And I use an RME UFX+ myself, and this tool was built from that experience.

Our Windows PCs aren't designed for real-time audio processing. Default settings like USB power saving and core parking exist to conserve energy, but they can cause CPU spikes and audio dropouts — even on high-spec machines. Fixing this requires changing settings scattered across the registry, Device Manager, Control Panel, and BIOS. It's confusing and tedious.

After years of music production, live streaming, and hands-on PC setup work across countless systems, I was able to narrow down the settings that actually matter. And I built a program that applies them in just a few clicks.

What it covers:

Core Parking disable (auto-detects Intel vs AMD)

Power plan optimization (High Performance / Ultimate)

Advanced power settings (HDD timeout, sleep, USB selective suspend, PCIE link state, processor states)

USB controller power saving disable (scans all controllers/hubs)

Sound device exclusive mode management

Windows system sound control

Network adapter power management (EEE + NIC)

BIOS settings guide — auto-detects your motherboard manufacturer with direct link to BIOS download

Full system diagnostics with one-click apply/reset

English + Korean UI

Free. No ads, no telemetry, no installer — single exe. Windows 10/11.

Note: This is an early release, so code signing isn't applied yet. You may see a SmartScreen warning on download.

Download: https://www.voiceandfilm.com/product/so … vilm-care/

I hope this helps anyone just starting out with music production or streaming, anyone unfamiliar with Windows system settings, or anyone dealing with audio-related issues.

Would love to hear your feedback or suggestions for what to add next.

-------------------------------------------------------------------------------------------------------------------------
v1.3.0 2026-05-07
Added live CPU/GPU temperature monitoring
Added internet speed test
Added DPC latency measurement (Beta)
v1.2.1 2026-04-14
Changed NVIDIA low-latency mode to neutral option (Off / On / Ultra)
Added GPU power management mode setting (Normal / Prefer Maximum Performance)
v1.2.0 2026-04-11
Added per-item help tooltips
Added NVIDIA HD Audio / low-latency mode settings
Added processor scheduling (FG/BG) setting
Moved Windows Fast Startup to Optimization tab
Redesigned sound device management UI
v1.1.0 2026-04-02
Fixed USB power saving disable error
Added server notice feature
Added system info clipboard copy
Added detailed debug log export
v1.0.1 2026-04-01
Fixed sound device rename error
Fixed exclusive mode status on some devices
v1.0.0 2026-03-28
Initial release

https://www.youtube.com/SynthaxInc

Every Wednesday
5 PM / EST

Ask your questions in our video live chat.
Get answers right away from our experts and product specialists.
Latest driver news. Professional RME support. Purchase advice.

We shot a very basic, simplified video about the Extreme Power output stages of Pro and DAC and how they work with some typical headphones. It is published in our YouTube channel:

https://www.youtube.com/watch?v=15Zyuo- … e=emb_logo

Requests for more information, guidelines and ideas were easy to predict, so I thought I share a more detailed overview on the headphones currently lying around in my lab.

This is by no means a full and objective review of them, nor a recommendation in any way. I have a really hard time to recommend headphones, as I am fully aware of the many different, fully subjective and personal preferences. This broad range of different tastes and likings is the simple reason that there exist so many headphones which all find their lovers.

So don’t take my descriptions and findings as gospel. They might help you – or not.

But first things first: as mentioned in The Manual, the Extreme Power output drives all headphones very well. Exceptions are models that are either outdated or rare, mostly the super-power hungry AKG K1000 and HIFIMAN HE6. Both require a real power amp (for speakers) to drive them to acceptable volume levels.

The photo below shows some of my more often used headphones. Some more are tucked away, and four more are mounted at my workstations for immediate use.

From left to right, alternating front/back, IEMs at the end:

Audio Technica ATH-M50X: closed, low impedance, high sensitivity dynamic headphone with a lot of oomph and higher mid/presence, which makes it sound open and special. Most distracting for me are the too small earpads (I have big ears). One can easily exchange them with bigger ones (Shure fit well), but this also changes the sound a lot to the worse. This phone is a studio standard, so kept as reference.

JVC HA-SZ2000: closed, low impedance, high sensitivity dynamic headphone, no longer available. The king of bass with a unique dual driver design per side. Sounds like shit without EQ. With EQ its thunderous and physically head-shaking bass down to 16 Hz is something one has to experience at least once. As these phones can easily eat 1.5 Watts at only 14 Ohms, and didn’t cost a fortune (now they do) they are also very good lab power test dummies. That’s why I have several of them, the Japanese and Europe versions from both HA2000 and HA1000.

Mr. Speakers Aeon Closed (now Dan Clark Audio): low impedance, low sensitivity planar headphone. I got one of the first Aeon. For a planar it is astonishingly light weight, has soft and big earpads and is therefore very comfortable to wear. Good, unique overall sound. Kept as reference. If I lived in the US I would have updated already to the Aeon 2, which promises a bit more bass, and has an ingenious, foldable headband, making it fully portable.

beyerdynamic DT-990 Pro: open, high impedance, low sensitivity dynamic headphone. Yes you see that right – a spiral headphone cable! This phone has been my favourite for many years, back to the times when the number of available headphones was small and their sound mostly horrible. The 990 is lightweight, has very comfortable earpads, and a v-style tuning (more bass and a bit peaky treble). Unfortunately this was the phone which made me aware of intermodulation distortion when trying to play music with very low bass content. Kept as reference.

Audeze LCD-X: open, low impedance, medium sensitivity planar headphone with a very special, clear sound. Super low distortion no matter the volume and bass. Soft and big earpads to compensate the high weight (planar design with double-side magnets cannot be lightweight). I really like its sound (with Treble shelf raised by 3 dB), no matter what measurements say about the erroneous treble region. Cabling is provided balanced and unbalanced, and as they can eat several watts they would be perfect power load test dummies too. But I keep that task for the much cheaper JVCs. Clearly a reference to keep.

JVC HA-SZ1000: see above, basically identical to the SZ-2000. Note that all five models that I have sound a bit different. I dare to say this is not because of real differences in design, but no matching and bad quality control during manufacturing.

HEDD HEDDphone: open, medium impedance, low sensitivity AMT (air motion transformer) headphone. This latest introduction to the ever expanding headphone market brings some real innovation instead of the usual marketing bla – an AMT full range driver as headphone speaker. This is the most heavy headphone I ever had. Very soft and comfortable, big earpads shift parts of the weight from the top to the sides. One has to applaud HEDD for trying to bring this technology to the market, and I bet in a few years we will see more lightweight versions of it. And yes, it sounds different (again, like any headphone) but in a very positive way, so worth to try.

Bluedio R+: Bluetooth multi-driver headphone. Uh oh – kept as bottom reference (insert facepalm here).

Fostex – Massdrop TR-X00 PH: low impedance, high sensitivity dynamic headphone. Lots of deep bass and max volume with little distortion, maybe a bit too much treble. Not enough inner earpad space. Absolutely horrible, stiff, tangling and noisy cable that is not easy to replace due to limited availability of the special Fostex connectors at the cups. Embarrassing how (Mass-) Drop brags about improving all ‘their’ versions in details and then can’t see or fix this annoyance. Kept to give the phones collection more colour.

A bit covered in the back: Audeze Mobius. This Bluetooth planar with tons of exciting additional features and even gaming support (including mic) was a definitive must-have when it appeared. Unfortunately the sound was rather disappointing in the beginning, as the included DSP-EQ did not provide any sound change through its multiple choices. Plus the most important/critical feature, the mic, failed. While all this is fixed nowadays my son still prefers his more lightweight, standard dynamic headset for battling through the hours. Kept as Bluetooth reference.

Sennheiser HD650 / Massdrop HD 6XX: open, high impedance, low sensitivity dynamic headphone. Lightweight, but a bit too stiff ear pads. Still comfortable. Praised for its linear, uncoloured sound, although quite bass-shy. Kept as reference. Tip: balanced cables can be done very cheap by ordering a standard replacement cable and replacing the TRS with whatever you need, as both sides run separately down to that connector.

Bose (Quiet Comfort) QC 15: closed Bluetooth headphone that can also work wired and was a reference for noise isolation. These days there are not only newer models but also tough competition. In fact the Sony WH-1000XM3 is much more popular in my family and therefore not seen in the photo. Kept as reserve.

Cut off from the pic, left down, not visible: beyerdynamic DTX-350. Closed, low impedance, high sensitivity dynamic on-ear headphone. My travel phone for many years. Sounds several classes better than its price predicts. Highs can be a bit problematic (the beyer peak), but nothing a simple EQ can’t fix (I use Neutron on my iPhone and iPad).

Front IEM in custom case: InEar ProPhile 8. Low impedance, medium sensitivity in-ears with 8 (!) drivers per side, crossover-separated into 4 bands. The best sounding, un-coloured and amazing headphone / earphone I ever heard (don’t forget, this is fully subjective and might not apply to your taste/opinion at all). Except for the very deep bass, where IEMs generally struggle due to not being able to physically transfer vibrations to the skull as some headphones can do. They replaced my DTX-350 as travel phones, and I couldn’t be happier.

Front second IEM: cheap Massdrop x NuForce EDC in-ear monitors. Well, you get what you pay for. Kept as low cost reference.

Not on this photo: Sennheiser IE400 Pro. Cheaper alternative to the ProPhile 8. With just one dynamic driver per side Sennheiser managed to provide an astonishing sound quality. Biggest drawback for me is that they don’t provide big enough ear tips to close the big holes of my ear canals. A personal annoyance that these share with many other manufacturers. Why nobody makes XXL tips?

Not on this photo (at my office desk): Sony MDR-Z7M2. Closed, medium impedance, medium sensitivity dynamic headphone with unusual big drivers (70 mm !). As I wrote in the beginning this topic is highly subjective and personal, but people will ask anyway, so I try to explain why this is currently my most often used headphone. It simply ticks most of my boxes. Design, craftsmanship, manufacturing, details, material, looks – all absolutely amazing. Comfort level 120% with medium weight and big, soft earpads. Screwed cable connection with the right angle, and a very soft, not tangling and noiseless cable (eat this, Drop!). Nice sound with very deep bass extension. Highs are a bit too recessed for my geriatric ears, so I dialed in a simple PEQ correction filling up two bands (see below) where measurements show a drop. Perfect. For me. YMMV, of course…

Also not on this photo: tons of other headphones that I tried over the years and especially in the time the ADI-2 Pro was developed. I had no reason to keep them when they did not meet my personal taste, but had passed the Extreme Power output test, and would then just lie around as dead money.

A quick note on headphone stands: the ones in the photo are mostly very cheap no-name ones from Amazon https://www.amazon.de/Kopfh%C3%B6rerst% … B07PHWNHP2 which do not leave marks on the headband, are high enough and still stable.

The heavier phones are parked on the king of stands – German engineering, of course – Rooms Audio Line FS Pro A BK Headphone Stand. Made of solid steel they stand like glued to the table. Height is adjustable, and even damped while adjusting. Available with two different radius to perfectly match the phone’s headband. Also available in silver and wood versions.

At my desk and workstations I use a simple clamp with curved surface, also not leaving marks on the headphone: https://www.k-m.de/en/products/accessor … lder-black

Measurements and setting up the PEQ

There is no final truth. Neither the diffuse field nor Harman curve is guaranteed to bring you the highest joy in listening. But measurements of headphones DO highlight the flaws in them. How much correction is needed and which flaws to tackle then is your personal decision. So you get some guidance, but you have to fine-tune yourself. For example the Sony’s MDR-Z7M2 recommended EQ boosts the mid area around 1.2 kHz by 2.5 dB. I don’t like that. To me it sounds better without that correction. But I agree in fixing the drop at 4.3 kHz and above 8 kHz. My simplified PEQ does this:

B4 G +6.5 dB F 4.3k Q 4.8 (Peak)
B5 G +5.0 dB F 10.0 kHz Q 1.3 (Shelf)

That already gives the Sony a much nicer and fully reconstructed treble sound. Now with

B1 G X.X dB F 65 Hz Q 0.8

I can dial in as much deep oomph as I want, without the bass getting bloated or leaking into the lower mids. Fun all around!

You can follow the recommended EQ corrections more closely, as the ADI PEQ has 5 bands, and even Treble and Bass can be included (as Oratory1990 successfully shows). But in my experience many of the smaller corrections do not give any audible change, or (see above) might even change the sound to something you don’t like. Just experiment with it. Doing so be warned, your ears quickly adapt to any change! You might end up in settings that scare the hell out of you when you put the phones on again one day later. What you need is a second source (headphone) as reference to compare the dialed-in EQ.

I know it is easy to get lost here, and understand why people look for the 'perfect' EQ setting so they don’t have to do all this. Still in my opinion there is no such ‘perfect’ one that suits all tastes and ears. So if you really wish Nirvana you will need to invest some time.

Sources

AutoEQ: https://github.com/jaakkopasanen/AutoEq … er/results

Oratory1990: https://www.reddit.com/r/oratory1990/wi … of_presets

I hope this little overview and experience report provides some insight and helps you to find the ideal headphone that matches your ears!

You can disable the noise suppression and send "original sound" with the settings below in Zoom settings > Audio > Advanced

With these settings in Zoom and a properly configured Loopback on Output 1/2 your audience can now hear the full mix of audio, not just your microphone. See here for more info about Loopback: https://youtu.be/1Bca_dn7u6w

RME drivers (kernel extensions) are signed and notarized, compiled with the latest XCode, and therefore in most cases an easy and quick installation. But for some it can mean multiple installations, manually deleting files and countless reboots until the dreaded Security & Privacy Allow dialog pops up. Worst: for some that never happens.

There is a working solution to this problem. If all the following steps have failed:

- manually performimg the 'allow' function in the Security & Privacy app. Will only work if the RME driver is shown there.

- reinstalling the driver including reboots several times until the dialog comes up

- deleting the driver's kernel extension manually prior to reinstalling it

then the user can add the RME driver manually to the Mac's Trusted List. Instead of dectivating the whole System Integrity Protection (SIP), which also works, but obviously is not the best approach, adding the RME kernel extension (precisely: the RME team identifier) to the trusted list disables the SIP only for RME kernel extensions. No harm done and the computer is still protected as most users want it to be.

Here's how:

- Boot into macOS Recovery mode. To do so turn on the computer and immediately press and hold the keys Command (⌘) and R. Release the keys when you see an Apple logo, spinning globe, or other startup screen. Enter your password if requested to do so. Startup is complete when you see the utilities window.

- Open Terminal from the Utilities menu in the menu bar.

- Run the command: /usr/sbin/spctl kext-consent add 67AK2U2X7M

- Reboot the computer

It should not be necessary, but in case the driver is still not loaded check that everything worked as expected by using this command in the Terminal (Recovery mode not required):

/usr/sbin/spctl kext-consent list

This should give the following result:

Allowed Team Identifiers:
67AK2U2X7M

Thanks to our forum user arikaspi for making us aware of this procedure.

Links:
Mac OS Recovery: https://support.apple.com/en-ca/HT201314
Sophos SKEL Troubleshooting: https://community.sophos.com/kb/en-us/132813
RME Forum thread: https://forum.rme-audio.de/viewtopic.ph … 65#p152665
Debugging SKEL issue: https://forum.rme-audio.de/viewtopic.php?id=30416

Documented are several ways to solve this:

- manually perform the 'allow' function in the Security & Privacy app. Will only work if the RME driver is shown there.

- reinstall the driver including reboots several times until the dialog comes up

- delete the driver's kernel extension manually prior to reinstalling it.

In seldom cases this all fails and the unit/driver stays blocked. There is a thread about this that includes a lot of details:

https://apple.stackexchange.com/questio … gatekeeper

There should be a way to find out why macOS denies loading the driver. Following is a description of it. Note that we do not have a computer where this mean behavior is happening, so the described method might not work. But it is worth a try. Here we go:

Experienced customers could try to load the driver manually with the command line tool "kextutil" and see what error message they get (ideally the driver is loaded once and available in the future).

If you want to load (or check) KEXT directly from the Library folder, you have to navigate to it within the terminal window using the change directory command: "cd ...", "cd ...", "cd library", "cd extensions" (the first two "..." go from the user directory to the roots directory, from there to the non-Apple drivers).

Then you can check with "ls" that there is a file "RMEFirefaceUSB.kext" in the directory and type "sudo kextutil -v 2 RMEFirefaceUSB.kext" and confirm with your login password (-v 2 sets the info level). If the KEXT was already loaded as on our computers the output is a bit cryptic:

user@User-MBP extensions % sudo kextutil -v 2 rmefirefaceusb.kext

Defaulting to kernel file '/System/Library/Kernels/kernel'
/Library/Extensions/RMEFirefaceUSB.kext appears to be loadable (not including linkage for on-disk libraries).
Loading /Library/Extensions/RMEFirefaceUSB.kext.
Reading load info for 19 kexts.
KextAudit initialized: audit=T
KextAudit didn't find a bridge: audit=F
Created mkext for architecture x86_64 containing 1 kexts.
(kernel) Notice - new kext com.apple.driver.usb.AppleUSBHostPlatformProperties, v1.2 matches prelinked kext but can't determine if executables are the same (no UUIDs).
(kernel) Received request from user space to load kext de.rme-audio.driver.RMEFirefaceUSB.
/Library/Extensions/RMEFirefaceUSB.kext successfully loaded (or already loaded)

If macOS blocks the kernel extension from loading maybe the info why is shown here.

Edit: Final solution to this issue ican be found here: https://forum.rme-audio.de/viewtopic.ph … 41#p152741

This is information from this thread:

https://forum.rme-audio.de/viewtopic.ph … 30#p150430

Using a newer computer with a second or later generation Thunderbolt 3 implementation, a TB3 to TB2 adapter (Startech or Apple) and a Sonnet Echo TB2 to ExpressCard adapter, or any TB2 to PCIe chassis, Windows throws a Blue Screen Of Death with Driver Verifier DMA Violation warning.

This issue seems not present on first generation Thunderbolt 3 implementations (the Thunderbolt Control Center is also different, and the Apple adapter won't work with that hardware).

Remedy (special thanks to forum user glittle):
It seems the BSOD caused by some strange interaction (because this issue has nothing to do with DMA) of the DMA Verifier process can be disabled via a registry entry.

HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Services\pci\Parameters\DmaRemappingCompatible

is set to 0x00000002(2) by default. This has to be changed to 0x00000000(0).

There is advice to change a second entry as well:

HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Services\USBXHCI\Parameters\DmaRemappingCompatible

which might not be necessary (at least it wasn't on our test systems), but as it could only improve reliability there is no reason not to change it as well.

Be careful with registry modifications as these can destroy your Windows installation easily! If you feel unsure let someone do it who knows this stuff!

I just installed latest catalina drivers for usb madiface at the warehouse. After installation and allowing restart OS loaderbar hangs up at around 60%. Anyone had the same issue before and/or tips to proceed?

https://www.forum.rme-audio.de/viewtopic.php?id=27796

This includes DIGICheck, and on units where the FX are rendered by the computer (Babyface Pro) also TotalMix FX. Here the FX output stays silent/dead as long as TM FX is not allowed to access the 'microphone'.

my first mistake was installing the firewire drivers instead of the usb drivers.  once I realized that i had installed the wrong drivers i installed the usb drivers.

the symptoms are:

• Fireface USB settings app sees the devices serial number and can change clock source but cannot change the sample rate (when I change clock source in app I se change reflected on device, but not sample rate).  If I turn off the device it and its firmware / driver info go away from the UI as expected. so the computer sees and can talk to the device

• Totalmix doesnt see the device at all (nothing appears in New Totalmix Window device selection list)

• Fireface Flash Tool sees the devices serial number but cannot flash the device, when I click Update nothing happens. also, under the Current Revision column, the flash tool sees the correct version for USB but reports 0 for the other firmwares.  When i test on different mac the correct values are shown

• Mac Audio MIDI setup does not see the device at all

Update: I have tried uninstalling all RME software and reinstalling drivers per this https://www.forum.rme-audio.de/viewtopic.php?id=9233 but to no avail.  Note I could not find all of the plists listed.  Maybe the post needs updating?

Update: have tested the device on my wifes Macbook Air (using 2.x drivers) and the device works great so we can rule out hardware errors, the problem is entirely the config of my Mac.  using her Mac I have flashed the device to the latest firmware.

If you are like me a heavy RME interface user, and also a Samplitude or Sequoia user, you will be familiar with the problem of ASIO channel naming.

I have developed a small utility to workaround this.  Currently, Magix use the name of the driver, not the driver-reported hardware name of the device channel.  We can rename these ourselves in the app settings panel, but in my situation, I have interfaces with multiple I/O types: analog, MADI, ADAT, AES.  Every time I change the sample rate (which I do often) or reconfigure the Madi ports, any custom naming becomes invalid as the number of device channels change.  Also, when the driver is updated, the names are reset too.  I needed something to more easily automate better naming of my ASIO channels.  This is it (for me, for now).  Please feel free to use it. 

FAC > Friendly ASIO Channels (for Sequoia & Samplitude)

thanks,
Simon

Design and form factor have not changed over the years - to the left you see the older PCIe to PCIe TurboBox. The Mac Mini in the background shows that Netstor's design fits nicely with Apple's design. For short audio cards the boxes are a bit longer than necessary, and the 3-slot version does not have spare slots to mount add-on cards for more I/Os or word clock. That is something to keep in mind, additional cards which do not use a PCIe slot need to be counted as if they do when ordering such an expansion box.

The NA211TB3 has three free PCIe 3.0 8x slots, an internal switching mode power supply, and two USB-C sockets. The package includes a 2 meter active Thunderbolt 3 cable with USB-C on both ends.

As you can read in the above mentioned reviews, and this is true even now years later, the expansion chassis are regularly used by RME developers and never failed. Not once. The available audio performance is defined by the used computer, not the chassis. Low latency at 32 samples is possible if the computer can do so. If not then the computer is at fault, not the TurboBox.

I am using the NA211TB3 regularly to use all our PCIe cards with a Dell XPS15 9550, a Skylake based laptop with a USB-C/TB3 connector. Usually a Startech TB3 to TB2 adapter is used to connect a wide screen monitor which has a TB2 hub, so I can connect a UFX+ via Thunderbolt as next device. Simple serial cabling that works a treat.

The NA211TB3 has a TB3 hub inside, which is absolutely nice to have. When connecting the expansion box to the latop,  its only USB-C port is occupied, which would mean no monitor and no UFX+ anymore. Instead of plugging the Startech adapter into the laptop it can now be plugged into the second USB-C port of the expansion chassis, and everything works like before - double treat! One just has to note that this is a TB3 hub. You can not connect USB based devices/adapters with USB-C connector, like Dell's own USB-C to VGA/Ethernet/USB3 dongle. It can't work as there is no USB on this hub. Adapters to USB/Ethernet etc. that are PCIe based will work, though.

The photo shows the inside of the box with a HDSPe MADI and a HDSPe AES working on 80 channels input and 80 channels output simultaneously (note this is the bus load. To sync the cards I mounted the MADI's word clock module, which meant I could not mount the second card of the AES anymore. So externally only 72 channels were available). Now these days a laptop won't even raise its eyebrows, so to say, as one can also plug in the HDSPe MADI FX with 194 channels input and 196 channels output - and it still works without noticable base load.

The really nice thing with UCB-C/TB3 and the TurboBox is the hotplug behaviour that one gets with PCIe cards - which still astonishes me regularly, as I have seen the opposite on so many systems over so many years. With Windows 10 running, simply plug in the USB-C cable, the TurboBox immediately turns on, and there are the cards and audio channels. Finished your audio work - simply unplug the cable, the TurboBox will turn off automatically, the cards/channels are removed from Windows. No crash, no hang, no problem - amazing times!

I intentionally describe this behavior in detail under Windows. Of course it works the same on a Mac, but nobody would expect it NOT to work perfectly on that platform. Which it indeed does, flawless operation there as well.

Regarding available latency - as mentioned it depends. The XPS15 is a real Dell - which means it won't give you lowest latency, no matter if USB2, USB3 or Thunderbolt. Currently the combination of HDSPe MADI and HDSPe AES needed 128 samples buffer size in ASIO to work click free. At 64 from time to time a small drop-out occured. The HDSPe MADI FX and MADIface XT also required 128 samples. I thought 64 samples worked before, but there are so many driver, OS and BIOS updates these days, you simply loose track what happens when...and I won't waste my time trying to improve a Dell's low latency performance! But else the XPS15 is a very nice unit, that's why I use it daily, both in home and when travelling.

How powerful modern laptops are became obvious when I connected the UFX+ via TB to the TurboBox second USB-C port, while a full channel test was running with the MADI and AES cards at 128 samples buffersize. I started to use the UFX+ with its TB driver on all 94 channels I/O, completely independent from the running audio test - and it did not cause a single glitch/drop out in that MADI/AES test. Wowza!

Using the MADIface XT via external PCIe (with Matrox PCIe/ADP card in the TurboBox) was another nice surprise, as those expansion cards never before were usable with hot plugging and unplugging. With TB3 in-between even that works.

Once again Netstor provides a perfectly working expansion chassis, this time for the latest USB-C/Thunderbolt 3 technology. It is fully compatible to all RME PCIe cards, from the old Multiface PCIe up to the HDSPe MADI FX. Well done, Netstor!

First try not impressed at all :-(

So I go thru all the old tweaks to see if something helps. Nadia, I get depressed. Time to do hard internet search.

Looking at the resource list of irq sharing on this computer reveals the problem at once: The graphic card is sharing irq with both USB and FireWire controllers all on irq 16! Reading on internet, irq sharing is supposed to be "OK" and something that drivers are supposed to completely support, right...

I started looking at the irq numbers, and being not young anymore, I see something new to me, negative numbered irqs, what is this? Answer: Message Signaled-Based Interrupts = MSI.

So what?

This is a more modern approach to interrupts that goes beyond irq sharing. Can all devices do this? No, but some can and this can have a HUGE HUGE impact on your system. In the link below I will share a document/link that describes the process of changing the way devices use old irq or the new MSI.

USB version 2 can not use MSI but 3 can.

First of NVIDIA support this and switching to MSI made irq sharing with the USB and FireWire controllers go away. This was the moment of eureka. Reboot and test, BINGO! Rock stable both USB and FireWire, not pops or clicks. Started Firefox with lots of tabs while playing guitar through the DAW just to be nasty to the system, still no pops or clicks. Started Latency Monitor because it can produce pops and clicks sometimes, same result, perfect audio!

So I tried to put the FireWire controller in MSI mode and it started up, but the UCX was not detected by the RME drivers, I wonder if this can be fixed, because it seamed that the FireWire card was detected without any errors by windows.

This is the cure to our problems with irq sharing, that still exist today. You can do this with lots of devices, do be careful with your system while tweaking.

Here is the resolution:

http://forums.guru3d.com/showthread.php?t=378044