AUDIO over IP - REDNET 3 & 16 Review. AES67 Sets A New Standard for Computer Audio
Oct 12, 2016 at 2:09 AM Post #2,161 of 3,694
Maximum for both settings.

Read this page for the proper settings in low latency environments: https://www.tradingtechnologies.com/help/tt-system-requirements/appendix-setting-descriptions/nic-settings/


And this section about low latency from Microsoft: https://technet.microsoft.com/en-us/library/jj574151(v=ws.11).aspx#bkmk_low

Thank you for the reply.  The recommendation to set the RX and TX buffers to max is opposite to read of have read about getting optimum performance.  Granted what I read was from the perspective of gaming.  Reducing the buffers helped to reduce lag which I assume was caused by the application being "starved" for data due to latency.  Of course gaming has much different network demands than audio.  I must admit that the idea of setting the buffers to max would seem to increase latency????
 
Oct 12, 2016 at 2:19 AM Post #2,162 of 3,694
  Thank you for the reply.  The recommendation to set the RX and TX buffers to max is opposite to read of have read about getting optimum performance.  Granted what I read was from the perspective of gaming.  Reducing the buffers helped to reduce lag which I assume was caused by the application being "starved" for data due to latency.  Of course gaming has much different network demands than audio.  I must admit that the idea of setting the buffers to max would seem to increase latency????

Your gaming example is, I guess as I'm no gamer, working basically on incoming streams.
This AOIP is working on outgoing streams. The speed of the computer is filling the buffers is much larger than the transmission speed over ethernet, you would have no risk of empty send buffers.
 
Though this recommendation comes from a Gaming site:
  Guide: Optimal Wired LAN adapter settings for latency (Windows)

Since I bought my Steam Link on Amazon and Steam won't let me post a guide unless I buy it through them, I'm posting this here:

If you are experiencing stuttering over wired, or you want to run games that have extremely low latency tolerances, here are a few settings in Windows you can tweak.

to change these, go to the network and sharing center, click on your Ethernet connection, Go to Properties, then Configure, then click the Advanced tab. Changing these settings will cause you to lose connectivity for a brief moment.

 
  1. Interrupt Moderation: This is the big one. Microsoft themselves[technet.microsoft.com]recommends turning this off for low-latency applications. Set this to Disabled.
  2. Recieve/Send Buffers: Some drivers do not have this set to the maximum level by default. Just click the up arrow on the value until it maxes out.
  3. Flow Control: Implementation is screwy from driver to driver and can mess up the Windows networking stack. If the above doesn't solve your problems, disable this.
  4. Receive Side Scaling: Ditto for the above.
  5. Energy Efficient Ethernet: Some driver implementations can cause the adapter to go into low power mode while running active applications. If you notice intermittent freezing or stuttering, disable this.
  6. Green Ethernet: Ditto for the above.

 
And also
 
Number of RX buffers = Maximum

This setting specifies the number of buffers used by the driver when copying data to the protocol memory. The recommended value is usually 16. In high network load situations, increasing receive buffers can increase performance. The tradeoff is that this also increases the amount of system memory used by the driver. If too few receive buffers are used, network performance will suffer. If too many receive buffers are used, the driver will unnecessarily consume memory resources. The valid range of values for Receive Buffers is 80-3000. TT recommends that this is set to the maximum (2048 for Intel and 3000 for Broadcom).

Number of TX buffers = Maximum

This parameter determines the number of transmit buffers that are allocated by the driver when loaded, and thus significantly impacts the resources the driver consumes in the system (allocated memory). More transmit buffers allow more send operations to be in progress at one time, and thus potentially increases system performance – this also consumes more system resources. Setting this parameter too high can prevent the driver from loading due to insufficient system resources and runs the risk of the NIC dropping packets. The range of possible values for this parameter is 80-2048. TT recommends this is set to the maximum (2048).


 
 
And
 
  1. Max Transmit/Send Descriptors and Send Buffers - This setting specifies how many transmit control buffers the driver allocates for use by the network interface. This directly reflects the number of outstanding packets the driver can have in its “send” queue. Set this value as high as possible for maximum performance. On servers with limited physical memory, this may have a negative impact as send buffers consume system memory. On most systems however, the maximum setting can be applied without significantly reducing available memory.
     
  2. Max Receive Buffers - This setting specifies the amount of memory buffer used by the network interface driver when copying data to the protocol memory. It is normally set by default to a relatively low value. Set this value as high as possible for maximum performance. On servers with limited physical memory, this may have a negative impact as receive buffers consume system memory. On most systems however, the maximum setting can be applied without significantly reducing available memory.

 
Oct 12, 2016 at 2:23 AM Post #2,163 of 3,694
Also, the first link reocmmends disabling offload task while all other links recommend enabling?
 
Oct 12, 2016 at 3:01 AM Post #2,164 of 3,694
Also, the first link reocmmends disabling offload task while all other links recommend enabling?


Yepp, some conflicting opinions there.

I have them (offload settings) all enabled.

But which Rednet do you have nowadays?
It is easier if you can try for real and see its effects.
 
Oct 12, 2016 at 3:29 AM Post #2,165 of 3,694
Hi Guys
After a couple of weeks go having my new Rednet 3 in my system, I thought I would share my opinions on it.
 
 
 

 
 

My System
Mac Mini with 12V Uptone Audio mod, SSD and 8 GIG RAM. LPS.
Audio Note DAC 5 Special (R-2R tubed DAC).
KGSShv Carbon Electrostatic headphone amplifier
Stax 009s
Passive Pre-amplifier
Plinius SA-13 Power Amplifier
Zingali Client Name EVO 1.2 Horn Hybrid speakers.
 
I play only 44.1 / RedBook AIFFs at bit prefect (no up sampling) with Audirvana+
 
Setting up
It took me 2 hours to set it up and download the software and set up the routing. It is basically pro-studio gear, so not so easy to set up.
 
If you are not buying the Dante Card, then you have 3 pieces of software to set up.
 
1. RedNet Controller
2. Dante Controller
3. Dante Virtual Sound Card
 
I use a Mac Mini and my setup was fairly straight forward. You need to hook the unit up to your system and switch it on. Then download the software from FocusRight, and Audinate for the 2 x Dante bits of software.
 
The allow the unit to update to the latest version (firmware). The software may need to update as well.

I got a bit stuck in Dante Controller routing the Mac source and RedNet SPDIF output. I got Sweetwater to do a remote access and go into my Mac and set that bit up, though seeing how easy it seas now could do it myself.
 
Playing Music
Playing music now WOW, hell, it is amazing. I don’t see many jumps up in sound quality this big ever.
 
I would say my previous USB stack (1.4K worth of M2Tech gear) was good as far as USB chains go I thought. But AOIP (audio over IP) is the way to go.
 
It is very very smooth, no digital edge, no bloated bass, massive soundstage and depth. Vocals and instrument stand right out in 3D front to back. Everything sound so real, I would call it liquid as opposed to digital.
 
I am playing super loud, windows rattling and sitting 4 feet from each speaker, but it sounds fabulous. I am stunned TBH. It is like getting super close to the master tape, that feeling I always had of listening to CD, or more recently listening to a PC feeding tunes is gone. The illusion of real music is so much better. It allows me to relax into the music and just enjoy it more, plus play it louder!
 
So, forget CDP or USB IMO
Ethernet is the way to go. This Rednet 3 inc software for a PC or Mac is 1K USD.
 
I am recommending it to anyone who asks on the forums now. 
 
The RedNet 3 has pro S/PDIF output and AES, so will hook up to any DAC. It can handle 44.1 up to 192K DSD.
 
USB is weak IMO because of:
1. It has a 5V feed in the lines, which even if the board in the DAC doesn’t use that (really bad idea if it does), it adds noise and distortion regardless
2. It was never designed for high end audio, only cameras, memory sticks and printers without needing a driver
3. It sends data in packets, with gaps, so this induces jitter which has then to be filtered out
4. The USB bus on a PC or a Mac is shared by other USB devices it uses, like a mouse, printers, keyboard, again causing noise and havoc , latency.
5. USB has poor latency, it gets going ok (USB2/3) but soon grinds to a slower rate causing latency issues
6. The last 5 years have seen a plethora of USB fixers, convertors, filters and cabling that add even more parts and problems to remove the grunge and high frequecny hash in USB and were only partly successful. Even the recent XIMOS convertors are still way behind Ethernet.
 
So IMO, USB is not good for true high level Audio.
 
Hope this helps folk to make the decision to go for Ethernet. I was nervous but it is way beyond my expectations. And TBH looking at other hifi add ons, is good value.
 
Oct 12, 2016 at 4:12 AM Post #2,166 of 3,694
   
It is very very smooth, no digital edge, no bloated bass, massive soundstage and depth. Vocals and instrument stand right out in 3D front to back. Everything sound so real, I would call it liquid as opposed to digital.

 
Good to hear that you are enjoying the RN3.
I can easily relate to this statement since that is exactly what I heard when I reported my findings back in May here: http://www.head-fi.org/t/803111/xmos-xu208-usb-bridges-the-latest-gen-has-arrived/1830#post_12606908
 
That REALISM is so compelling and really draws your utmost attention.
 
Cheers!
 
Oct 12, 2016 at 5:22 AM Post #2,167 of 3,694
  Hi Guys
After a couple of weeks go having my new Rednet 3 in my system, I thought I would share my opinions on it.

 
Very nice review, thanks for posting, and finally this thread gets some photos :) :)
Looks like a very impressive setup you have there. Rednet is really appealing, but I'm still holding off for something smaller and cheaper.
 
Oct 12, 2016 at 11:56 AM Post #2,168 of 3,694
  Playing Music
Playing music now WOW, hell, it is amazing. I don’t see many jumps up in sound quality this big ever.
 
I would say my previous USB stack (1.4K worth of M2Tech gear) was good as far as USB chains go I thought. But AOIP (audio over IP) is the way to go.

 
Thank you for your review. One aspect that caught my attention is that you only changed from usb stack to Rednet (versus changing several components at once) and that you are going direct from mac mini to rednet which is a greatly simplified system. The point is that in your instance we're not considering a host of variables which may have contributed to the improvement in sound quality; it must be the Rednet. 
 
I have a mac mini running Audirvana and perhaps this Aoip route is a worthy next step as soon as I figure out which gear I should choose. Like others here, 
I'm sort of waiting to see what "consumer" or prosumer AOIP devices emerge. I'm intrigued by the Arrakis simple AoIP--rb2013 mentioned this unit--because it has a DC input and the option for an AES ouput (AES or ethernet option). Ideally, I wish it had BOTH AES out and ethernet out as that would be more future proof. 
 
I'm also pondering what "loss" might exist for me given that I'm running a Comet Exogal and the Exogal's usb input slightly exceeds the capabilities of the AES input. And, the Comet regenerates the entire signal avoiding a lot of usb Gremlins. It's quite a sophisticaed FPGA process (well above my neophyte knowledge). 
 
 The Comet uses an FPGA (Field Programable Gate Array) for its DAC. That means that unlike the competition that uses a commercially available DAC chip from from companies like ESS, Wolfson or Burr brown, the Comet is all programmed by Exogal and is upgradable over the web. So it offers built in NON Obsolescence


 
 
Thank you for the post. 
 
PS I have no affiliation with Comet. I do love the dac, and the customer service has been excellent. 
 
Oct 12, 2016 at 1:36 PM Post #2,169 of 3,694
  Hi Guys
After a couple of weeks go having my new Rednet 3 in my system, I thought I would share my opinions on it.
 
  Playing Music
Playing music now WOW, hell, it is amazing. I don’t see many jumps up in sound quality this big ever.
 
I would say my previous USB stack (1.4K worth of M2Tech gear) was good as far as USB chains go I thought. But AOIP (audio over IP) is the way to go.
 
It is very very smooth, no digital edge, no bloated bass, massive soundstage and depth. Vocals and instrument stand right out in 3D front to back. Everything sound so real, I would call it liquid as opposed to digital.
 
I am playing super loud, windows rattling and sitting 4 feet from each speaker, but it sounds fabulous. I am stunned TBH. It is like getting super close to the master tape, that feeling I always had of listening to CD, or more recently listening to a PC feeding tunes is gone. The illusion of real music is so much better. It allows me to relax into the music and just enjoy it more, plus play it louder!
 
So, forget CDP or USB IMO
Ethernet is the way to go. This Rednet 3 inc software for a PC or Mac is 1K USD.
 
I am recommending it to anyone who asks on the forums now. 
 
The RedNet 3 has pro S/PDIF output and AES, so will hook up to any DAC. It can handle 44.1 up to 192K DSD.
 
USB is weak IMO because of:
1. It has a 5V feed in the lines, which even if the board in the DAC doesn’t use that (really bad idea if it does), it adds noise and distortion regardless
2. It was never designed for high end audio, only cameras, memory sticks and printers without needing a driver
3. It sends data in packets, with gaps, so this induces jitter which has then to be filtered out
4. The USB bus on a PC or a Mac is shared by other USB devices it uses, like a mouse, printers, keyboard, again causing noise and havoc , latency.
5. USB has poor latency, it gets going ok (USB2/3) but soon grinds to a slower rate causing latency issues
6. The last 5 years have seen a plethora of USB fixers, convertors, filters and cabling that add even more parts and problems to remove the grunge and high frequecny hash in USB and were only partly successful. Even the recent XIMOS convertors are still way behind Ethernet.
 
So IMO, USB is not good for true high level Audio.
 
Hope this helps folk to make the decision to go for Ethernet. I was nervous but it is way beyond my expectations. And TBH looking at other hifi add ons, is good value.


Nice review!  Thanks for posting this
 
Oct 12, 2016 at 2:01 PM Post #2,170 of 3,694
   
Thank you for your review. One aspect that caught my attention is that you only changed from usb stack to Rednet (versus changing several components at once) and that you are going direct from mac mini to rednet which is a greatly simplified system. The point is that in your instance we're not considering a host of variables which may have contributed to the improvement in sound quality; it must be the Rednet. 
 
I have a mac mini running Audirvana and perhaps this Aoip route is a worthy next step as soon as I figure out which gear I should choose. Like others here, 
I'm sort of waiting to see what "consumer" or prosumer AOIP devices emerge. I'm intrigued by the Arrakis simple AoIP--rb2013 mentioned this unit--because it has a DC input and the option for an AES ouput (AES or ethernet option). Ideally, I wish it had BOTH AES out and ethernet out as that would be more future proof. 
 
I'm also pondering what "loss" might exist for me given that I'm running a Comet Exogal and the Exogal's usb input slightly exceeds the capabilities of the AES input. And, the Comet regenerates the entire signal avoiding a lot of usb Gremlins. It's quite a sophisticaed FPGA process (well above my neophyte knowledge). 
 

 
 
Thank you for the post. 
 
PS I have no affiliation with Comet. I do love the dac, and the customer service has been excellent. 

Nice looking DAC - the FPGA's are not processing USB - but pre-processing the digital bit stream before sending on to BB DAC chips
 
The USB processing is handled by a XMOS chip  - and not the latest gen XU208 (like the Singxer F-1, or SU-1), or XU216(DIYinHK Pro3z).  But the older XMOS U8.  It is hard to see if the XMOS USB has it's own PS feed - but DAC internal USB designs can have issues with 'packet noise' making it;s way back into the PS that feeds the DAC clocks.  Not good.  Almost everyone who has tried an external XMOS USB DDC - agrees they sound better then the internal board (including the Yggie Gen3). At least this was commonly reported on my XU208 thread.  You do give up the highest SR though.
 
The AOIP Dante should be a major step-up in SQ - even when feeding the AES digital input.
http://hifilive.es/2016/03/23/exogal-comet-2/
 

 
Oct 12, 2016 at 3:34 PM Post #2,172 of 3,694
I'm wondering, if I don't have drop outs, whether further tuning my Ethernet settings would improve sound quality or not.

Opinions?

Thanks.

Joel

It's to my understanding that if you're not getting dropouts and keep a low latency - below software settings (actual latency can be checked with Rednet software) then no. Unless you're using the Rednet in recording where you want minimum latency between gear then I see no gains. Doesn't hurt to play with the settings though, just make sure you take note of your 'good' settings so you can always revert to the combination.
 
Oct 13, 2016 at 4:52 AM Post #2,173 of 3,694
   
Good to hear that you are enjoying the RN3.
I can easily relate to this statement since that is exactly what I heard when I reported my findings back in May here: http://www.head-fi.org/t/803111/xmos-xu208-usb-bridges-the-latest-gen-has-arrived/1830#post_12606908
 
That REALISM is so compelling and really draws your utmost attention.
 
Cheers!

I too agree that the sense of realism is compelling and dare I say it, addictive.
It can suck you in, to a degree that until you do experience it first hand, a simple written description does not adequately convey the experience itself.
EVERYTHING you have ever heard in your music library will sound enticing and compelling and fascinating, even the poorly recorded/mastered/performed music you NEVER play because it was just not up to snuff.
Perhaps especially this entire section of your music library, will delight you even more, perhaps even more than the 'good stuff'.
 
This level of inner detail and focus can, in one sense, FORCE your attention into the music to the exclusion of all else.
SuperDuperGlue indeed.
 
And on a side note I will be adding a Mutec 3+ into the system hopefully this weekend.
Does anyone know what value of fuse it uses?  
atsmile.gif

 
JJ
 
Oct 13, 2016 at 8:06 PM Post #2,175 of 3,694
That looks slick!! I may follow up on the plastidip -- trick is waiting enough time between coats and getting 6-7 costs so peeling is easier (if you want to go back to red).

Revisiting the "make RedNet into BlackNet" idea again
 
Has anyone here successfully made a D16AES faceplate black?  If so, what is the best and sanest way to go about doing this?
 
Is the faceplate easily removable?  If so, it would seem spraying plastidip or paint might work.  If not, are we stuck with masking and spraying?
 
Any info/guidance greatly appreciated.
 
I wish Focusrite would sell black faceplates -- because I believe only the faceplate itself is red, the casework and enclosure is all black anyway.
 

Users who are viewing this thread

Back
Top