General Question - Latency through XITE on auto DSP alloc

A place to talk about whatever Scope music/gear related stuff you want.

Moderators: valis, garyb

Post Reply
User avatar
grappa
Posts: 223
Joined: Mon Aug 23, 2004 4:00 pm
Location: In a hard town by the sea....

General Question - Latency through XITE on auto DSP alloc

Post by grappa »

Hi all,

Before I waste any of my time is it possible to give some insight on the consistency of latency through Scope/XITE. I don't do any DSP allocation work in my projects and just let everything load as-is - never had any issues with this or needed to do anything as projects are relatively lightweight in comparison to XITE power. This being the case is there potential for a project to load and the latency through scope to change based on where a device is placed on DSP? If so can the device be manually allocated to a DSP and will this result in a consistent result on project load after the project is saved?

Just be interested in some insights..

Thanks,

Si
User avatar
garyb
Moderator
Posts: 23246
Joined: Sun Apr 15, 2001 4:00 pm
Location: ghetto by the sea

Re: General Question - Latency through XITE on auto DSP all

Post by garyb »

generally, most devices will add a few samples of latency, about the same as any hardware digital device.
generally, it is pretty consistent.

if you are VERY picky about such things, then you would need to keep an eye out for changes due to differences in how things are loaded. if you use any external hardware in your mixes/production, then the problem will be minimized. everything creates some latency, including runs of wire and hardware circuits. only something that is 100% in the box, and offline processed, can be corrected to the point of 0 degrees of phase shift on EVERYTHING.
fra77x2
Posts: 414
Joined: Sun May 03, 2015 3:23 pm

Re: General Question - Latency through XITE on auto DSP all

Post by fra77x2 »

Devices that are loaded and interconnected on the same dsp add 0 samples of latency.

Connections between dsps, add 5 samples of latency.

Automatic dsp distribution may not load exactly the same and so the latency may differ by a tiny amount of samples.

Manual dsp allocation always loads exactly the same.

Devices that have look ahead functionality will add latency to the path.
User avatar
grappa
Posts: 223
Joined: Mon Aug 23, 2004 4:00 pm
Location: In a hard town by the sea....

Re: General Question - Latency through XITE on auto DSP all

Post by grappa »

Thanks guys.

The end to end latency isn't a concern for me as once ascertained is fine. This potential issues for me would be to have a circumstance where a project would load with dynamic DSP allocation that significantly altered the latency each time in a random manner.

It would appear that if I manually allocate and save the project I shouldn't have anything change?

Regards,

Simon
fra77x2
Posts: 414
Joined: Sun May 03, 2015 3:23 pm

Re: General Question - Latency through XITE on auto DSP all

Post by fra77x2 »

Yes manual allocation is always consistent.
User avatar
Sounddesigner
Posts: 1063
Joined: Sat Jun 02, 2007 11:06 pm

Re: General Question - Latency through XITE on auto DSP all

Post by Sounddesigner »

fra77x2 wrote:
Connections between dsps, add 5 samples of latency.
Thanks for that info! 5 samples seem to be higher than the old pci cards but still very low considering overall latency and the fact it isn't always added like when loading multiple zero-latency effects on the same dsp (plus I work at 96khz wich makes it even shorter time-wise and makes latency not even worth mentioning). Do you know if 5 samples is from old dsp's to new ones in XITE-1 or just with new dsp's to other new ones? I was curious about this since the last debate on it (latency is not a problem for me I'm just curious about some geeky things).
User avatar
astroman
Posts: 8410
Joined: Fri Feb 08, 2002 4:00 pm
Location: Germany

Re: General Question - Latency through XITE on auto DSP all

Post by astroman »

such latency figures only become a problem if critical sounds are layered
on hi-hats or the click of a parallel processed kick drum 5 sample can be a huge difference
and of course if a complex process is spread over multiple DSPs
(some legacy devices suffer from this, but today developers generally check it in advance)
5 samples seems similiar to the old PCI cards, at least according to what I've measured in special cases
in general multitracking it can be ignored

cheers, Tom
Post Reply