Latency issues on an Intel Duo.

Tig · Post by **Tig** » Sat Nov 03, 2007 4:19 am

Can anyone lend me any insight on this. I just bought an new computer with an Intel Core 2 Duo . I have a TC Electronic 24D running into live. I use ASIO 24bit with 128 buffer size. After 2,61ms of driver error comp my overall latency is 11.2. I have seen peaople with similar systems getting much lower latency, why is mine so much higher? I was hoping for somthing <5ms. 64 buffer craps out, so I can't use it.

Computer Specs:
Intel Core 2 Duo E6750 (2.66GHZ)
2 GB Corsair 800mhz RAM
Asus p5n-e

Thanks.

nebulae · Post by **nebulae** » Sat Nov 03, 2007 4:47 am

I run the RME Multiface at 128 samples buffer:

3.63ms input
4.37ms output
0ms correction
7.98 total

If my correction was almost 3ms, then I'd be runnning at about 11ms as well. So I think the math is correct.

Tig · Post by **Tig** » Sat Nov 03, 2007 7:23 am

Well, how are other capable of getting such low latency. My machine is fast yet I can't seem to get under 5ms.

rbmonosylabik · Post by **rbmonosylabik** » Sat Nov 03, 2007 7:52 am

Using a Motu Ultralite on an iMac G5 1.6 GHz:

Buffer Size: 144 Samples
Input: 3.54 ms
Output: 3.56
Driver Error Compensation: 1.57 ms
Overall: 8.67

I run Live at 24 bits/48 kHz.

Raising the sample rate lowers the latency, but increases the CPU use.

nolus · Post by **nolus** » Sat Nov 03, 2007 9:40 am

From what I've managed to find out most of the variability comes from the device driver software and the way it interacts with the firewire port.

rikhyray posted this link in another latency thread that I started.

rikhyray wrote:http://www.centrance.com/about/pr/pr060130.shtml
There is free utility from these people, no idea about their products but the utility let you measure exactly your cards latency, there is also PDF that you might find interesting to read.

They have some driver software that they claim to work optimaly with any firewire interface. unfortunately no demo version is available which makes it a bit of a gamble. and no Vista support which scuppers it for me.

I don't think CPU power is an issue these days.

For reference I'm running a Firepod from the built in FW. PC is Vaio 2Ghz Core 2 Duo.

measured with the LTU tool downloaded from above link:

With buffer size of 88 sample I get 11.2 mS round trip delay at 44.1Khz.

Also have FW410 which puts in good performance at 64 sample buffer size. (was unusable on my old PC below 256 samples)

At 44.1KHz It has 6.8 mS overall latency with a 64 sample buffer and 9.9mS with a 128 sample buffer

I think M-Audio worked some magic in the most recent driver releases, not sure if the 410 still deserves it's evil reputation.

nolus · Post by **nolus** » Sat Nov 03, 2007 11:03 am

Try using the ASIO4ALL driver. It might help and it's free.

ASIO4ALL can operate the Firepod with 64 sample buffers which gets the latency down to 9.6 mS - still worse than FW410 though

mdk · Post by **mdk** » Sat Nov 03, 2007 11:59 am

this is a completely wild guess but i have read about some firewire hardware not liking certain chipsets on the computers firewire interface. (i read about it for the motu 828 mk II which i own)

fwiw, i got a new core 2 duo (e6600) and i can run with a buffer size of 32 and my overall latency is 2.59ms (1.72 / 0.86) thats at 44.1Khz

havent done the error correction thing yet to see how that affects it..

but obviously you should be able to do better. what do TC Electronics say?