A novel way to massively improve the SQ of computer audio streaming

the_doc735 · October 18, 2018

@marce

Roughly translated it says:

Soon available Scrion

The PinkFaun Scrion is a Roon endpoint

Equipped with a linear supercap power supply

Standard outputs AES/EBU, coax and optical

Expandable with multiple digital output variants such as i2s-TTL or LVDD

...more LPS and not SMPS.

seeteeyou · October 18, 2018

Cool, they just added LDAC support for Linux recently so we could stream 24/96 files from an Intel NUC to Sony DMP-Z1 / TEAC NT-505 etc. via bluetooth

https://github.com/Arkq/bluez-alsa/commit/3a381ef3d81a7d363e9f9067050258ef8f1fcd7a

https://github.com/Arkq/bluez-alsa/commit/a849756c1de5a418d772908ddb8abc313facae9d

misterspense · October 18, 2018

18 hours ago, lmitche said:

Roy, I don't know enough about the differences between Intel and ARM cpu designs to offer an informed opinion. My conjecture is that ARM CPUs most often operate in mobile environments and Intel designs in more complex power and connectivity environments on desktops and in datacenters, with laptops an exception. If it is correct that Intel cpus are designed for harsher environments it wouldn't surprise me if there are engineering differences that harden the Intel design. Perhaps those translate into a SQ advantage?

The Asus Tinker board looks a lot like a Raspberry Pi. I have almost zero experience with this class of machine with audio except for the old SMS-100 and a short time with the original microrendu. Both are greatly surpassed by today's standards. It will be fun to learn more about the Asus solution over time.

We can't discount the impact of a low latency OS in either environment. I don't know of a low latency OS in the ARM world, but haven't done much, if any research into this topic.

There is a Audiolinux version available for the Beaglebone black, that has an ARM processor. Don't think it will run in RAM mode, due to the low amounts of it in the BBB. But Audiolinux focusses on low latency.

jabbr · October 18, 2018

16 hours ago, seeteeyou said:

The term low latency could be used quite casually, sometimes they might simply use that for marketing purpose without stating what the value is or maybe in other cases how a particular value is obtained.

Here's the latest from Mellanox with hundreds of nanoseconds

http://www.mellanox.com/page/press_release_item?id=575

http://www.mellanox.com/blog/2018/05/ethernet-storage-fabric-part-2/

http://www.mellanox.com/related-docs/solutions/SB-breaking-the-low-latency-trading-barrier-with-next-gen-intelligent-interconnect.pdf#page=2

Though their latest ConnectX-6 EN (Ethernet adapter) turned out to be much higher

http://www.mellanox.com/page/products_dyn?product_family=268&mtag=connectx_6_en_ic

Then there's always something like this

https://blog.metamako.com/how-to-measure-the-latency-of-a-4ns-switch

The Solarflare 7322F (which I have several of) has a rather low phase error onboard clock: https://www.solarflare.com/Media/Default/PDFs/SF-111380-CD-LATEST_Solarflare_SFN7322F_Product_Brief.pdf

Solarflare is known for low latency.

That switch looks nice but I can't justify that price ... all the 10 Gbe gear is pretty good

austinpop · October 18, 2018

@flkin - thanks for your valuable impressions of the Pink Faun 2.16 and Antipodes CX/EX servers.

And Roy, it's always fascinating to hear about your continuing saga! I feel you and Larry are on track to unlock another "massive" leap in sonic performance.

23 hours ago, romaz said:

During this time, I also had access to SOtM's latest sMS-200ultra Neo. When I last owned an sMS-200ultra, it was before I owned the Ref10 and so I felt it was important to revisit the latest incarnation of the sMS-200ultra to see how it would compare to the NUC. As a similarly small device designed from the ground up for audio, I always felt these types of devices should rule the world. WIth the sMS-200ultra Neo connected via Habst cable to my Ref10, even though I no longer had an original sMS-200ultra on hand for direct comparison, I very much liked what I heard and it was indeed better than what I recall my sMS-200ultra sounded like. Where I felt the original sMS-200ultra sounded incisive, airier, and more detailed compared against my microRendu, there was a beguiling smoothness and naturalness to the microRendu that was also very pleasant. What is interesting is that the ultraRendu has gone more in the direction of the sMS-200ultra and in some ways, sounds even more detailed and more mechanical whereas the new Neo has gone more in the direction of the original microRendu as it has given up some of its incisiveness for a harmonically richer and more natural sounding presentation. In isolation and when powered by my SR7, my immediate inclination is that the new Neo is a winner and it definitely is.

Comparison of sMS-200ultra vs sMS-200ultra Neo

In the flurry of activity pre- and post-RMAF, I have not had a chance to report on a comparison Eric @limniscate and I did. We were able to compare the sMS-200ultra Neo (the same one Roy mentioned above) with Eric's vanilla unit. Both units have a reference clock input, so we were able to run both with a reference clock from the Mutec Ref-10. We did this in my system, so the chain was:

Vanilla: Router > TLS OCXO switch > Zenith SE (Roon Server) > TLS OCXO switch > sMS-200ultra > tX-USBultra > DAC
Neo: Router > TLS OCXO switch > Zenith SE (Roon Server) > TLS OCXO switch > sMS-200ultra Neo > tX-USBultra > DAC

In both cases, the sMS was running the Roon Ready application. The nice thing about this test is we did not have to rely on auditory memory to do this comparison; we were able to A/B them in real time.

Sure enough - there is a difference in sound signature between the two. I would not characterize this as a large, knock-your-socks-off difference. But certainly, SOtM have managed to alter the tonality of the Neo to be mellower, and not as "bright." I think Roy is spot on when he says it's reminiscent of the smoother tonality the original mR has. The Neo has smoother treble, and slightly improved lower mids and upper bass. As I said, it's reasonably subtle, but very welcome.

The second finding is even more interesting. May had suggested I try their beta firmware, where they have a new kernel that they have tweaked. She wouldn't (and hasn't) tell me what exactly has changed, but I have to say, I felt the beta version sounder better for sure. In fact, it seemed to be doing the things Roy and Larry are reporting with their experiments - more dynamic and more expansive.

If you want to try it - AT YOUR OWN RISK, obviously - here are the instructions from May. I strongly recommend you make a backup of your SD card before you do.

1. Go to the following link

URL: <your eunhasu ip address>/beta.php

2. Enable the Beta server and save.

3. Go to the Eunhasu upgrade page.

URL: <your eunhsu ip address>/upgrade.php

4. Click the kernel check. After that, reboot the device and test.

After testing,

5. Go to the beta page again.

URL: <your eunhasu ip address>/beta.php

6. Disable the Beta server and save.

7. Reboot the unit and use.

flkin · October 19, 2018

3 hours ago, austinpop said:

@flkin - thanks for your valuable impressions of the Pink Faun 2.16 and Antipodes CX/EX servers.

And Roy, it's always fascinating to hear about your continuing saga! I feel you and Larry are on track to unlock another "massive" leap in sonic performance.

Comparison of sMS-200ultra vs sMS-200ultra Neo

In the flurry of activity pre- and post-RMAF, I have not had a chance to report on a comparison Eric @limniscate and I did. We were able to compare the sMS-200ultra Neo (the same one Roy mentioned above) with Eric's vanilla unit. Both units have a reference clock input, so we were able to run both with a reference clock from the Mutec Ref-10. We did this in my system, so the chain was:

Vanilla: Router > TLS OCXO switch > Zenith SE (Roon Server) > TLS OCXO switch > sMS-200ultra > tX-USBultra > DAC

Neo: Router > TLS OCXO switch > Zenith SE (Roon Server) > TLS OCXO switch > sMS-200ultra Neo > tX-USBultra > DAC

In both cases, the sMS was running the Roon Ready application. The nice thing about this test is we did not have to rely on auditory memory to do this comparison; we were able to A/B them in real time.

Sure enough - there is a difference in sound signature between the two. I would not characterize this as a large, knock-your-socks-off difference. But certainly, SOtM have managed to alter the tonality of the Neo to be mellower, and not as "bright." I think Roy is spot on when he says it's reminiscent of the smoother tonality the original mR has. The Neo has smoother treble, and slightly improved lower mids and upper bass. As I said, it's reasonably subtle, but very welcome.

The second finding is even more interesting. May had suggested I try their beta firmware, where they have a new kernel that they have tweaked. She wouldn't (and hasn't) tell me what exactly has changed, but I have to say, I felt the beta version sounder better for sure. In fact, it seemed to be doing the things Roy and Larry are reporting with their experiments - more dynamic and more expansive.

If you want to try it - AT YOUR OWN RISK, obviously - here are the instructions from May. I strongly recommend you make a backup of your SD card before you do.

1. Go to the following link

URL: <your eunhasu ip address>/beta.php

2. Enable the Beta server and save.

3. Go to the Eunhasu upgrade page.

URL: <your eunhsu ip address>/upgrade.php

4. Click the kernel check. After that, reboot the device and test.

After testing,

5. Go to the beta page again.

URL: <your eunhasu ip address>/beta.php

6. Disable the Beta server and save.

7. Reboot the unit and use.

Thanks Rajiv, good comparison with the new sMS200 Evo. It always amazes me how one tiny component can make such a difference but I'm glad that SOtM is still pushing the existing model with these kinds of mods and keeping the hardware fresh.

I'm going to have to revisit using the the sMS200evo as a NAA again with the new software. Sounds promising!

Bruce Orr · October 19, 2018

Rajiv - I don’t think you use a sMS Vsnilla/Neo in your regular setup do? You use the SE for both Core & Roon Ready and then strait into the tx-USB ultra?

I’m taking a step back here, but does the idea @romaz presented - and that is at play in your neo vs. vanilla test - that adding the additional switch and Roon endpoint after the Server (Core) improves SQ?

If I recall romaz preferred the mR to the uR in this configuration?

flkin · October 19, 2018

Some more info about the Pink Faun Scion device.

The Scion is a ROON endpoint designed to replace both Pink Faun's entry model audio streamer and mid-end model streamer 3.4. The older units were starting not to be commercially viable with the motherboards costing too much to make. In effect the Scion will operate much like the sMS200evo or the UltraRendu endpoints except I think Pink Faun is aiming higher.

It's an entry model to be used with a NAS or other PC/Mac streamer for a two box solution. A low noise OCXO in the Scion will be made as an upgrade too. The power supply is a Super Capacitor PSu containing 12 Farad! This super capacitor tech is supposed to provide very fast power response. In the discussion with Jord, I get the feeling that he intends to incorporate super caps into their other offerings as well over time.

Price is not fixed yet but should be not more than €2000 Inc. Tax.

The plan is that in a couple of years a higher end model for a two box solution to the 2.16x server will be coming but not for the time being.

austinpop · October 19, 2018

39 minutes ago, Bruce Orr said:

Rajiv - I don’t think you use a sMS Vsnilla/Neo in your regular setup do? You use the SE for both Core & Roon Ready and then strait into the tx-USB ultra?

I’m taking a step back here, but does the idea @romaz presented - and that is at play in your neo vs. vanilla test - that adding the additional switch and Roon endpoint after the Server (Core) improves SQ?

If I recall romaz preferred the mR to the uR in this configuration?

Bruce,

I didn't mean to confuse you. You're right, I no longer use an sMS-200ultra. I use the SE straight to tX-USBultra. The only reason for the above comparison was to directly compare the vanilla to Neo - because it was possible - as a service to the community. Most people who upgrade don't have the old unit still in hand to do this kind of comparison.

The NUC approach that Roy is taking goes a step beyond the sMS-200ultra, by using an in-memory OS using the ramboot feature of audiolinux. As Roy said, it's not clear why endpoints like the sMS or the uR do not give the same level of performance as his/Larry's ramboot NUCs, although we suspect it has to do with the low latencies of operations when running the OS in memory.

seeteeyou · October 19, 2018

On 10/17/2018 at 12:43 PM, romaz said:

Unless it's a typo, it looks like the new Sound Galleries Evo will have a storage capacitance of 3.76 million uF. Let the server PSU wars begin.

I went back and read again, then I was still scratching my head. 3.76 million µF might be considered fairly high for relatively small capacitors, larger ones from Nichicon could go up to 2.2 million µF

http://www.nichicon.co.jp/english/products/alm_larg/index.html

And then there are supercapacitors like what Rob Watts were putting inside the original Hugo TT as well as Hugo TT 2, they didn't even tell us anything about Hugo TT 2 but it's 10 million µF for Hugo TT

https://chordelectronics.co.uk/product/hugo-tt/

Quote

Super capacitor energy storage: 10,000,000μF

There were only 2 pieces of supercapacitors inside Hugo TT and exact model is PHB-5R0V505-R @ 5 million µF

https://www.digikey.com/product-detail/en/eaton/PHB-5R0V505-R/283-3520-ND/2770536

http://www.cooperindustries.com/content/public/en/bussmann/electronics/products/eaton-supercapacitors/cylindrical/phb-series.html

Basically some audiophiles in Taiwan ordered like two dozens of PHB-5R0V505-R (instead of two) plus 16 pieces of Rubycon capacitors as I mentioned below, then they modified the original Hugo accordingly and it's sounding as good as Hugo TT afterwards

https://www.head-fi.org/threads/chord-hugo.702787/page-826

Then there's something from Gryphon with 12.5 million µF

http://www.gryphon-audio.dk/products/da-converters/kalliope.aspx

Quote

12.5 Farad SuperCap powersupply for the Kalliope USB module which acts like a true battery supply

In addition to Mini Pure DC-4-EVR from Vinnie Rossi (discontinued) and LPS-1 / LPS-1.2 from UpTone Audio, we could also find 2 × 310 million µF ultracapacitors here

http://audiobyte.net/products/hydra-zpm

Quote

The Hydra ZPM makes a step further, letting behind the Hydra X concept, which was battery powered. Instead, ZPM is using a ultracapacitor bank consisting of 2 x 310 Farads (1 Farad equals 1.000.000 uF) which will provide lower internal resistance and longer life and reliability, compared to a battery solution.

Please pardon my ignorance here. Even though supercapacitors / ultracapacitors could offer a MUCH higher storage capacitance, are they only capable of providing a pretty low voltage + amperage? For instance, the ones from Eaton could only go up to 5V or so while we're getting merely 2.5V from Nichicon

https://www.mouser.com/new/nichicon/nichiconevercap/

https://www.mouser.com/search/refine.aspx?Ntk=P_MarCom&Ntt=145879225

If that were the case, most likely they're no good for something as hungry as Xeon or Ryzen etc.

OTOH, LPS-1.2 could only reach 12V so we'll need two of them in order to achieve 19V or 24V for an Intel NUC. If we're running headless AudioLinux in ramboot mode, most likely we could get away with either 20.9W or 26.4W.

Now the $64,000 question is, where do we actually get those Mundorf caps with the highest capacitance?

http://www.mundorf.com/en/?category=hifi&menu=caps_power&content=mlytic_ag_2pin

http://www.mundorf.com/en/?category=hifi&menu=caps_power&content=mlytic_ag_4pin

Quote

470µF to 3 300 000µF

Dr. Sean Jacobs should have picked Mundorf for his ToTL PSU for a reason, it's just a matter of finding out if going all the way up to 3.3 million µF were actually worth it and of course the size of that thing.

romaz · October 19, 2018

22 hours ago, seeteeyou said:

And then the BEST part is an internal (lower impedance) Molex connector in addition to an external (higher impedance) DC connector, therefore it should be easier to try power cables with higher gauge number / lower resistance.

Whoa, thanks for bringing this up! Larry and I had been discussing trying out the Intel NUC7i7DNBE board as a RoonServer:

https://www.intel.com/content/www/us/en/products/boards-kits/nuc/boards/nuc7i7dnbe.html

I actually selected this particular board purely because it is only 1 of 2 i7 NUC boards that Akasa is making a fanless chassis for at this time. This is a commercial class NUC board and as you stated, it accepts 12-24V but what is more important to me and what I didn't realize until you brought it up is that this is one of only a few Intel NUC boards that can be powered via a 2x2 Molex connector which is a better alternative to the high impedance 5.5mm x 2.5mm barrel connectors that most NUCs and small audio endpoints use.

22 hours ago, seeteeyou said:

NUC7i3DN and NUC7i5DN came with only 1.5MB cache per core, and we're still getting 2MB cache per core from NUC7i7DN.

This is not entirely accurate but the news is good. The NUC7i3DN and NUC7i5DN you mentioned above actually incorporate 3MB of SmartCache. Probably for cost reasons, most Celerons and budget CPUs incorporate L2 or L3 caches that are not SmartCaches which means the 4MB of cache used by the Celeron J4005 CPU in my NUC shares this cache evenly among its cores. Since this is a 2-core Celeron, then each core gets 2MB of cache. If only a single core is being utilized by an app, that core can only access 2MB of cache and not the entire 4MB.

The advantage of Intel's SmartCache is the entire cache can be used by whichever cores are active and so with a single-threaded app that uses only a single core, that single core now has the potential of accessing the entire cache. With the NUC7i3DN or NUC7i5DN, this means a single core can access the entire 3MB of cache and generally speaking, the larger the cache, the lower the CPU latency. With the NUC7i7DNBE board I have been eyeing, the i7-8650U CPU that it utilizes has a very healthy 8MB of SmartCache!

What is further appealing with the NUC7i7DNBE board is that the CPU's TDP can be downrated to 10w just like my current Celeron NUC. What I am hoping is that SpeedStep and Turbo can also be turned off through the BIOS. If this is possible, this would allow me to very closely mimic the Celeron CPU in my NUC but with a much larger and smarter cache. This would mean that this board could be well suited as both a RoonServer board and a RoonBridge board. As a RoonServer board, as this board has the capacity for 32GB of RAM, it should be possible to store both the entire OS as well as the entire Roon database in memory allowing for a completely diskless RoonServer. Obviously, this would mean you would need a NAS to store your music files but at this time, this is my preference anyway provided I have access to a good network switch.

RickyV · October 19, 2018

43 minutes ago, seeteeyou said:

I went back and read again, then I was still scratching my head. 3.76 million µF might be considered fairly high for relatively small capacitors, larger ones from Nichicon could go up to 2.2 million µF

http://www.nichicon.co.jp/english/products/alm_larg/index.html

And then there are supercapacitors like what Rob Watts were putting inside the original Hugo TT as well as Hugo TT 2, they didn't even tell us anything about Hugo TT 2 but it's 10 million µF for Hugo TT

https://chordelectronics.co.uk/product/hugo-tt/

There were only 2 pieces of supercapacitors inside Hugo TT and exact model is PHB-5R0V505-R @ 5 million µF

https://www.digikey.com/product-detail/en/eaton/PHB-5R0V505-R/283-3520-ND/2770536

http://www.cooperindustries.com/content/public/en/bussmann/electronics/products/eaton-supercapacitors/cylindrical/phb-series.html

Basically some audiophiles in Taiwan ordered like two dozens of PHB-5R0V505-R (instead of two) plus 16 pieces of Rubycon capacitors as I mentioned below, then they modified the original Hugo accordingly and it's sounding as good as Hugo TT afterwards

https://www.head-fi.org/threads/chord-hugo.702787/page-826

Then there's something from Gryphon with 12.5 million µF

http://www.gryphon-audio.dk/products/da-converters/kalliope.aspx

In addition to Mini Pure DC-4-EVR from Vinnie Rossi (discontinued) and LPS-1 / LPS-1.2 from UpTone Audio, we could also find 2 × 310 million µF ultracapacitors here

http://audiobyte.net/products/hydra-zpm

Please pardon my ignorance here. Even though supercapacitors / ultracapacitors could offer a MUCH higher storage capacitance, are they only capable of providing a pretty low voltage + amperage? For instance, the ones from Eaton could only go up to 5V or so while we're getting merely 2.5V from Nichicon

https://www.mouser.com/new/nichicon/nichiconevercap/

https://www.mouser.com/search/refine.aspx?Ntk=P_MarCom&Ntt=145879225

If that were the case, most likely they're no good for something as hungry as Xeon or Ryzen etc.

OTOH, LPS-1.2 could only reach 12V so we'll need two of them in order to achieve 19V or 24V for an Intel NUC. If we're running headless AudioLinux in ramboot mode, most likely we could get away with either 20.9W or 26.4W.

Now the $64,000 question is, where do we actually get those Mundorf caps with the highest capacitance?

http://www.mundorf.com/en/?category=hifi&menu=caps_power&content=mlytic_ag_2pin

http://www.mundorf.com/en/?category=hifi&menu=caps_power&content=mlytic_ag_4pin

Dr. Sean Jacobs should have picked Mundorf for his ToTL PSU for a reason, it's just a matter of finding out if going all the way up to 3.3 million µF were actually worth it and of course the size of that thing.

Hi c2u,

You can't compare the Sean Jacobs psu/ conventional psu to a super/ultra capacitor psu as far as capacitance goes. Does are totally different beasts.

Rick?

seeteeyou · October 19, 2018

8 minutes ago, romaz said:

As a RoonServer board, as this board has the capacity for 32GB of RAM, it should be possible to store both the entire OS as well as the entire Roon database in memory allowing for a completely diskless RoonServer. Obviously, this would mean you would need a NAS to store your music files but at this time, this is my preference anyway provided I have access to a good network switch.

There's always ACARD with up 64GB of DDR2 (via SATA) if we're so inclined, though we might have to take compatibility / reliability into consideration

https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?page=396&tab=comments#comment-886052

True latency wouldn't be all that different when compared to DDR4, though going through SATA should most likely take a hit even if both CPU and chipset were integrated inside the SoC

http://www.crucial.com/usa/en/memory-performance-speed-latency

seeteeyou · October 19, 2018

2 hours ago, romaz said:

Whoa, thanks for bringing this up! Larry and I had been discussing trying out the Intel NUC7i7DNBE board as a RoonServer:

https://www.intel.com/content/www/us/en/products/boards-kits/nuc/boards/nuc7i7dnbe.html

I actually selected this particular board purely because it is only 1 of 2 i7 NUC boards that Akasa is making a fanless chassis for at this time.

You're very welcomed. There's yet another bonus because of these slots

https://www.intel.com/content/dam/support/us/en/documents/mini-pcs/nuc-kits/NUC7i7DN_TechProdSpec.pdf#page=15

Quote

M.2 2230 Module Connector (Key Type E)

M.2 2280 Module Connector (Key Type M)

Fortunately both of them are connected to PCIe lanes, therefore we could actually give tX-USBexp and/or NET Card FEMTO a shot as mentioned below

https://www.techinferno.com/index.php?/forums/topic/10209-m2-link-via-pe4c-v41-experiences/&tab=comments#comment-146893

Quote

Three weeks ago I purchased the skull canyon nuc. As I expected Intel graphics are fine for the browser and desktop, but not for games.

In anticipation of this I had researched alternatives.

I ordered (from California USA) the bplus v4.1 m.2 to pcie egpu adapter on Saturday, and it arrived on Thursday via DHL.

I connect ed the m.2 card, plugged in my gtx 750ti, moved the HDMI cable, added my atx power supply and it worked the first time and every time with no modifications. Even sleep works.

Also good for laptops so most likely we're doing fine here

http://forum.notebookreview.com/threads/egpu-results-clevo-p870dm-g-gtx1080-tb3-hdk-pe4c-v4-1-m-2-ngff.796773/

https://imgur.com/a/BtVwJ

As usual they aren't the cheapest options, PE4C-M2060A V4.1 would cost $140 and PE4C-M4060A V4.1 would cost $160 respectively

http://www.bplustech.com/Adapter/PE4C V4.1.html

These guys are untested at the moment but the price is right

https://www.aliexpress.com/item/Riser-M-2-wifi-A-E-Key-To-PCIe-x4-Extender-Adapter-Card-Riser-Cable-5cm/32840447397.html

https://www.aliexpress.com/item/M-2-NGFF-NVMe-M-Key2280-To-PCIe-3-0-4x-Riser-Card-Cable-PCI-Express/32833359557.html

Even cheaper ones as linked below but they're only good for the 2280 slot with Key Type M, not to mention the awkward position with the fanless case opened

https://www.amazon.com/Leboo-Key-M-Adapter-Converter-Riser/dp/B075WZY3LY

https://www.microsatacables.com/pci-e-1x-4x-card-to-m-2-m-key-4-lane-pcie-slot-adapter

https://www.fasttech.com/product/2401800-ep-091-m-2-ngff-to-pcie-x4-desktop-pcba-coverter

Forehaven · October 19, 2018

System report update: Larry and I have spent the last week changing from using my single high power/upsampling machine to be used as a Roon Server, and the NUC as the Roon Bridge. From what I felt was a super optimized, shielded and powered cleanly machine that had taken my system to a new level of enjoyment. So at first, after reading about the use of NUC's and Ram booted Audiolinux, I was a bit skeptical. But after hearing from Larry how superior the new NUC sounded, I bit the bullet so to speak.

I want to Thank Larry so much for all the help! I couldn't have tried this without his generous time! (further optimization to incl. better than stock power to the NUC, LPS 1 powered USB card in the big machine and the reintroduction of my IsoR which is being repaired) Even with these 'limitations', the SQ has taken a stupendous step forward. Improved presence, harmonic richness, detail, dead black background, and an added sense of reality in voices and instruments. The slightest inhale, pluck, spitter from horns...all incredibly REAL now!

Thanks to Roy and Larry for bringing this new approach to audio nirvana to CA, and my home!!

Chris

Confused · October 19, 2018

17 hours ago, austinpop said:

Sure enough - there is a difference in sound signature between the two. I would not characterize this as a large, knock-your-socks-off difference. But certainly, SOtM have managed to alter the tonality of the Neo to be mellower, and not as "bright." I think Roy is spot on when he says it's reminiscent of the smoother tonality the original mR has. The Neo has smoother treble, and slightly improved lower mids and upper bass. As I said, it's reasonably subtle, but very welcome.

Thanks for report Rajiv. Romaz's comments re the mR reminded me of my own post when first trying the Neo, where I found the Neo to be a bit like an mR but with more details, there is certainly some commonality with what we are hearing. An exact quote from my earlier post "First impressions though, it sounds like a more dynamic higher resolution version of the microRendu. I really like that old microRendu, so that is very good first impressions."

Anyway, I have a question. When feeding the tX-USBultra, how would you say the Neo compares with your SE? I am presuming the SE is still ahead but the Neo has closed the gap a bit? Can you advise any observations regarding relative sonic performance and characteristics?

austinpop · October 19, 2018

34 minutes ago, Confused said:

Thanks for report Rajiv. Romaz's comments re the mR reminded me of my own post when first trying the Neo, where I found the Neo to be a bit like an mR but with more details, there is certainly some commonality with what we are hearing. An exact quote from my earlier post "First impressions though, it sounds like a more dynamic higher resolution version of the microRendu. I really like that old microRendu, so that is very good first impressions."

You're welcome. Yes, my findings agreed with yours and others who reported this. You guys were more impressive, as you noticed this from memory. I had the advantage of a direct A/B.

34 minutes ago, Confused said:

Anyway, I have a question. When feeding the tX-USBultra, how would you say the Neo compares with your SE? I am presuming the SE is still ahead but the Neo has closed the gap a bit? Can you advise any observations regarding relative sonic performance and characteristics?

We did look at this. No, even with the beta firmware on the sMS-200ultra Neo, the SE direct to the tX-USBultra still sounded better. I have yet to experience the nirvana that Roy and Larry are reporting.

austinpop · October 19, 2018

41 minutes ago, Confused said:

Can you advise any observations regarding relative sonic performance and characteristics?

Ah I didn't answer this. The SE direct to the tX-USBultra is better in 2 key ways:

dynamics: music sounds more energetic and alive
darker, calmer, more relaxed sounding. Some call this a wetter or creamier sound. It just makes you sigh and listen deeper and longer.

Dev · October 19, 2018

I followed Larry's route to NUC based streamer approach and sound quality is a lot better than the Jetway mobo I was using earlier (both were powered using Mojo audio Illuminati LPS). A big thanks to Larry.

Currently I am using a bridged network on my server which is an old 6th generation i7 mobo having a 4-port Intel server NIC and powered from smps. The plan is to replace the NIC with Jcat NET card (on order and scheduled to receive next week) and power it using a quality LPS.

Since I upsample everything to DSD256, I needed a bit of horsepower and looked at the NUC7i7DNBE earlier. This would be perfect for my application and if I can attach the NET card, like this:

It would be ideal if Akasa can accommodate a PCI slot in their fanless case. I wrote to them if they would consider making a model but didn't hear back yet.

seeteeyou · October 20, 2018

16 hours ago, romaz said:

The advantage of Intel's SmartCache is the entire cache can be used by whichever cores are active and so with a single-threaded app that uses only a single core, that single core now has the potential of accessing the entire cache.

For the NUC form factor, most likely 8MB SmartCache from an i7-8650U should be as good as it gets for now.

For 3.5" SBC form factor, there's Avalue ECM-CFS with 4-pin Molex internal power connector

http://linuxgizmos.com/intel-coffee-lake-appears-on-3-5-inch-sbc/

https://www.avalue.com.tw/Product/Embedded-PC/Single-Board-Computer/3.5"/ECM-CFS_2709

We're getting 12MB SmartCache from an i7-8700T (25W with TDP-down) and the price is about $305 at the moment

https://www.provantage.com/intel-cm8068403358413~7ITEP5YR.htm

https://ark.intel.com/products/129948/Intel-Core-i7-8700T-Processor-12M-Cache-up-to-4-00-GHz-

Next up is an unlocked i9-9900K with 16MB SmartCache and 95W TDP, it would be interesting to see what's happening to power consumption when 7 outta 8 cores are disabled. We only have Mini-ITX motherboards with Z390 chipset at the moment, not sure if there were any Micro-STX or Mini-STX in the future

https://ark.intel.com/products/186605/Intel-Core-i9-9900K-Processor-16M-Cache-up-to-5-00-GHz-

There's a little birdie ASRock X299E-ITX/ac for i9-9980XE with 24.75MB SmartCache / 165W TDP

https://ark.intel.com/products/189126/Intel-Core-i9-9980XE-Extreme-Edition-Processor-24-75M-Cache-up-to-4-50-GHz-

Somehow they still managed to keep unlocked 28-core processors (38.5MB SmartCache / 255W TDP) running without a single fan

https://wccftech.com/intel-28-core-cpu-lga-3647-socket-6-channel-memory-aorus-motherboard/

https://www.asus.com/Motherboards/ROG-Dominus-Extreme/

https://wccftech.com/asus-rog-dominus-motherboard-skylake-x-28-core-cpu-announced/

How do we take advantage of dual 24-pin ATX connectors?

seeteeyou · October 20, 2018

https://www.supermicro.com/manuals/motherboard/C620/MNL-1907.pdf#page=78

Quote

LLC (Last Level Cache) Prefetch

If set to Enable, the hardware prefetcher will prefetch streams of data and instructions from the main memory to the L3 cache to improve CPU performance. The options are Disable and Enable.

https://www.supermicro.com/manuals/motherboard/X299/MNL-2001.pdf#page=97

https://www.supermicro.com/manuals/motherboard/X299/MNL-2078.pdf#page=79

Quote

LLC Prefetch

This feature enables LLC Prefetch. The options are Disabled or Enabled.

https://medium.com/software-design/why-software-developers-should-care-about-cpu-caches-8da04355bb8a

Quote

L1 cache access latency: 4 cycles
L2 cache access latency: 11 cycles
L3 cache access latency: 39 cycles
Main memory access latency: 107 cycles

They're only providing BIOS options for Xeon and Core X-Series processors, not sure if there were another avenue or otherwise.

seeteeyou · October 20, 2018

Are you guys familiar with this?

https://forums.naimaudio.com/topic/rips-are-just-rips?reply=77314878515682648#77314878515682648

Quote

And of course as described on the forum many times now I have measured and mused with Naim how different media servers can sound different transferring the same sampled data... and this has been down to the Ethernet interframe timing spacing consistency between media server and streamer... this is mitigated on the newer streamer architecture.. I discovered this difference very early on with Naim streaming.. and took me many months to discover and measure the reason why.. it was quite a Eureka moment!

zettelsm · October 20, 2018

On 10/19/2018 at 9:55 AM, austinpop said:

Ah I didn't answer this. The SE direct to the tX-USBultra is better in 2 key ways:

dynamics: music sounds more energetic and alive

darker, calmer, more relaxed sounding. Some call this a wetter or creamier sound. It just makes you sigh and listen deeper and longer.

Thanks for this.

Did you also try comparing SE direct to DAC, either this trial or in the past?

The reason I ask -- I recently tried a tX-USBultra powered variously by an Uptone JS-2, Uptone LPS-1, and Sonore Signature power supplies between my Innuos ZENith and Chord Blu MKII feeding a Chord DAVE DAC and didn't hear any particular benefit with the tX-USBultra, regardless of power supply.

I'm confident I'm familiar with the sound of my system, and the system itself is fairly revealing, so my experience would seem to fly in the face of common experience, at least here on CA.

Thanks for your comments,

Steve Z

Bruce Orr · October 20, 2018

On 10/18/2018 at 11:45 PM, austinpop said:

Bruce,

I didn't mean to confuse you. You're right, I no longer use an sMS-200ultra. I use the SE straight to tX-USBultra. The only reason for the above comparison was to directly compare the vanilla to Neo - because it was possible - as a service to the community. Most people who upgrade don't have the old unit still in hand to do this kind of comparison.

The NUC approach that Roy is taking goes a step beyond the sMS-200ultra, by using an in-memory OS using the ramboot feature of audiolinux. As Roy said, it's not clear why endpoints like the sMS or the uR do not give the same level of performance as his/Larry's ramboot NUCs, although we suspect it has to do with the low latencies of operations when running the OS in memory.

Bruce Orr · October 20, 2018

Rajiv, it’s not you. I’m just easily confused! Thanks for the additional clarification.

A novel way to massively improve the SQ of computer audio streaming

Recommended Posts

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Link to comment

Create an account or sign in to comment

Create an account

Sign in