Building a DIY Music Server

[StreamFidelity]

7 hours ago, genvirt said:

Only one thing that I can say to be considering maybe - the width - in my case 48cm will not enter to my rack, and most of audio equipment has 43-45 in its width...

 

I am excited about this beautiful new project. I share the chassis size concern.

 

I am also not convinced of putting the power supply in the same chassis. The DIY community lives from trying out different linear power supplies. It is also a question of costs. A transformer that does not hum must be of particularly good quality and encapsulated. Who would want vibrations in the chassis?

 

If there is no difference to the Taiko Extreme, who else will buy it? 😉

[StreamFidelity]

1 hour ago, guiltyboxswapper said:

Have any of the Solarflare SFN8522

 

The Solarflare SFN8522 is absolutely awesome! A dream in terms of sound. I reported here:

 

 

 

[StreamFidelity]

53 minutes ago, ASRMichael said:

Would be great if there was a comparison with new JCAR Net XE card!

 

The JCAT NET Card FEMTO is well behind Solarflare. I don't have the XE version.

 

27 minutes ago, guiltyboxswapper said:

Its worth making an effort with OnLoad - you're only half utilising the card at the moment 

 

Yes, I am aware of that. Solarflare is a Diva. 😄

 

I was happy to get it to run well. According to the manual, I have set the network card adapter to the lowest possible latency. Have a look at this: DSD 256 x 48. The sound is fantastic.

 

 

Can OnLoad also be set in Windows? As far as I know, this only works on Linux.

[StreamFidelity]

1 hour ago, guiltyboxswapper said:

Did you externally power your NET Femto card too?

Yes, with the JCAT Optimo 3 Duo. An absolutely excellent LPS. And yet, powered by the motherboard, the Solarflare 8522 card is so much better.

 

42 minutes ago, Zauurx said:

LantecyMon gives the overall latency, not that of the network and your card ?

That's right. But I think the Solarflare 8522 card have their part in that. The latency is reportedly in the very low range of less than 1μsec.

[StreamFidelity]

1 hour ago, guiltyboxswapper said:

I would also try the "max throughput" option as that invokes a certain amount of in-memory caching/batching.

 

I think that runs contrary to low latencies. The manual says:

 

There are three metrics that should be considered when tuning an adapter:
• Throughput
• Latency
• CPU utilization
Transactional (request-response) network applications can be very sensitive to latency whereas bulk data transfer applications are more dependent on throughput.
...
Turning off interrupt moderation wants:
- generate the most interrupts
- give the highest CPU utilization
- give the lowest latency
- give the biggest reduction in peak throughput.

[StreamFidelity]

5 minutes ago, MarcelNL said:

I also toyed with the idea of installing an SFP based network card, but I'm dreading the fight with Linux...

 

Linux scared me off too. I don't know about it. if I then read through the numerous Linux installation problems with the HQPlayer, I don't feel like doing it. Solarflare provides the appropriate Windows drivers. Except for OnLoad, everything can be used.

[StreamFidelity]

29 minutes ago, guiltyboxswapper said:

I think myself and @StreamFidelity use the non-plus version.

 

I don't have the plus version. And I'm just noticing that this card doesn't support OnLoad at all. One less reason to switch to Linux. 😄

 

Here is an overview: 8000 Series 10/40GbE Network Adapters

 

 

[StreamFidelity]

 

@Nenon I very much apologize for disturbing your thread with my Solaflare Card reports. Questions about this are perhaps better off in my thread.

 

I suggest we all focus now on the great DIY Taiko Audio products.

 

1 hour ago, Nenon said:

Dimensions are approximately 170 mm x 78 mm. 

Here are the HDPLEX dimensions for comparison.

 

 

[StreamFidelity]

11 hours ago, seeteeyou said:

Dual Xeon 4114 will only boot with RDIMM like D51.23244S.001 as shown below, please be cautious and DO NOT get the regular UDIMM

 

I think the latencies are quite high.

 

Apacer [email protected] MHz RAM

1,000 / 2,666 * 19 = 14.25 nanoseconds

 

It can be done better, for example [email protected] MHz RAM

1,000 / 2,400 * 10 = 8.33 nanoseconds (0.00000833 ms)

The 2.400 MHz is not very high, but in combination with the CL value, the latencies are very low.

[StreamFidelity]

25 minutes ago, MarcelNL said:

why does Optane sound better than a regular SSD?

 

It could also be due to the latencies. See image.

Source: PC Perspective INTEL’S OPTANE DC PERSISTENT MEMORY DIMMS PUSH LATENCY CLOSER TO DRAM

 

I consider the lowest possible latency to be one of the keys to audio PCs.

 

Latencies can also be set in the software. With ASIO drivers (USB), Taiko Audio experimented with the buffer settings to my knowledge.

 

If you are interested, I wrote something about latencies (click on translate): Audio PC latencies

[StreamFidelity]

2 hours ago, Zauurx said:

But the latency remains high and once the drivers are up to date...

 

Is it the result of LatencyMon? Which most disruptive process does the application indicate?

[StreamFidelity]

15 hours ago, Zauurx said:

The report is attached

 

The Ntoskrnl.exe takes on numerous tasks that are difficult to understand. The process has, among other things, an impact on hardware virtualization, as well as on memory and process management.

 

I suspect the culprit is more likely to be storport.sys.

 

MS wrote: storport.sys is your chipset device driver, check the support page for your device on the manufacturers website to see if there is an updated chipset driver

 

It helps to deactivate unneeded devices in the BIOS. I have deactivated WLAN, Bluetooth and Audio. And of course a BIOS update always has to be made. That solves a lot of problems.

[StreamFidelity]

13 hours ago, Gavin1977 said:

SFN8255 now added - immediate increase in resolution / definition.

I assume you mean Solarflare SFN8522?

 

13 hours ago, Gavin1977 said:

Big thanks to everyone and @StreamFidelity for reporting on this one.

You're welcome and I look forward to your positive experience.

 

13 hours ago, Gavin1977 said:

I gave in and added a fan to the Solarflare  (noctua 40x20mm)... too hot for my liking in a passive system.

This is an interesting fan solution. With the new cover, however, it also works without a fan.

 

[StreamFidelity]

1 hour ago, Gavin1977 said:

However, the Formula has "a dedicated onboard clock generator that supplies dynamic reference clocks to the CPU cores and major subdomains".  Any idea if this provides any benefit?

 

I think it depends on the purpose. With the HQplayer and the EC modulators I need at least 4GHz. Overclocking runs great on the ASUS ROG Maximus XI Formula board. I've never had any problems with it.

 

I chose the motherboard because of its very good passive cooling properties. The backplate not only improves stability, but is also responsible for cooling 20 transistors. The Crosschill EK III cooler with copper fins cools the voltage converters and VRM and the ten coils very well even without water.

[StreamFidelity]

2 hours ago, ASRMichael said:

Hi, are you using a switch between your home network & solar?

 

Yes, this is my current setup:

 

 

At some point I will buy a switch with a SFP connection. One box less. 😄

[StreamFidelity]

Hi Matt,

 

It is not a commercial modification. Beci does it for members in the German forum. A field report is here: Treffen der Generationen. FritzBox 7390-7530-7590 (Beci). He can be reached via the forum: https://www.open-end-music.com/member/2818-beci.

 

Gabriel

 

[StreamFidelity]

13 hours ago, matthias said:

I am curious how even the modded Fritzbox cable modems can sound good as ALL use the "bad" Intel Puma chipset.

 

Hi Matt,

 

the modification was more about the power supply. As far as I know, Beci installs higher quality voltage regulators and capacitors. He also offers linear power supplies. I use a Keces P3. Otherwise I don't have much confidence in the chipset and clocking capabilities of the FritzBox.

 

I perform a reclocking with Selllarz clocks on the Ethernet link. These are the devices labeled G-Switch and 11Gtek.

 

Gabriel

[StreamFidelity]

On 3/19/2021 at 1:59 PM, Gavin1977 said:

Motherboard which has true VRM phases (avoid the use of doublers which can affect transient performance of power supply to the CPU).

 

Thanks for this very good explanation.

 

In my opinion, ASUS is taking a different approach. Using the example of the MB ASUS ROG Maximus XIII Extreme, a teamed power stage design is used. In contrast to the conventional phase-doubled design, the power supply takes place in parallel and not one after the other.

 

Source: https://rog.asus.com/de/motherboards/rog-maximus/rog-maximus-xiii-extreme-model/

 

ASUS writes: These improvements necessitated a re-evaluation of power design priorities, because phase doublers add a propagation delay that hampers transient response.

[StreamFidelity]

32 minutes ago, Topk said:

@Streamfidelity did recommend enabling HT for upsampling and should be able to provide more details. 

 

It depends on the purpose. As some have already written, it is highly recommended for upsampling with the HQPlayer. Hyperthreading does not increase the computing power as some think. Instead, the parallel processing is distributed over more (virtual) cores.

 

The manufacturer also recommends hyperthreading for the Solarflare SFN8522 NIC to reduce latencies.

[StreamFidelity]

12 hours ago, Topk said:

Anyway, to cut a long story short, we both use the EF PCIe power filter.

 

I don't understand how it works.

 

Audiophonics writes: The AXF-107 module from Elfidelity is a switch mode power supply filter for computers connected in PCI Express or PCI Express 16x and designed to provide cleaner power to all connected cards (sound card, interfaces, GPU ...).

 

My questions:

 

1. Is the PCIe card simply plugged into a free slot and "sucks" EMI interference?

2. How is the power consumption?

3. Are there any recommendations as to which slot to use?

[StreamFidelity]

39 minutes ago, Topk said:

1. it’s supposed to suck some noise from the power supply.

 

Many Thanks. I'll just try it out now. 👍

[StreamFidelity]

12 hours ago, ASRMichael said:

Tried with a PC PSU. Interestingly boots no problem, that indicates maybe my LPS.

 

During the boot process, high current peaks can overload weak power supplies. The boot process is aborted or repeated. This can destroy the motherboard. Which CPU are you using?

[StreamFidelity]

1 hour ago, ASRMichael said:

i9 10900k

 

Is the CPU multiplier set to Auto? If so, try a fixed assignment for all cores, for example 42 (4.2GHz).

[StreamFidelity]

11 hours ago, Topk said:

If you look at the photo attached below, I circled in red that in some MB (not all), you have both a 8 pin and 4 pin CPU connectors for 12V.
 

My experience is that you need to feed both connectors with power. Gives a significant better sound and jump in sound quality: bigger, fuller, more dynamic, more effortless. 
 

Do not leave the CPU 4 pin connector unpowered or you leave big sound quality gains on the table! This CPU needs to be well fed, it’s key to SQ.

 

It would have been nice if it worked. I had done this experiment before and have now repeated it with Taiko Audio DC-ATX and 2 Keces P8. I use ROG MAXIMUS XI FORMULA.

 

The nice thing about the Keces P8 is the display of the voltage and power consumption. This means that there can be no mistake about where the electricity comes from.

 

1. Use case:
- Keces P8 20V to Taiko Audio DC-ATX
- Keces P8 12V to 8 pin CPU EPS
- Taiko Audio DC-ATX 12V to 4 pin CPU EPS

 

No power was drawn from the Keces P8 12V! The audio PC booted successfully anyway.

 

2. Use case:
- Keces P8 20V to Taiko Audio DC-ATX
- Keces P8 12V to 4 pin CPU EPS
- Taiko Audio DC-ATX 12V to 8 pin CPU EPS

 

No power was drawn from the Keces P8 12V! The audio PC booted successfully anyway.

 

Beware of placebo effects and measure specifically where the electricity is coming from. 😉

 

I think it only works if the 8 pin and 4 pin connections are fed from the same LPS. Then the larger cable diameter may help something.

[StreamFidelity]

2 hours ago, Topk said:

But for some reason, it did work with the 2 Keces. If someone knows why, I’m interested to know.

 

We may not be talking about the same things. I also use two Keces Single P8. But I only use the 8 pin CPU EPS.

In my opinion, the question revolved around using the 4 pin CPU EPS as well. So ultimately to use the CPU EPS over 12 pins in total. That is only possible with an LPS.

 

14 minutes ago, NanoSword said:

Can you use 4pin from taiko EPS 2 to power the 4pin CPU or it will not work like this ?

 

Sure it's possible. I use an 8 pin Molex cable on both sides for this. 4 pins remain free on the side of the motherboard. But I only did this for my experiment with 2 Keces P8. Otherwise, of course, I took the 8 pin on the motherboard. What is the point of only using the 4 pins?

 

 

However, this constellation cannot keep up with a direct CPU supply bypassing the Taiko ATX as shown above. Nenon has said that this requires an unregulated LPS to take advantage of the Taiko ATX.