Strange latency issue from a newbie

  • Thread starter Thread starter wendy
  • Start date Start date
W

wendy

New member
Hi all! I've been learning a lot here as I work to set up my new Saffire LE to work with Cakewalk on my Vista laptop (Yes, it works great with the downloaded drivers!)
I've got a strange thing going on that maybe someone can explain to me:
I measured the latency using both WDM and ASIO drivers and came up with 13msec with ASIO and 8.5msec with WDM. After setting the record latency adjustments accordingly for each and doing tests with a recorded click track looped back into the input everything lined up just right. So I figured the WDM would be the way to go, right? Because it measured faster? Wrong.... When i tried to plug my guitar into the Saffire and use a few plugin effects in real time I got a noticeable lag that made it really tough to play. So I tried switching to the ASIO driver and lo and behold I could play with no perceptible lag. So of course I'm sticking with the ASIO driver, but I don't understand why it measured slower but performs faster when monitoring in real time with the guitar. Everything's working great now with the ASIO, but
I would just like to have a better grasp on how these things work.
Thanks for any ideas!
 
How did you measure it? You cant really rely on what the driver tells you, you need to measure it with a real world app like the cetrance LTU. I have yet to see a WDM driver that outperforms an ASIO one

http://centrance.com/products/ltu/
 
I measured the latency by recording a click track from a drum machine and then looping that track's output back into the input of a second track and recording it. I then looked at the waveforms and counted the sample difference between the two. From what I understand that gives you an accurate measurement of latency. Maybe I'm missing something??

I did this several times for each driver and got consistent results : the WDM measured faster than the ASIO, but the ASIO actually performs faster when monitoring a VST plugin and live guitar.

I tried this on two different computers, one running XP and one Vista, and the results were similar for both.

?????
 
I'll give it a try, but it only works for ASIO so I'll still have to measure the WDM latency in some other way. It'll still be interesting to see how the Centrance tester compares with my measurements. I'll report back when I get a chance to try it.
Thanks.
 
OK. The LTU measured 16.51msec with the Saffire ASIO driver! The measurement I made in Sonar by looping the output back in and counting the sample difference gave me a latency of 13msec. So I don't have any idea how to explain that. Are these reading generally high for a firewire interface or are they in a range that would be expected?
Just for kicks I installed ASIO4all and it measured 11.43msec w/ the LTU. And it works great while monitoring the guitar w/ VST effects in real time.
Still nothing explains why the WDM driver measured so low but didn't perform well in the real world.

I've just about decided that the best philosophy in the world of msecs is find something that works and use it. Don't sweat the details.
Right now that seems to be ASIO4ALL. It's easy to get obsessive with this stuff!
 
the wdm driver may have been reporting the software latency, one way trip, or some other obscure result
 
I understand what you're saying, but I wasn't relying on what the WDM driver reported. I physically connected out to in and measured the actual latency for both WDM and ASIO drivers. As I said in the first post, the WDM driver measured much faster than the ASIO driver using this method. I know it sounds wrong or like I misread something, but I rechecked this several times. The part I would like to understand is why the ASIO driver actual performs much better (faster)with live monitoring using VST effects.
 
OK - I may have found an answer - from some reading in some other posts I'm gathering that different drivers deal with plugins differently. So even though one driver (in my case the WDM) may measure faster strictly in terms of rountrip in and out, a different driver (in my case the ASIO) may handle plugins faster.
I was working under the (wrong?) assumption that a driver works consistently under different demands.

Anyway, ASIO4All seems consistently faster under all conditions than the supplied SaffireASIO driver so that's what I'm going to work with.
 
You must log in or register to reply here.
Back
Top