than 10 msec.), which you do by adjusting the Buffer Size slider in Edit > Preferences > Audio - Driver Settings."


That's all I'm saying. From the OP's question, I was very sure he was referring to the MIDI latency experienced when playing a softsynth in real time with the hardware buffer not being set low enough or when the audio drivers for the audio interface aren't good enough to run SONAR at the needed lower latency setting for good performance when playing a softsynth in real time. :) That's all I'm saying. And this is true whether one has a technical understanding of MIDI latency or not. The latency experienced when playing a softsynth in real time is what most common folk understand as MIDI latency, whether technically correct or not. :) 

slartabartfast


We have a chronic problem on this forum with the term "MIDI latency." Logically that should mean the amount of delay introduced by only the MIDI limb of the MIDI controller to MIDI interface to processor/software execution to softsynth to effects execution to DA convertor loop. So when John says MIDI latency is insignificant, he is following that definition i. e. the delay involved in getting a midi message from a controller into the box. What many if not most people on the forum mean by "MIDI latency" seems to be the delay over that entire loop. Most of that loop delay is introduced in the box, long after MIDI data has been handed off to digital processing. The delay most people are concerned about is not MIDI related. It is related to mathematical processing in the box and deliberately programmed buffer delays (especially look-ahead buffers).

Thank you so much for your great post.. 

There is such a thing as MIDI latency but what is being discussed here is not MIDI latency.

Basically people encounter latency in two forms and are confused by it. One is the audio latency we get when we monitor signals via our DAW. The other is when we use soft synth. Both of these latencies have at their cause the audio driver and its settings. MIDI has such a very low data impact on a system that it has no noticeable latency. If you use a hardware synth and it is controlled by a controller that runs through your DAW via MIDI hitting a key on the controller will be almost instant, no latency. Even though the MIDI data is going from the controller through the DAW and on to the synth. When we use any processing within the computer and then use an audio interface latency is introduced. The better the driver the better the latency. When you include the internal processing of a soft synth you increase the need for a faster computer to help lower the time it takes to process that MIDI data.  But it isn't the MIDI data that is the cause of the delay. Its the audio processing. Therefore it is incorrect to call it MIDI latency which does exist. Call it latency.

i find it interesting that sonars latency varies automatically regardless of what it is set to in the asio panel.  when i open a new project, soft synths play with NO noticable latency.  as a project grows, plugins add up, synths add up, the latency inreases to account for the load.  if i freeze some synths and turn of plugins in the master bus etc, the low latency returns.  years ago, the latency was manually controlled bythe asio or options panel.. 
i use a profire 2626 and octane. 

Yeah - I can see that with the addition of non-frozen soft syth tracks, effects, etc (requiring processing time for Sonar (or any DAW) that the latency would rise.  Read something about that last night.

Jim

mikespitzer
In fact, you may find most USB interfaces choking once you try to move too far beyond 16/44.1
 
Jim Roseberry
Typical USB audio interface offer nowhere near this performance.

It used to be like this, but I think we have a different story now with the latest USB 2.0 interfaces. I always wanted a PCIe card + a breakout box, but could not find anything that would work with my new mobo, an EVGA SR2. This is the fastes and most powerful PC you can build.
I got myself a MOTU 828 mk3 USB / Firewire hybrid.
I record using Sampling rate 88200, with 192 samples, 16 bit, which gives 2.2 ms latency. I can bring latency down to 1.5 ms by lowering the samples, but then I get a few crackles now and then.
USB 2.0 High Speed is faster than Firewire. Pity they did not jump to USB 3.0 immediately, since I have that too on this mobo. 
 
I was against USB all thew time, and never thought I could use it. But since my first purchase, the ESI 1010e breakout box with its corresponding PCIe card, did not work (my PC froze) I had to go for the MOTU, having fought problems almost one month. And I don't regret it. For me it is a very good solution. And using Sonar X1 for multi media in general (not pure music production), I find 88200 / 16 good enough. If I am overlooking something here, please advice me. I always want the best result, of course.
Ingvar

I record using Sampling rate 88200, with 192 samples, 16 bit, which gives 2.2 ms latency.

 
FWIW, You're talking about one-way (playback) latency... not total round-trip latency
The MOTU FW units are good (I have two of them)... and the new mkIII "hybrid" USB/Firewire units are also good.
The RME USB units are particularly nice... as they let you go down to a 48-sample ASIO buffer size (and they use a small hidden safety buffer).  If your machine can keep up with the load, that translates to incredibly low round-trip latency.