Option for optimized performance for a single core and single zone playback [Not on Roadmap]

Same machine? Is Roon running will you listen to A+? If not, try that…

Hey Danny.

Same machine yes. Each app is closed while the other is in use.

Both using exclusive mode and integer mode and bit perfect mode.

Two men say they’re Jesus one of them must be wrong

The above lyric from a Dire Straits song popped into my head when I read the first post. Both players claim ‘bit perfect’ playback, so if there is a difference…


I really don’t want this thread to go into why they may or may not sound different. There’s already the RAAT vs DLNA threads that cover this I think.

This is just based on some observations I’ve seen around the web and now experienced myself and just a friendly request to keep pushing the sound quality forward.

I didn’t believe that Darko article when I read it 2 months ago but hearing is believing.

It doesn’t have to be about RAAT vs. DNLA or any such comparison. I imagine it would be about Mac specific optimisations Audirvana may be able to make that you wouldn’t necessarily have the opportunity to do for an application ported across a number of operating systems. The nearest you might get is the symbiosis of Roon with ROCK.

1 Like

I have an idea of what’s going on… and both apps can still be bitperfect.

Our audio stack is custom for Mac with no portable code – it is built using the Mac OS APIs, just like A+, including Darin’s paper explaining Integer Mode. As for RAAT vs DNLA, there is no debate on sound quality between the two. RAAT and DLNA are equals when it comes to SQ. Here is our problem with it:


@dabassgoesboomboom – I’d love for you to repeat your test but with this:

  • run A+ and play
  • run Roon while A+ is playing and just use the UI, but do not play.

Do you still hear the difference or does A+ sound the same now? Can you have someone else do the actions while you listen?

I’m wondering if your DAC + computer are not shielding RFI properly and your GPU is being driven harder by Roon, and RFI is leaking into your D->A process inside the DAC.

When Roon speaks to Core Audio in exclusive/integer mode, the apps can not be doing anything different here (assuming you are playing the same content w/ no changes like DSP. There must be something environmental, and the biggest thing I can think of is that Roon has a very heavyweight UI compared to A+.

Normally I would argue your device should be protecting you against this, but every device handles RFI differently. This is why we suggest you take the “Core” and “Remote” out of the listening device and use a network device to render audio. Even a USB bridge can isolate RFI generated from a GPU. On my Macbook Pro from a few years ago, I can visibly hear on disconnected speakers the scrolling of Chrome. Unreal.

I think I can offer my own experience here as I evaluated A+ and Roon on the same machine feeding the same AV chain, etc. My setup was:

A+ (on MBP Touchbar 2016) -> ethernet -> microRendu -> USB -> KEF LS50W

Roon Core (same laptop) -> ethernet -> microRendu -> USB -> KEF LS50W

No upsampling or DSP stuff going on. All vanilla. One app closed while the other open.

I listened at length to several different tracks and I could not hear a difference.

I find Roon has improved over the years in SQ when connecting the DAC directly to the computer running Roon Core via USB. It’s decent now. However, Audirvana has always come out ahead for me.

I find using ethernet and separating Core from endpoint player is the best way to go. To me that sounds the best. I do this even though the computer is near the audio gear and the USB connection would be much simpler. All for SQ. Ideally, I wouldn’t have any computer in the room and would just be using the iPad for control.

Where does your wireless router fit into this setup? Does that also get ethernet from router to Core?

I don’t know if that’s addressed towards me or Sean?

If you’re asking me. I have a separate closed router/switch setup for Roon.

My router is in another room. Unfortunately, I don’t have ethernet running yet through the place. I tried Powerline adapters, but found them noisy. My endpoint BDP-1 only takes ethernet.

So I use my iMac which allows me to have 2 network connections. One with wifi and other with ethernet port. I use the Wifi connection to connect to the internet and run things as you would normally expect. This ensures Roon is also connected to the internet.

I have a spare Linksys router and D-Link switch. I use them both in the room. The router is needed to assign the IP addresses. I’ve turned off its wireless capabilities though along with other things. It’s very stripped down. The router connects to a gigabit switch. To the switch I connect the iMac and the BDP-1. Just the bare essentials.

iMac’s Wifi for internet connection. iMac’s ethernet port reserved for Roon output. Runs perfectly. Never a hiccup.

You and anybody else who wants to respond. I am trying to set up my roon system for best SQ. I have the core on a 2009 Mac Pro that’s installed on an ssd for the os drive. The Mac Pro is Ethernet connected to my wireless router. My end point is an Auralic Mini which is wirelessly connected to roon and internet. So am I best advised to run Ethernet from my Mac Pro to Mini or do I run Ethernet from router to Mini? Is there a difference? Thank you.

I don’t have experience with your Auralic Mini so I don’t know if wireless or wired is preferred. Are you having any problems currently with the wireless connection like dropouts, or are you just trying to improve for sound quality?

In my setup that I had described above, I had both the router and switch in the room, so I experimented by connecting the BDP-1 and iMac directly to the router vs. having the switch in the middle. I did notice a slight difference in SQ between the two connections. I also found benefit in feeding the switch with clean power vs. the stock switching power it came with.

I guess I just wanted to know does it mater if I ethernet connect from end point to router or Mac Pro (core).

I too have experienced this difference between A+ and Roon when running directly to USB and have raised it in prior threads. The best defense was electrical noise, as @danny mentions above. But I have monitored the CPU and the difference that the Roon UI adds is minimal. I find it hard to believe that increasing CPU load from say 11% to 14% is having such an obvious impact on SQ. Even though Usb carries power, I simply don’t believe it could make that much of a difference. Really frustrating!

I have often wondered if A+ output is not bitperfect and there is other stuff going on. How all media players can sound so different when they all claim to have bit perfect output doesn’t add up. I also wonder if the A+ direct/integer modes (where one can choose between integer mode 1 and 2) are affecting the actuals bits coming out of the Mac. A+ directly bypasses CoreAudio does it not? So there might be a difference there.

That said, I am now comparing A+ -> ethernet/dlna -> Kef LS50W vs Roon > SMS-200 -> USB -> LS50W and there is very little between them. The A+ route might be a tad airier and tighter i.e. more controlled - underlying theoretical cause could be that USB carries 5v power whereas Ethernet is galvanically isolated and carries none. That seems more plausible than the above small differences in CPU load between Roon UI and A+ UI. This makes me hopeful that Room -> ethernet > LS50W will offer at least the same if not better performance than an endpoint when it is released. I also need to test A+ > SMS-200/dlna -> USB to see how that compares.

Of course, it’s crucial that one is doing Apples to Apples comparisons. All DSP/upsampling/AU Units need to be disabled in A+ and Roon.

A+ still has the option for “Direct Mode” and 2 integer modes if this is enabled and it’s not entirely clear to me how they impact the bits coming out other than vague descriptions of different rounding functions. DirectMode bypasses CoreAudio and some claim that it makes an improvement so maybe that is a difference if not controlled for?

Whats the closest match to Roon in the following? I am assuming it’s as below.

It may be that Roon is a heavier load on the GPU (not CPU) when compared to A+. Hence the source of electrical noise is the GPU??

I suppose it’s possible but sounds pretty dubious to me. I am using a MacMini - there’s no souped up turbo GPU running on this thing that consumes much power. Maybe for minute/tiny differences in sound but not such an obvious difference.

I also wonder whether adding an endpoint, which seems to level the playing field, is actually improving the Roon SQ or if it’s decreasing the A+ sound, or a bit of both. I know that Danny says that RAAT and DLNA offer the exact same SQ, but doesn’t dlna A+ have to convert the FLAC/etc file to .WAV before sending to DLNA and thus it might negate the A+ direct mode stuff?? Guess I need to make sure DirectMode is controlled for when testing …

Thanks @danny , I will happily do this test later this weekend. I completely get Roon’s recommendation about having the Core outside the listening Room and using a light networked endpoint. in my previous setup I had a sonicTransporter i7 powering 4 Roon endpoints are the house - those werehappy days. However having moved into a small apartment circumstances have changed and with that, some compromise. But at the same time, A+ is having to deal with this same setup as Roon - same machine. Again this post is all out of love for Roon. I’m not a hater. I’m in this for the long haul as a lifetime member and looking to work out why things may be different and hope things can continue to improve in this space (sound quality).

Regarding other’s points about whether the apps are actually playing bit perfect. Before I just got my Hugo2, my DirectStream DAC has a stereo 24/192 bit perfect test file which has the FPGA verify that playback is bit perfect - both Roon and A+ are capable of true bit perfect playback so that’s not a concern of mine at all. Obviously a very good question to ask though and not many DACs are able to test you if your playback chain is truely bit perfect but I was able to verify this.

Regarding my WiFi router, it’s in the same office room but when I’m in there on the Mac I always turn WiFi off.

So, to clarify, you have heard a SQ difference between A+ and Roon over USB using this bit perfect playback test on your DirectStream DAC? If so, that’s very very interesting.

Not using this bit perfect test file. The test file is not music. It’s a file created by the DirectStream’s designer, that his FPGA uses to verify playback is bit perfect or not. There may be a couple other companies that do something like this, but it’s not common. It’s one thing for Roon (or any software) to tell you it’s output is bit perfect but another for the DAC to verify for you also.

I’m sure many DIrectStream owners reading this would also think it would definitely be interesting if I could hear a difference of this test file, between different software, given the test file is not music :smile:

So to be clear, the bit perfect test file was used just once with A+, just to verify it’s output (and the entire chain up to the DAC) is bit perfect. Which it is. This thread is based on observations based on listening to music in A+ and Roon.

I also don’t want this thread to turn into ‘software can’t possibly sound different’. There are enough observations out there and Danny has already hinted at a possible cause which I am familiar with also - the old Blackberry near the speaker is an extreme example but along the lines of what he is talking about. I’ve also heard the fast scrolling being heard on a nearby speaker but that was many years ago too.

I’ll test Danny’s suggestions later this weekend.

1 Like