You are not logged in. Please login or register.
RME User Forum → AVB/Dante - Network based devices → 2nd Aox-D not working on PC
I've noticed that the MADI IO of my 2nd Aox-D stops working after a while (> day).
I usually leave my PC on 24/7. But usually the next day after switching on my MADI devices (UFXIII, ADI-648 and H9000R) connected to Aox-D#2 are not registering any activity in Totalmix or any DAW. A reboot solves this. It's rather consistent so it's not a glitch but reproducable.
Does anyone have any similar issues?
Reproducible means it is happening on all systems.
Why don't you tell us what system you have.
There are many bugs with overnight sleep and constant on on windows. Cubase had the nasty habit on crashing after a few hours of inactivity for example, but if course not on all systems.
In any way it is important to not have any power saving options on. There is a utility here on the rme forum that can set all power saving options correctly and also indicates when one is not set properly. I am sure @ramses has a link for it.
Cheers
I won't fully rely on sleep and personally prefer a freshly booted system each day.
What energy profile is in use by you?
It can't hurt to check every energy-saving profile to disable power saving for USB and PCIe.
This tool can be of help for such things and also to optimize some other things.
https://forum.rme-audio.de/viewtopic.php?id=43146
What also can cause issues is "Fast Boot" (or "Fast Startup"). Instead of performing a clean shutdown, Fast Boot logs off user sessions and hibernates the Windows kernel and system drivers to a file (hiberfil.sys). When you turn the PC back on, it restores that exact snapshot into RAM.
In recording environments, it is strongly recommended to turn Fast Boot off, as it can cause many different issues.
Another positive side effect of disabling fast boot is that the shutdown of the system runs much faster, as the state has not to be written to disk. It really nerved having to wait for shutdown if you are in a hurry.
If you want to save energy, there is an alternative approach.
I use Process Lasso Pro by Bitsum to switch power profiles dynamically, depending on which applications are open and on whether the system is idle (no keyboard or mouse activity).
The default power profile for office work and normal operation is "Balanced".
The so-called "IdleSaver" automatically switches the system to "Power saver" when there has been no mouse or keyboard activity for xx seconds. I use 15 seconds.
You can also enable an option that disables IdleSaver while "Performance Mode" is engaged.
This is important for a DAW, because while working with one, you may not touch the mouse or keyboard for a while but still need full performance.
As soon as I start applications with high-performance demands, such as a DAW or photo editing, the power profile switches automatically to "High performance" (or any other profile you prefer, e.g. "Ultimate Performance").
In addition, I enable "Induce Performance Mode" so that IdleSaver is ignored. However, you can also assign a power profile only, so that IdleSaver remains active.
I tweaked the "Power saver" profile so that it turns off the display after 3 minutes. When I am idle, Power Saver kicks in, and three minutes later the screen goes off. This helps reduce wear on the OLED screen.
In addition, I used Bitsum's ParkControl to maximize CPU core parking and to enforce the lowest clock frequency.
The result is a responsive system that automatically selects the power profile my work requires.
For you, it could have the advantage of saving some energy here and there and of avoiding sleep states during the day.
Overnight I would shut down so that the next day starts with a freshly initialized system.
You can implement this step by step to see whether the tool and disabling Fast Startup have already improved things.
Process Lasso Pro is an excellent tool; the dynamic switching of power profiles works flawlessly.
In case you suspect it might slow the system down: it doesn't. It only parameterizes the process scheduler through a defined API that Microsoft created for controlling the scheduler, plus some extra functions like e.g. ProBalance.
ProBalance can help to rein in certain runaway processes, both in terms of memory and network load. I would use ProBalance but disable it for the DAW. It helps to reduce DPC latency spikes, for example when Firefox is generating a heavy load.
Thanks for thinking out loud with me.
After a turn of events I think the issue is somewhat more discomforting. But first let me explain.
The system is based on a Pro WS WRX90E-SAGE SE with a AMD Threadripper Pro 9975X.
Today I found out that the Asus tool I was using for checking driver/bios updates was superceeded by 'MyAsus'. Which is a dreadful application filled with marketing and AI datacollection stuff which you can opt out to. But it showed a new BIOS and driverset update which I just installed. It's really nice you have to find out this by yourself.
After reboot my 2nd Aox-D didn't respond to any incoming MADI messages from the 648. My 1st Aox-D did from an A32 pro.
While just before the driver and BIOS installation it all was working. Even after a nights sleep where only the computer stayed on.
Thinking back the issue with the 2nd Aox-D started after a Windows 11 security update. (Which I always check for not messing up my settings to 'defaults', which happens all the time)
I'm using Project Lasso btw and use the performance profile. (And I eddited to be the 'balanced' profile to be exactly the same if a Windows update ever decides to 'help me save then environment'.)
But my theory now is that after every power cycle its either at random that the 2nd Aox-D works just fine. Or only for every odd powercycle. Both are annoying, but I cand work with that if I have to. But it would be most welcome if RME would look into this and may provide a solution.
edit: It appears to be at random, which is even more annoying.
And have you tried to place the card to another PCIe slot? Or swap the cards?
... and test each card individually.
Sure I can do that. But it's not that easy to swap cards. First I have to take out the 5090 with the riser cable and then take out the cards or swap them. All PCI-e slots should be (theoretically the same). And cold booting with 256Gb of memory takes a few minutes every time. So that's about half a day or a full day of testing.
OK, not simple. But there, probably, is not other way.
I do not know, how it is with Threadripper, but Zen has two kinds of PCI-e, some of them go directly from CPU, some from chipset. And even sockets of the same kind may slightly differ, here and there. Or certain socket may be defective or out of specs.
One more thing, you might try, if that is possible without removing CPU. Try to remove and insert the card, or just move with it a bit, sort of wiggle. Not much hope, it would help, but it might.
Nope, Threadripper chipset only supplies USB and other periphials. All PCI-e are connected to the CPU. Some PCI-e have a repeater because of the longer distance but that's that. That's the main reason I got this motherboard. Because of the PCI-e lanes so I can stack them with cards. And in the future even add a 3rd Aox card by swapping it with a firewire controller.
Wiggling the card doesn't make sense when the card correctly works at random after a reboot.
Swapping the cards kinda makes the best sense because then you have a clear indication if it's the PCI-e slot or not.
But I need to plan this ahead since it will take some time.
Wiggling the card doesn't make sense when the card correctly works at random after a reboot.
It can make sense if the issue comes from bad contact.
In this case take out the card and clean the contacts with a pencil rubber.
These are brand new cards - cleaning makes no sense here.
Swapping the cards is a good idea. Personally I expect the mobo/OS having issues to handle the required resources correctly. The enormous number of channels already might screw things up...
Wiggling the card doesn't make sense when the card correctly works at random after a reboot.
It does make sense exactly because of random behaviour. I do not say, it will help. Chance is rather small as the card is new. Placing card to another slot has higher probability to help.
Wiggling the card doesn't make sense when the card correctly works at random after a reboot.
Nope. You could see whether the issue moves with the card or sticks to the slot.
And by testing one card only, you could find out if it's an issue of resources - if swapping doesn't produce a clear result.
Just a question, why do you need 2 of these in one system?
Because I have 7 Madi devices connected and plan to have Dante connections in the future as well.
RME User Forum → AVB/Dante - Network based devices → 2nd Aox-D not working on PC
Powered by PunBB, supported by Informer Technologies, Inc.