SAC interface reports...

[AudioAstronomer]

Here's what I've tried in SAC and the average workable latencies. Main systems are a 3.6ghz p4 on p4p800 w/4gb, and new macbook pro core 2 duo 2.6ghz, 4gb ram and bootcamp 2.0

ASIO unless otherwise noted. 24/48khz

*Alesis io26 2x256
Edirol ua-1000 1x128
Motu Fast Track (usb, pro and ultra have all be tried) 3x128
Motu 896 (not HD) 2x128
RME fireface 1x64, prefer 2x64. 1x128 at 96khz
MOTU Traveller 1x64, 1x128 at 96khz
Motu Ultralight 3x128
M-Audio lightbridge 3x64
Macbook pro ASIO4all 2x256
Presonus Firestudio 1x512


*I've had some minor issues with BSOD with the alesis drivers. I would avoid that interface currently.


The best working interfaces, in my experience, are in bold.

Feel free to post your experiences as well

[mako]

Gigabyte K8NNXP AMD3200 1Gig ram

RME PCI - Digiface - ADA8000
Works at 64x2 but slips buffers - sometimes bad enough to cause a delay effect of around 600ms.

RME 9632 - works well @ 64x2 but slips over time (20 mins).

LynxOne - rock solid @ 64x2

I hope to get back to the RME stuff today for more testing.

mako

[bcorkery]

LynxOne - rock solid @ 64x2

Wow! This is the 2nd mention I've read of LynxOne getting this low. I've never trusted mine to get too low in SAW. Great news!

[DaveS]

Here's what I've tried in SAC and the average workable latencies. Main systems are a 3.6ghz p4 on p4p800 w/4gb, and new macbook pro core 2 duo 2.6ghz, 4gb ram and bootcamp 2.0

ASIO unless otherwise noted. 24/48khz


MOTU Traveller 1x64, 1x128 at 96khz


The best working interfaces, in my experience, are in bold.

Robert - regarding the Traveler - was it tested on the PC or the Macbook? If on the PC - which version driver did you use?


Thanks!

Dave
Recordable Media Services

[Mountain Media]

ASUS B533-E Mobo, Intel 2.6MHz P4, 1Gb RAM, XP Home Edition W/SP1 (old system)

RME DIGI 9652 as Inputs (ADAT x3)

LynxONE as Output (Analog x2)

Using Std MM Dvr

Using Alesis HD24 for Input Device W/2Hrs of 22-Tracks Previously Recorded Live Concert

24-Input Tracks being fed from HD24 (2-Null Input)
All 24 SAC Inputs set with EQ and Comp on each channel (just for testing)

(I'm feeding analog out from HD24 to External Mixer, mixed to 2-track fed into a small mixer feeding Monitors. Two other channels of small mixer feeds Monitors with 2-track from LynxONE - (used to check for slip between HD24 output and SAC output)

Two hours runs without any aural glitching or buffer re-syncs/drops at 2x64
W/less than 26-27% CPU load

Two hours runs w/out glitching or drops at 4x32 at about 48-50% CPU load

Buffers at 1x64, occasional aural glitches (didn't run long enough to check buffer drops)

Sure impresses me, considering a 5-6 Year old system!

I know this wasn't the setup that had meter issues, but all 22 meters seem to keeping perfect sync with audio

Oh, yeah - SAC Pre-1a

Edit Update: 3x32 buffers seems fine without scene swaps - scene swapping intermittently glitches audio at 3x32. Fast Scene swapping does NOT glitch audio at 2x64 or 4x32.

[AudioAstronomer]

Robert - regarding the Traveler - was it tested on the PC or the Macbook? If on the PC - which version driver did you use?


Thanks!

Dave
Recordable Media Services

Both.

I use a firewire pci card with a texas instruments 1394 chipset now, and it greatly improved all my firewire interfaces.

Right now I'm deciding on what to use live once we get SAC rolling, and I'm leaning towards the traveller.

[Bob L]

That's great John... you might want to try ASIO and see if that corrects for the glitches with the F-Key view switching.

Also... try the Realtime priority option... that may make a difference also.

Bob L

[Mountain Media]

That's great John... you might want to try ASIO and see if that corrects for the glitches with the F-Key view switching.

Also... try the Realtime priority option... that may make a difference also.

Bob L

Yeah, I've already set Realtime Priority since posting results and it reduces glitching, but not completely gone. I was trying ASIO, but haven't gotten it setup for RME input and LynxONE output yet. I see input/output on meters, but don't hear nuthin'. It's been a while since I've used RME inputs, 'cause I usually direct copy HD24 files to HD on SAWStudio, so I just gotta take a few minutes and play with ASIO setup.

I did find, based on response of Solo'ing a channel, at 1x64 the posted setup is off by about 1-2 seconds after about 30 minutes. No problems at 2x64.

And though I'm not using the path through a mic/pre-amp/A-D as I would if I were a vocalist listening through headphones, I don't hear any latency at 2x64.

[Bob L]

There is no question that if the buffers are set too low, you will get drift because each time a buffer is glitched a blank buffer gets inserted by the driver.

I am actually searching for a way to detect this, but since the system does not report an error or anything out of the ordinary, I have yet to be able to detect when it happens... I am working on a concept though to detect and self correct for glitched buffers... at least as far as the drift is concerned... obviously I can't fix the gitched buffer... but it would be nice to be able to absorb back the extra buffer latency added in to keep the whole system running in sync forever... regardless.

Bob L

[Mountain Media]

There is no question that if the buffers are set too low, you will get drift because each time a buffer is glitched a blank buffer gets inserted by the driver.

I am actually searching for a way to detect this, but since the system does not report an error or anything out of the ordinary, I have yet to be able to detect when it happens... I am working on a concept though to detect and self correct for glitched buffers... at least as far as the drift is concerned... obviously I can't fix the gitched buffer... but it would be nice to be able to absorb back the extra buffer latency added in to keep the whole system running in sync forever... regardless.

Bob L

Great thoughts, Bob. Especially at these low buffer settings, an audience would probably seldom hear a dropped buffer, or the glitches, which will happen anyway, but to know they will stay in sync would be awsome!!

[Craig Allen]

There is no question that if the buffers are set too low, you will get drift because each time a buffer is glitched a blank buffer gets inserted by the driver.

I am actually searching for a way to detect this, but since the system does not report an error or anything out of the ordinary, I have yet to be able to detect when it happens... I am working on a concept though to detect and self correct for glitched buffers... at least as far as the drift is concerned... obviously I can't fix the gitched buffer... but it would be nice to be able to absorb back the extra buffer latency added in to keep the whole system running in sync forever... regardless.

Bob L

That would definately be a breakthrough. My biggest fear is that buffers will get dropped and the audio will get 'behind'. I had that happen with Saw once or twice when using it to mix live.

One thought on that, would it be posible to cycle the engine every once in a while. Maybe do it super-quick (a sample or two) so it wouldn't be noticed? Because starting and stopping SAW fixed the problem back when I had it.

[Bob L]

Well... I am going to create a reset command for the next update that will cycle the engine as fast as possible manually... that will have to suffice until I can come up with a reliable way to detect when it needs to be corrected automatically.

Bob L

[Leadfoot]

the 0.1a seems perfectly stable on my machine now. meters and audio are instant and solid at 2x64. i even had it running at 2x32 and it seemed fine.
it may have drifted ever so slightly at 32 over time, but a quick stop and restart of the engine fixes that. i'm quite impressed, no pops when hitting buttons like solo, mutes, even changing in and out assigns, totally clean.
the system/audio overall is amazingly clear, i would love to use it at the gig i'm doing tomorrow night, i just haven't had time to get myself set up properly yet. very soon though, got a lot of shows lined up. thanks again bob!

tony

p4/1gb-ram/xpsp2/matrox-g450
2x ada8000
rme hdsp9652/asio driver 3.057

[Bob L]

Good news.

Thanks,

Bob L

[AudioAstronomer]

Updated some of my experience.

[mako]

I'm now using a dual core during rehearsals and find it does slip a little over time (Asus P5K Pro, C2D 6300 with RME Digiface/PCI) but overall is clean and rock solid at 32x2.
1 vocal channel (EQ, Comp) 1 guitar channel, 1 backing channel = 0% load

Updated info - Yesterday I added my RME 9632 to the mix (previously, in my 64 Athlon, they didn't work well together, the PCI/Digiface suffered badly) now it runs just as well as stated above - I'm very pleased.

Asus P5K Pro / XP Pro
C2D 6300
RME Digiface/PCI -ADA8000
2gig memory
ASIO 32x2

Great fun

mako

[Bob L]

Read my thread about the dual cpu issues... the slippage can be contained now... I am working on a permanent fix.

Bob L

[mako]

Thanks Bob - all the best.

mako

[Iain Westland]

bit dissapointed!!!

i have two 9632 cards and decided to try and push them...

opened 72 inpots (mono) added the console eq with high and low filters on, plus gate and compressor on all channels. opened all the auxes on each channel (thats all 72 - did i mention that before)

47% load

ok, do all of that on the monitors then

monitor 1 - yep

monitor 2 - yep

bugger it, add 20 instances of the eq to the first 20 foh chanels

1 buffer 64 samples, running 2 hours 98% MT!!!

should it not have crashed before i managed to get this far? thats a bit too stable aint it mate? come on

something did catch me out though ... the mutes, solos and key listen are global, solo'd out a feed when playing with my headphones on monitor 1 and nearly blew my monitors ... that could be an expensive mistake fo somebody to make

[Bob L]

I'm glad you are seeing the performance possibilities... wait till this new engine that will properly utilize the dual cpus.

There is a PRE SOLO bus level trim... or use the Pst Solo bus option... in fact there will soon be a popup fader for each solo console section that you can adjust on the fly.

Features are all lined up... but the engine stability and dual cpu issues are first in line so everyone can start using it in real life setups without being afraid of slipping or crashing and burning the show.

Bob L