Low Latency SoundCard Recommendations

[tomcat]

I am looking to get my latency down.......so I am selling my presonus firestudio. I am however looking to keep my presonus digimax fs for its 8 preamps and adat i/o. I will need a total of 16 ins so I will either purchase another digimax fs or (?suggestions?). I am not very familiar with adat i/o so I am looking for suggestions for a soundcard that would meet my needs and give me the 16 adat i/o that would be needed. I would prefer something that is stable and is known for its low latency(rme or something like it). It also needs to be PCI-e. I have a xeon dual quad-core processor.
Thanks for all of the help on this forum! I am learning more than I have in quite some time and always enjoy reading all of the great post on this forum.

[Craig Allen]

RME. There are a lot of options, but all are rock solid. This is the page for their PCI-e cards -
http://www.rme-audio.de/en_products_overview_pci_express.php

[DominicPerry]

The Digimax FS has 96KHz capability using S/MUX meaning that you need 2xADAT to support 8 channels at 96KHz. Do you want 16 channels of audio at 96KHz? If so, and you want a PCIe card, I recommend the RME RayDAT which will allow 32 channels in + 32 channels out at 48KHz and 16 + 16 at 96KHz. It will also be low latency (depending upon your machine) and can use 32 sample buffers.
Historically, I think we have always found RME interfaces to provide the lowest latency (and best stability) here in SAW land.

There are other good products including Sydec and M-Audio and Lynx, but my personal recommendation is RME.

Dominic

[DominicPerry]

I doubt whether there is any audible reason to run FoH at 96KHz. I would expect 44.1 or 48 to be fine. And it will use lower CPU and you can get away with half the number of ADAT ports.

Dominic

[Mark Edwards]

I also recommend RME. Many years ago I bought an RME Hammerfall DIGI 9652 on the advice of fellow Saw users. It has been a solid performer to this day. It is now a "legacy" card, but the support and driver updates have always been stellar. A new RME RayDat would be the only reason I would consider parting with it. 32 ch of lightpipe I/O, and even lower latency is very tempting.:D Good luck.
Mark

[Bob L]

The RME cards running with the ASIO driver should get you down to 1 x 64 or 2 x 64 very stable... many have even been getting 2x32 in SAC with those cards.

The DigiMax FS should do great... or you can add more chans for less money with the Behringer ADA8000 units... some will argue about the quality difference... you will have to decide... no argument about the price difference... less than 1/3 the cost.

Bob L

[IraSeigel]

The Focusrite OctoPre LE is also in the same price range as your Digimax FS. Might be higher quality, but I'm not familiar with their Windows drivers.

http://www.gigasonic.com/focusrite-octopre-LE-bstock.html

Ira

[DominicPerry]

I had a Focusrite Sapphire io26 (which is Firewire) and the drivers were rubbish. But the Octopre doesn't need drivers - it doesn't interface with the PC directly. Come to think of it, the Sapphire had poor mic pres too.

Dominic

[HapHazzard]

I've been looking high and low for low latency and overall performance
and I figure that I'll spend the money because of what I've experienced with my present card, a RME HDSP9652. My choice is the RME RayDAT (http://www.rme-audio.de/en_products_hdspe_raydat.php) because I will get to those 1 X 64 and 2 X 64 without any problems. It maybe over kill with channels but 32 channels in, can just cut most of the jobs I do live.

It all centers around your sound card.:D

Hap

[mycorn]

i got a steal on 2 older rme9652's

absolutely no complaints

and you can use the firestudio in stand alone
"adat mic pre" mode if you dont mind running 48k
[it works fine live]

if you set the mixer output patch to send mic 1 to adat1,1
2 to 1,2 etc in the matrix it will pass strait thru like any 8x adat mic pre

it remembers this patch if powered up not hooked to a firewire

the "problem" is it defaults to 48k
rather than remembering 44.1, 88.2, 96 etc.

i use mine like this now and let it clock the system
[currently thru adat but will switch to WC
in the next couple weeks when the new box is built]

fwiw
mike

[Steve L]

I know Bob doesn't like the firewire stuff but recently had a system using an M'Audio Profire 2626 and 2 ADA8000's (24 chan) run at 2X64 for 4+ hours of live SAC and SS recording with nary a glitch!..... The recording was awesome and the comments on the 10 monitor mixes were all positive to say the least. This was on a stock Win XP Pro setup with SP2 and no firewire tweaks. I routinely use this as a test setup to show customers on my laptop also.
Steve L.
Audio-Video Electronics
Mountain View Recording

[Warren]

I've been looking high and low for low latency and overall performance
and I figure that I'll spend the money because of what I've experienced with my present card, a RME HDSP9652. My choice is the RME RayDAT (http://www.rme-audio.de/en_products_hdspe_raydat.php) because I will get to those 1 X 64 and 2 X 64 without any problems. It maybe over kill with channels but 32 channels in, can just cut most of the jobs I do live.

It all centers around your sound card.:D

Hap

Hap:

You sound very disgusted with the HDSP9652 and its latency.
I know others here are using it as well, what are you concerns?

I have had latency issues but had not tracked it to the HDSP9652,
I have been massaging my system to the point where there is only the audio card left to replace. I certainly would expect more from a dual Xeon 3.2 Gig w/4 Gig of ram.
I tried DPC Latency Checker and it shows flat line perfect, tried all the tweaks out there.
Disabled parallel ports, Com, USB ports, onboard audio. I have even used micro versions of XP. I seem to only get 24 channels in & 4 out Native plugs only, 3 at best at 4x64. I use the latest ASIO drivers from RME. I use very little EQ, Comp and Gating. It acts stable solid 18's to 24's (CPU CYCLES), but I still drop buffers, sometimes even just by moving the mouse which is not USB.

At times I have felt like the final adjustment for it, 5 shots of high speed lead from my shotgun. BUT! A cooler head has prevailed.

I will take any advice offered on this one.

Thanks for letting me vent! And sorry for the hijack of the thread.

Chris

[Bob L]

You may have an IRQ sharing problem with a network card or something similar on the motherboard.

Check your irqs in system information and move the rme to a different slot and see what happens... try to get it not sharing with anything.

Try the ASIO drivers and set SAC to Realtime priority.

Bob L

[Warren]

You may have an IRQ sharing problem with a network card or something similar on the motherboard.

Check your irqs in system information and move the rme to a different slot and see what happens... try to get it not sharing with anything.

Try the ASIO drivers and set SAC to Realtime priority.

Bob L

Thanks Bob:
I did set Affinity's both CPU's, on one cpu then the other no help, also I set Realtime priority. No change.

I will check the IRQ sharing posibility, if no go, I will try to disable the on board NIC (fingers crossed).

Thanks for the quick reply!

Chris

[tomcat]

I would love to have the RME RayDat, but it looks like $849 is about the lowest I can find.......anybody found it cheaper. If not what is another RME low latency card that would fit my 16 channel adat needs! Thanks

[Warren]

I would love to have the RME RayDat, but it looks like $849 is about the lowest I can find.......anybody found it cheaper. If not what is another RME low latency card that would fit my 16 channel adat needs! Thanks

RME9632 will do 16 in and 16 out

[Leadfoot]

I will try to disable the on board NIC (fingers crossed).

Chris

The on board NIC was the biggest offender in my case. I always had it turned off in the bios until I wanted to use my wifi setup for sac with a laptop. When I activated the NIC I started having problems. I checked the Irq's and they were sharing the same as my RME hdsp9652. Haven't tried changing slots yet, that should do it.. then it will probably share with something else.
After all these years you'd think we would have better control of our hardware resources.

Tony

[Bob L]

Tomcat... look around the web for an old RME Hammerfall Lite... (digi9636)... you may find one for very cheap... it will do 16 ins and outs... as Warren mentioned.

Bob L

[HapHazzard]

I would love to have the RME RayDat, but it looks like $849 is about the lowest I can find.......anybody found it cheaper. Nope not yet.
Hap

[HapHazzard]

Hap:

You sound very disgusted with the HDSP9652 and its latency.
I know others here are using it as well, what are you concerns?

Chris

Very Happy with my old 9652. I'll be doing a major build real soon and a upgrade to full SS from my SSL. Might as well get the PCIe so I can at least have a few years at .7 ms with SAW/SAC combo.
Hap

[Warren]

You may have an IRQ sharing problem with a network card or something similar on the motherboard.

Check your irqs in system information and move the rme to a different slot and see what happens... try to get it not sharing with anything.

Try the ASIO drivers and set SAC to Realtime priority.

Bob L

Hey Bob:
Thanks for pointing me in the right direction.
I had found that two non-system processes other than SAC where set to RealTime priorty, and thus stepping on SAC when they made a request of the processor. Fixed

We are now "In like Flint" and smooth as silk!

Very Nice, Very Stable!

Thank you so much! :D :D :D

Chris

[Bob L]

Good news.

Now enjoy the world of virtual consoles. :)

Bob L

[IraSeigel]

Hey Bob:
...
I had found that two non-system processes other than SAC where set to RealTime priorty, and thus stepping on SAC when they made a request of the processor. Fixed

Chris

Hi Chris,
How did you find this out? Task Manager, or another utility?

Ira

[Warren]

Hi Chris,
How did you find this out? Task Manager, or another utility?

Ira

Hi Ira:

I checked it using Task Manager (TM). Just monitored the affinity and priority of all processes that show up in TM. Then used a utility called process.exe off the web, that allows you to control those funtions as well as a few others. Then I use process.exe in a batch file to force affinity & priorty of the processes the way I need them. I had also had to disable USB mouse and keyboard and only use PS2 type, Still at this point I feel I don't get to use all the VST plugs I would like, so I stay mostly Native (Makes me think something else is a foot here, hardware wise) Still pooking at that issue but where up and running stable enough to get the job done.

Chris

[Naturally Digital]

Warren, you mentioned you are running a dual xeon machine...

Which motherboard and/or computer is it?

[Warren]

Warren, you mentioned you are running a dual xeon machine...

Which motherboard and/or computer is it?

Hi Dave:
Thanks for asking
Yes

Its a IWILL DH800 MB with DUAL 3.2 G XEON (Nocona's) Processors, 4Gig ram DDR400 PC3200. MATROX P750 3 head video card 64 Meg me think, Onboard NIC and Audio. (Disabled)

Any way’s it runs SAC smooth enough with 24 channels of input and FOH and 3 Monitor consoles as outs (which are setup as Pre fader from FOH No effects at all on monitor consoles), as long as Native echo and Verb are the only effects running FOH and the buffers are set to 4x64. Its great no dropped output buffers. Then if you add EQ, GATES, COMP, even a second instance of the native reverb to FOH or another Monitor console it starts to pop slowly then the processor will runaway up to 99% as its dropped hundreds of output buffers, input buffers is always fine. When a second instance of native reverb is added it sounds like a broken ten cent flanger.:eek:

It just seems strange to me, being a dual Xeon with HP disabled, all the tweaks and all. Come to think of it, it’s kind of got me tweaked as well. But hey up to that point it sounds wonderful and I am not doing all the work I used too, I guess I just thought it should do more, I am sure its my confuzer and not at all SAC.

If you have any ideas I sure would be glad to hear it!

Thanks
Chris

[IraSeigel]

... I am sure its my confuzer and not at all SAC.

If you have any ideas I sure would be glad to hear it!

Thanks
Chris

Ah, I think that's your problem. The Confuzer is a VSTi. You've probably patched it in the wrong place. Use it in Pre-Patch, route it to the talkback channel, and then everyone will be Confuzed.

Ira

[Warren]

Ah, I think that's your problem. The Confuzer is a VSTi. You've probably patched it in the wrong place. Use it in Pre-Patch, route it to the talkback channel, and then everyone will be Confuzed.

Ira

Hey Thanks for that indepth thought provoked responce to my concerns.
One thing

Huh?

Im more confuzered than ever now.

Thanks for that.:p
Chris

[Bob L]

Chris... the runaway cpu to 99% is definitely a plugin with the floating point problem not being properly handled in the code.

There is a bug in the floating point processor that will logarithmically increase its calculation time on infinitely small data values... so if a plug is placed in SAC that has this problem, during silence the plug will start eating the cpu as you describe.

This is not a load thing... it is a problem in the code of the plug not paying attention to this floating point issue.

put the plugs in one at a time till you find which one is causing the problem... and get rid of it.

You can use multiple verbs of mine or others that have handled this correctly... you should have more than enough power to handle many more channels and more monitor mixes.

Bob L

[Warren]

Oh I also wanted to say:

Tomcat:

I hope you have gotten the answer to question?

I am sorry that it seems as your thread has been hi-jacked at this point. :o mostly by me. :o

It seemed as though it had traveled down a path, toward a direction of concerns I also was having. Sorry!

I am sure we will be on track soon.

Chris

[Warren]

Chris... the runaway cpu to 99% is definitely a plugin with the floating point problem not being properly handled in the code.

There is a bug in the floating point processor that will logarithmically increase its calculation time on infinitely small data values... so if a plug is placed in SAC that has this problem, during silence the plug will start eating the cpu as you describe.

This is not a load thing... it is a problem in the code of the plug not paying attention to this floating point issue.

put the plugs in one at a time till you find which one is causing the problem... and get rid of it.

You can use multiple verbs of mine or others that have handled this correctly... you should have more than enough power to handle many more channels and more monitor mixes.


Bob L

Thanks Bob:
It sounds like what I am seeing but it even happens when native plugs only are used, it is much more stable, with less cpu runaway than if had non-Native plugs been used, but its still there. Every time I arm another Monitor console (Prefader NO EFFECTS) the load goes up 10%. Second instance of your verb gets added at any point the the thing will sound just like a real cheap flanger, remove the second verb and it goes away.

On a side note the presets on the second instance of the reverb sometime defaults back to default settings when you try to go to a different room sound.

The things that did make a big difference and are Much better since I :
(1) Dumped USB devices (keyboard and mouse), and went PS2.
(2) forced affinity to stay both processors and set SAC/SAW priorty to realtime and all other processes to normal.
(3) Went into CMOS (BIOS) and set memory timings to the ram specs as well as setup the remaining BIOS settings properly.

Still I am scratching a bald spot on the back of my head trying to figure this one out.

Looking forward to your responces.
Thanks for shooting back some ideas.

Chris

[Naturally Digital]

Have you messed around with PCI latency settings? That MB has a pretty good bios for overclocking, you may have control of the PCI latency in the bios. If not you would need a utility for it (such as doubledawg or similar).

Have you tried using 2x128 or some other combination, just to see if things operate differently? I find some drivers/cards work better with one block/buffer size than others.

On that generation of machine (875 chipset) I don't think you can expect cutting-edge performance. I have a number of machines of similar vintage and they're all stable and solid but I don't necessarily get great performance at buffer sizes of 64 samples.

I would not recommend buying a new RME card for use in that machine. I think you'd be better off getting another computer and keeping the card you have. Might cost about the same.

[Warren]

Have you messed around with PCI late


ncy settings? That MB has a pretty good bios for overclocking, you may have control of the PCI latency in the bios. If not you would need a utility for it (such as doubledawg or similar).

Have you tried using 2x128 or some other combination, just to see if things operate differently? I find some drivers/cards work better with one block/buffer size than others.

On that generation of machine (875 chipset) I don't think you can expect cutting-edge performance. I have a number of machines of similar vintage and they're all stable and solid but I don't necessarily get great performance at buffer sizes of 64 samples.

I would not recommend buying a new RME card for use in that machine. I think you'd be better off getting another computer and keeping the card you have. Might cost about the same.

Well that doesen't give me a warm fuzzy.
If I try to adjust the PCI latency lower than 64 the machine won't boot, The machine does much better at 4x64 than 4x32 but I've not used x128 since I hear the delay way to much. I would hate to think all I got here is a heavy paperweight, a pricy one at that:eek:

Chris

[Bob L]

When you add a second verb and it sounds flanged... that seems to me you are looping the signal back around on itself somehow or feeding the same signal to both verbs set to the same kind of sound which then stomp on each other creating a flange effect.

You should have no problem using a second or third or fourth verb on a different monitor mixer, with the signals feeding different outputs... make sure you are feeding the monitor mixers to different devices and make sure that something like TotalMix in the background (RME) is not combining out devices into a common mix.

Your 10***37; figure sounds troublesome for just adding a 24 chan monitor mix... I just tested my E6600 Duo Core 2 processor system... I have a 42 chan mix up with 7 monitor mixes running at 45%... if i duplicate and add another 42 chan monitor mixer, it adds only 3%.

Sounds like your setup still needs some tweaking... something is taking it down. I forgot... is the soundcard firewire... if so that could be your bottleneck.

Bob L

[Naturally Digital]

Well that doesen't give me a warm fuzzy.
If I try to adjust the PCI latency lower than 64 the machine won't boot, The machine does much better at 4x64 than 4x32 but I've not used x128 since I hear the delay way to much. I would hate to think all I got here is a heavy paperweight, a pricy one at that:eek:Chris, 2x128 should be the same delay as 4x64.

You don't necessarily want to set the PCI latency "low". You want to optimize it. If you have control over these settings then set all your cards to 64 and see how it operates. Once you've done that, start setting the RME card to higher PCI latency -128, 256 etc. This is the number of clock cycles the card 'holds on to' the pci bus. Higher numbers allow the card to 'hog' the pci bus.

Some have said the RME card defaults to 256. One key is to make sure all your other cards are set to a lower setting (64 is pretty standard).

[Warren]

When you add a second verb and it sounds flanged... that seems to me you are looping the signal back around on itself somehow or feeding the same signal to both verbs set to the same kind of sound which then stomp on each other creating a flange effect.

You should have no problem using a second or third or fourth verb on a different monitor mixer, with the signals feeding different outputs... make sure you are feeding the monitor mixers to different devices and make sure that something like TotalMix in the background (RME) is not combining out devices into a common mix.

Your 10% figure sounds troublesome for just adding a 24 chan monitor mix... I just tested my E6600 Duo Core 2 processor system... I have a 42 chan mix up with 7 monitor mixes running at 45%... if i duplicate and add another 42 chan monitor mixer, it adds only 3%.

Sounds like your setup still needs some tweaking... something is taking it down. I forgot... is the soundcard firewire... if so that could be your bottleneck.

Bob L

Bob can you send me a copy of the .mxs so I have something to compare it too?

No Totalmix looping at all, As far as Echo, an Verbs all effects are use pre-fader on the AUX returns fed from the various channels, I use a RME HDSP9652 on totalmix no assignments are made no inputs go anywhere so everything is don in the Host software. the only thing running are HDSP config process,Totalmix process and SAC, and the least amount of system processes to run XP and still have network. Another thing Remote SACing this is a sure fire output buffer hog.

Chris

[Bob L]

The session is linked to a large session in SAWStudio so the mxs will not be useful in itself.

But... another thing you mention concerns me... SAC Remote should add no output buffer load at all... I wonder if you do have the network activity stomping on the RME card.

I wonder if the issue has to do with the quad core... have you tried forcing SAC to one cpu only? See if the performance is worse or better.

Also... make sure to try the ASIO drivers if you are currently using MME.

Bob L

[Warren]

The session is linked to a large session in SAWStudio so the mxs will not be useful in itself.

But... another thing you mention concerns me... SAC Remote should add no output buffer load at all... I wonder if you do have the network activity stomping on the RME card.

I wonder if the issue has to do with the quad core... have you tried forcing SAC to one cpu only? See if the performance is worse or better.

Also... make sure to try the ASIO drivers if you are currently using MME.

Bob L
Hi Bob:

I am using the latest ASIO 3.073 RME drivers, I not tried MME because RME stated I woule have better performance with ASIO even though I don't use the attached WDM driver that comes with it.

I have Buffers drop even when the NIC is disabled in BIOS or not.

This is not a Quad core, but DUAL XEON Nocona's which are single core w/HT
and I have disabled HT per statments made by you in the past.

I also tryed it with two cpu's and one no difference.

At present I am looking into a PCI latency config utility as Dave discussed.

I am comfident that I well get this under control in time and I am very thankful for the help you guy's are offering. Thank you very much!

Chris

[DominicPerry]

Have you run dpclat.exe to see whether you are getting nasty peaks in dpc latency? If not, give it a go, and shut down devices one by one until the nasty peak goes away.

Dominic

[Warren]

Have you run dpclat.exe to see whether you are getting nasty peaks in dpc latency? If not, give it a go, and shut down devices one by one until the nasty peak goes away.

Dominic


Thanks Dominic:

Yes I have and do use it to see if anything pops up while testing, at present its ruler flat at 5us the only time it got crazy was with USB mouse and keyboards.

Later tonight or in the morning when I get back to the studio I am going to check out the PCI Latency adjustments that Dave mentioned, fingers crossed.

Thanks for brainstorming this issue!

Chris

[DominicPerry]

Chris, sorry, I just re-read the whole thread and realised you already mentioned dpclat.exe.
I would say that your RAM seems a bit slow - 400MHz is about 4 years old isn't it? Perhaps that's the problem. Also, the Xeon processors are great but they are really designed for servers - I don't know enough about the instruction sets for the different CPUS, nor what Bob uses, but Xeons may not be optimized for what SAW/SAC does. Is this a board from 2004? Perhaps it's just a bit old for this kind of work - a difficult pill to swallow considering how expensive I'm sure it was, especially the CPUs.

Dominic

[Warren]

Dave:

Thanks I changed the PCI Buss Latency from 64 to 255, I seemed to me to be much more stable, still only Native plugs will work for any lenght of time as slowly the output buffers will begin to rise with at first a pop here and there then after 35 min its a popcorn machine. Thats with 24 ch and 5 Mon, 2 native reverbs (now no Flanger sound YEH!),and Echo. I know its a 4 year old machine but should have a lot more funtion than it is showing at this time.

Chris

[Naturally Digital]

Thanks I changed the PCI Buss Latency from 64 to 255, I seemed to me to be much more stable, still only Native plugs will work for any lenght of time as slowly the output buffers will begin to rise with at first a pop here and there then after 35 min its a popcorn machine. Thats with 24 ch and 5 Mon, 2 native reverbs (now no Flanger sound YEH!),and Echo. I know its a 4 year old machine but should have a lot more funtion than it is showing at this time.Are you running SAW and SAC at the same time during these tests?

[Warren]

Are you running SAW and SAC at the same time during these tests?

No not yet.
Hopeful thou.:o

Chris

[Naturally Digital]

Can you post a screen shot of your cpu graph from Task Manager (during the runaway buffer droppage)?

[Bob L]

If you take out all plugins... native or otherwise... do you still get the buffer runaway thing happening... if not... put one plugin back at a time and check... see if you can find the culprit... if it is, in fact, a plugin.

Bob L

[Warren]

Can you post a screen shot of your cpu graph from Task Manager (during the runaway buffer droppage)?

Dave:

Im not there now but SAW will run up to 99-100***37; all other processes are normal. I can send something later this evening but thats what you would see. I looked at that myself in hopes to find somthing and as well during this runnaway dcplat.exe stays flat at 5us.

Bob:
I will try the with/without and one by one plug test as you suggest, maybe during the plug install it got wedged somehow however I am almost certain I have tryed it before.

Thanks you all for your help on this one, My hope is Dominic is wrong, the last thing I need at this time is a server.

The reason I went with this configuration for a DAW was a HiEnd DAW builders web site showed these componets on there top banana machine.
If all the ones with this config acted as mine, this company is got to be out of business by now.

Thanks again!

Chris

[DominicPerry]

Chris, is this machine connected to the internet? If so, I'd be happy to log on and take a look - if you are prepared to install the VNC server or run remote desktop. If you are interested we can pick a time and I can log on, look around and make some suggestions. We can get on the phone (or Skype or MSN maybe?) and look around together. Clearly, I'll understand if you'd prefer not to. PM me if you're interested.

Dominic

[Warren]

Bob, Dave, and Dominic:

Thanks guys for all the help! I am still having issues, but it appears much more stable after all your suggestions. While using the PCI Latency tool. I set the processes that showed up well below the RME HDSP9652, I had also noticed that the Onboard NIC had a pulsing problem (via TM) that when the system went into output buffer runaway, I then disabled the NIC, at that point the buffers slowed their dropping dramaticly but where still continued so I am sure I have more than one issue here. (My computer not me :o) I just can't seem to find what is stepping on the RME.

This goofy thing will go for 15 min. rock solid 35%, 4x128. 3 native effects (ECHO, and 2 VERBs), while playing GTR through 24 ch at once w/3 monitor feeds and very little effects. Then a few pops here then here comes the popcorn machine. As I stated much better now but still its got to get better to be reliable.

With all the progress thus far, I am sure well get there soon, Thanks all for all the help, its gotting a lot better.

Chris

[Rick Stansby]

I doubt whether there is any audible reason to run FoH at 96KHz. I would expect 44.1 or 48 to be fine. And it will use lower CPU and you can get away with half the number of ADAT ports.

Dominic

I know this is an old thread, but I just wanted to make a comment about live sound, at 96k vs. 44.1k or 48k. The Yamaha PM5D is the $60,000 industry standard Digital sound board in large concerts. It can be run at 96k, 44.1k or 48k. Almost everyone seems to run them at 48k or 44.1k, because being able to use the expansion cards at full channel count is much more important than the sound improvement of running at 96k.

[Bob L]

I would venture to say that SAC at 44.1k is quite an an incredible audio experience due to its internal math design than other consoles at 96k... the very first thing you will most likely notice when connecting SAC to a high-powered live speaker rig is the difference in audio quality from what you have been used to hearing with the current digital boards.

44.1k or 48k shouold be all you would ever need in the SAC arena.... but... you are free to go higher if you want to spend the money on converters and a killer high power machine that can take you there... SAC doesn't care. :)

Bob L