Building a DIY Music Server

[Nenon]

1 hour ago, MarcelNL said:

I recently discovered that playing files from RAM sounds better when they sat on Optane rather than the (externally fed) SSD...no, don't ask me to explain

 

We've posted about that more than two years ago :). Good to see you catching up :). 

I suppose you fixed your Optane issue then? 

[MarcelNL]

did you? good to see I can confirm it ;-)

I read the info as saying that Optane as SSD sounded better, and that made me try Optane,  in a small dose as the cost of the large PCIe disks is abhorring. I tried loading to the RAM Disk both from SSD and Optane and learned that the latter 'storage' options sounds better, which makes me 'park' files on Optane to then load them into RAM.

 

That was testing done with the first Optane NVME stick that mysteriously died a while ago, he new one (32Gb) is not recignized by the new MB.

 

I may have misunderstood the info in the thread, anyway, the Optane is not yet working I just got my gear put back together and playing after a very long nightly session comparing loads of stuff with a couple of friends.

[m.i.c.k.e.y]

Just posting as an alternative and that it works.

 

My stop gap solution while wating for the Audiobyte Hydra.Hub

 

Android Box (Beelikn GT King) running UAPP + LPS -> USB Reclock/Regen (LH Labs Revive) -> Singxer SU-6

 

[CJH]

I don't think Z590 MB supports optane--you would need Z490.

CJH

[MarcelNL]

1 minute ago, CJH said:

I don't think Z590 MB supports optane--you would need Z490.

CJH

I'm using a X570...for which NVME PCIe 4 is a function, https://www.gigabyte.com/nl/Motherboard/X570-AORUS-ELITE-rev-10#kf

Somehow my 32Gb MEMPEK NVME Optane is not recognized, BUT linux tells me the NVME is 'not ready', which is a bit cryptical.

[MarcelNL]

so...today I swapped the switching external PSU for the SSD for a linear ultra low noise and I cannot believe what a difference in musicality, dynamics, space, power slam prat etc .that makes. Is digital really about 0 and 1 or is there a -0.9 and 1.1 too?

Guess i'll be trying mercury filled tube rectifiers for this anytime soon.

 

[genvirt]

On 3/23/2021 at 1:30 AM, Nenon said:

The Taiko ATX was delayed and scheduled for the end of April now

Great news @Nenon, if that so, can You please share final dimensions (H:D:W) and would be the holes the same as on HDPlex?

 

[Topk]

Impact of CPU on audio quality (going from 3700x to 3900x to 3950x in a fanless case)

 

I wanted to report my findings going from 3700x to 3950x in a H5 case.

 

Since I have HDplex H5 case I started with 3700x since it’s 65W TDP CPU (Asus rog strix b550-E MB and I can explain why vs. Gygabite Aorus master). It sounded good but for me was lacking some body and weight. I then upgraded to 3900x. No other change. Immediately the sound became much weightier, bigger, fuller, more inner detail, more calm and analogue, more realistic tone, more dense, much more depth, better soundstage.. etc. There were only upsides. If I had to put %, I would say sound quality increase of 50%. 

The reason I did this is because of reports here on the impact on CPU on SQ (eg Romaz) and because a friend who upgraded from i7 to 3900x reported a shocking increase in SQ in his system. I was able to make this 105W TDP work in H5 with undervolting done the right way (you can also use manual PBO, I might explain if people are interested)

 

Since I was able to make 3900x (12 cores) work in H5, I then decided to go for 3950x (16 cores but the same 105w TDP) and use same BIOS techniques, assuming more cores, more processing power = better/bigger sound.

 

Well guess what, the sound quality increased again. Same exact type of change in sound going to 3900x to 3950x. Bigger, denser, more musical content inside the sound, fuller, better tone, better depth, better decay... Very clear and obvious. Sounds a bit like you have a big amplifier upgrade, the sound is so much more powerful and real. In %, I would say plus 25% this time.

 

So my recommendation to anyone doing a DYI PC is start with the biggest CPU. I like AMD because they use 7nm technology and therefore generate less heat (even though they might have slightly higher latency than Intel).
 

I don’t know a way to create this sound without a bigger CPU, and at the back end, no USB card or LPS can re-create what’s not there in the first place...
 

Reported temp on the MB LCD is about 45-48. I do use a Keces P8 12v single rail on the CPU so YMMV. The 3950x is only drawing 2.1 amps on the P8 while playing music (no upsampling) which is less than 3700x and 3900x (boost etc all activated).

 

For context I have:

Asus MB B-550E MB (fanless chipset) 

AMD 3950x (16 cores and 70mb of cache)

Apacer 2666Mhz ECC RAM 8gb (latest model)

Euphony O/S

Jcat XE USB and Femto USB, El Fidelity USB, Pink Faun SPDIF bridge 

Jcat Net Femto

Uptone JS2 LPS

2 custom LPS with supercaps etc

KECES P8 dedicated to the CPU with FIS audio CPU cable

Uptone USB isoregen

Jcat Ethernet isolator

Etc

McIntosh and Harbeth 

I will post photos later on.

[bit01]

Apacer has 3200MHz DDR4 industrial wide temp modules out now. I have just received a 16GB one! In due time I will compare it to the G. Skill Ripjaws V (3200MHz). The Ripjaws produced a 'bigger' sonic  in my setup than the Apacer standard temperature 2666MHz modules, but not as 'sweetly polished'.

[Topk]

9 minutes ago, bit01 said:

Apacer has 3200MHz DDR4 industrial wide temp modules out now. I have just received a 16GB one! In due time I will compare it to the G. Skill Ripjaws V (3200MHz). The Ripjaws produced a 'bigger' sonic  in my setup than the Apacer standard temperature 2666MHz modules, but not as 'sweetly polished'.

Wow interesting!! Looking forward to your review! Where did you buy?

 

By the way we can talk about RAM: I alternate between

- Apacer 2666Mhz C19 ECC wide temp

- Corsair 3200 C15 and

- Patriot 4000 C16

 

It’s really a matter of taste. Also 1 stick is different (probably better for most) than 2 sticks (even if 2 or even 4 sticks are dual/quad channels and higher bandwidth).

I can also share my experience with latency, CAS/RAM timing, ram speed, ram voltage...etc but nothing really new vs. what others have reported before me.

Overall: Jury is out for now... I wouldn’t say Apacer is unambiguously better for everyone. 

[bit01]

17 minutes ago, Topk said:

Wow interesting!! Looking forward to your review! Where did you buy?

 

...

Hi,

I ordered it via one of the e-mail links that was posted in these forums earlier:

Vincent Chen (Apacer Memory America)

[email protected]

 

1x16GB:  p/n D31.27261S.002

 

ATB

b.

[Mr Morris]

Hi looking for some help with my DIY server. Hope this question is not too off topic. I have an old I3 based system, but with a new PSU using Sean Jacobs parts. Very pleased with it and his excellent support.

 

Funds don't permit MB changes currently. So tweaking internal wiring. My Jcat USB card is connected by a PC Riser card and a USB 3 Male to male cable.  I want to try making my own. Is the first configuration listed in the link below  the correct one? My DMM probes won't fit in the current cables connectors to work it out.

 https://www.newnex.c...ring-option.php
Thanks

[ASRMichael]

4 hours ago, Topk said:

Impact of CPU on audio quality (going from 3700x to 3900x to 3950x in a fanless case)

 

I wanted to report my findings going from 3700x to 3950x in a H5 case.

 

Since I have HDplex H5 case I started with 3700x since it’s 65W TDP CPU (Asus rog strix b550-E MB and I can explain why vs. Gygabite Aorus master). It sounded good but for me was lacking some body and weight. I then upgraded to 3900x. No other change. Immediately the sound became much weightier, bigger, fuller, more inner detail, more calm and analogue, more realistic tone, more dense, much more depth, better soundstage.. etc. There were only upsides. If I had to put %, I would say sound quality increase of 50%. 

The reason I did this is because of reports here on the impact on CPU on SQ (eg Romaz) and because a friend who upgraded from i7 to 3900x reported a shocking increase in SQ in his system. I was able to make this 105W TDP work in H5 with undervolting done the right way (you can also use manual PBO, I might explain if people are interested)

 

Since I was able to make 3900x (12 cores) work in H5, I then decided to go for 3950x (16 cores but the same 105w TDP) and use same BIOS techniques, assuming more cores, more processing power = better/bigger sound.

 

Well guess what, the sound quality increased again. Same exact type of change in sound going to 3900x to 3950x. Bigger, denser, more musical content inside the sound, fuller, better tone, better depth, better decay... Very clear and obvious. Sounds a bit like you have a big amplifier upgrade, the sound is so much more powerful and real. In %, I would say plus 25% this time.

 

So my recommendation to anyone doing a DYI PC is start with the biggest CPU. I like AMD because they use 7nm technology and therefore generate less heat (even though they might have slightly higher latency than Intel).
 

I don’t know a way to create this sound without a bigger CPU, and at the back end, no USB card or LPS can re-create what’s not there in the first place...
 

Reported temp on the MB LCD is about 45-48. I do use a Keces P8 12v single rail on the CPU so YMMV. The 3950x is only drawing 2.1 amps on the P8 while playing music (no upsampling) which is less than 3700x and 3900x (boost etc all activated).

 

For context I have:

Asus MB B-550E MB (fanless chipset) 

AMD 3950x (16 cores and 70mb of cache)

Apacer 2666Mhz ECC RAM 8gb (latest model)

Euphony O/S

Jcat XE USB and Femto USB, El Fidelity USB, Pink Faun SPDIF bridge 

Jcat Net Femto

Uptone JS2 LPS

2 custom LPS with supercaps etc

KECES P8 dedicated to the CPU with FIS audio CPU cable

Uptone USB isoregen

Jcat Ethernet isolator

Etc

McIntosh and Harbeth 

I will post photos later on.

Thank you for the post. Can I ask what clock speed have you set? 
 

I’m now using Intel i9 10900k. I had to restrict to 8 cores, no matter if frequency was set at 3.5ghz or 5.3ghz it would not work on 10 cores. Doesn’t get past boot. 
 

Currently CPU is fully open on turbo 5.3ghz. Boots fine with 8 cores. 
 

I’m using 10a LPS to power it. 
 

My temps are 45-48c like yours. @ 5.3ghz. 
 

41c @ 3.7ghz

 

much higher than my i9 9900k. @ 5.0ghz idle was around 37c. 
 

What doesn’t help is the XII Extreme MB has a heatsink close to cpu, its bit higher than previous MB. As a result the H5 copper pipes just fit but are touching the MB heatsink. 
 

No listening tests as yet. Just powered it up, wait few days and will revert back. 

[Dev]

28 minutes ago, ASRMichael said:

Thank you for the post. Can I ask what clock speed have you set? 
 

I’m now using Intel i9 10900k. I had to restrict to 8 cores, no matter if frequency was set at 3.5ghz or 5.3ghz it would not work on 10 cores. Doesn’t get past boot. 
 

Currently CPU is fully open on turbo 5.3ghz. Boots fine with 8 cores. 
 

I’m using 10a LPS to power it. 
 

My temps are 45-48c like yours. @ 5.3ghz. 
 

41c @ 3.7ghz

 

much higher than my i9 9900k. @ 5.0ghz idle was around 37c. 
 

What doesn’t help is the XII Extreme MB has a heatsink close to cpu, its bit higher than previous MB. As a result the H5 copper pipes just fit but are touching the MB heatsink. 
 

No listening tests as yet. Just powered it up, wait few days and will revert back. 

what software/OS are you running ? I too use i9-10900k with a dedicated SJ 10A rail and it works fine with AudioLinux 2.0 and Win 10Pro but NOT Euphony. With Euphony I had all sorts of problems. If you are using Euphony, try disabling Hyperthreading in order to use 10 cores. 

[Topk]

54 minutes ago, ASRMichael said:

Thank you for the post. Can I ask what clock speed have you set? 
 

I’m now using Intel i9 10900k. I had to restrict to 8 cores, no matter if frequency was set at 3.5ghz or 5.3ghz it would not work on 10 cores. Doesn’t get past boot. 
 

Currently CPU is fully open on turbo 5.3ghz. Boots fine with 8 cores. 
 

I’m using 10a LPS to power it. 
 

My temps are 45-48c like yours. @ 5.3ghz. 
 

41c @ 3.7ghz

 

much higher than my i9 9900k. @ 5.0ghz idle was around 37c. 
 

What doesn’t help is the XII Extreme MB has a heatsink close to cpu, its bit higher than previous MB. As a result the H5 copper pipes just fit but are touching the MB heatsink. 
 

No listening tests as yet. Just powered it up, wait few days and will revert back. 

First of all, 10900K is 125W TDP, that’s 30% above what HDplex recommends for H5. You have low temps but the 10900k does need power. It can pull up to 225W if I recall correctly.

 

Right now with 3950x, my CPU is un-throttled for speed within the limits of the core voltage which is set manually at a pretty high 1.25V (I’ll explain what I mean by that).

 

All the rest including Precision Boost Overdrive, Asus performance enhancement, CPU clock speed are all on Auto. Which means the CPU is free to boost frequency as needed (well

above 4 GHz) as long as core voltage is no more than 1.25V. And based on Linus Tech Tips tests and others, the 3950x runs around a very low 1V voltage under most conditions / benchmarking tests. Also, keep in mind music is not a CPU intensive task (without upsampling). It’s not like you’re running Cinebench R20 or Prime95.
 

To cut a long story short, my 3950x is pretty much running full speed in a fanless case....

 

On the other hand, the very powerful 10900K is rated 125W TDP which means it draws a lot of power. Intel was very smart in optimizing with thinner die but it’s still 14nm.
 

If I were you, I would first:

1) run all cores - it’s critical for sound quality

2) pick a decent clock speed - start with the stock speed and manually lock it (manual override in the BIOS). If needed you can decrease later on 

3) disable turbo boost (for now) - that can potentially put strain on your LPS 

4) Decrease CPU core voltage. Start with stock voltage and decrease in 0.25v increments, and check for stability 

 

Another way to do it is manually set TDP (if you can on Intel) to 95W or less and let the BIOS adjust everything else. 

Now - you mention you can’t go past boot. So a few more tips:
 

It could be because your CPU is boosting on the 10 cores and exceeds 10A of power usage during boot. In that case, the locked lower voltage might help. The lower fixed TDP might help. Or if you manually lock the power going to the CPU (PPT for AMD), max peak amperage (EDC for AMD) or max amperage (TDC), that should help too. Under AMD terminology:

 

- Package Power Tracking (“PPT”): The PPT threshold is the allowed socket power consumption permitted across the voltage rails supplying the socket. Applications with high thread counts, and/or “heavy” threads, can encounter PPT limits that can be alleviated with a raised PPT limit.

- Thermal Design Current (“TDC”): The maximum current (amps) that can be delivered by a specific motherboard’s voltage regulator configuration in thermally-constrained scenarios.

 

- Electrical Design Current (“EDC”): The maximum current (amps) that can be delivered by a specific motherboard’s voltage regulator configuration in a peak (“spike”) condition for a short period of time.

i don’t know the intel equivalents of the above but that could fix the issue

 

And I agree hyper threading should be disabled (for sound quality) if you’re not upsampling  but I don’t think it’s the root cause

 

I couldn’t boot in euphony (maybe like you if you use it too) and I could see on the Keces P8 the CPU pulling much more than 8A (which is the max for P8), probably boosting the CPU during boot. When I manually locked the voltage to 1.25V, the issue went away.

 

while you’re at it, disable spread spectrum as well since it measurably creates jitter. 

[ASRMichael]

1 hour ago, Dev said:

what software/OS are you running ? I too use i9-10900k with a dedicated SJ 10A rail and it works fine with AudioLinux 2.0 and Win 10Pro but NOT Euphony. With Euphony I had all sorts of problems. If you are using Euphony, try disabling Hyperthreading in order to use 10 cores. 

Hi

 

I’m using Euphony worked no problems. Using onboard LAN didn’t work but I use JCAT net anyway. 
 

I’m just glad back up & running. So I’ll try 10 cores again in few weeks. 
 

Actually playing music for a while (no upsampling) my temps are 65c at 5.3ghz. What are yours? 

[bit01]

@ASRMichael - the peak current capacity of your 10A LPS that is supplying the i9-10900K should be in consideration - I cannot give the exact requirement but the higher the better? 

[ASRMichael]

2 hours ago, Topk said:

First of all, 10900K is 125W TDP, that’s 30% above what HDplex recommends for H5. You have low temps but the 10900k does need power. It can pull up to 225W if I recall correctly.

 

Right now with 3950x, my CPU is un-throttled for speed within the limits of the core voltage which is set manually at a pretty high 1.25V (I’ll explain what I mean by that).

 

All the rest including Precision Boost Overdrive, Asus performance enhancement, CPU clock speed are all on Auto. Which means the CPU is free to boost frequency as needed (well

above 4 GHz) as long as core voltage is no more than 1.25V. And based on Linus Tech Tips tests and others, the 3950x runs around a very low 1V voltage under most conditions / benchmarking tests. Also, keep in mind music is not a CPU intensive task (without upsampling). It’s not like you’re running Cinebench R20 or Prime95.
 

To cut a long story short, my 3950x is pretty much running full speed in a fanless case....

 

On the other hand, the very powerful 10900K is rated 125W TDP which means it draws a lot of power. Intel was very smart in optimizing with thinner die but it’s still 14nm.
 

If I were you, I would first:

1) run all cores - it’s critical for sound quality

2) pick a decent clock speed - start with the stock speed and manually lock it (manual override in the BIOS). If needed you can decrease later on 

3) disable turbo boost (for now) - that can potentially put strain on your LPS 

4) Decrease CPU core voltage. Start with stock voltage and decrease in 0.25v increments, and check for stability 

 

Another way to do it is manually set TDP (if you can on Intel) to 95W or less and let the BIOS adjust everything else. 

Now - you mention you can’t go past boot. So a few more tips:
 

It could be because your CPU is boosting on the 10 cores and exceeds 10A of power usage during boot. In that case, the locked lower voltage might help. The lower fixed TDP might help. Or if you manually lock the power going to the CPU (PPT for AMD), max peak amperage (EDC for AMD) or max amperage (TDC), that should help too. Under AMD terminology:

 

- Package Power Tracking (“PPT”): The PPT threshold is the allowed socket power consumption permitted across the voltage rails supplying the socket. Applications with high thread counts, and/or “heavy” threads, can encounter PPT limits that can be alleviated with a raised PPT limit.

- Thermal Design Current (“TDC”): The maximum current (amps) that can be delivered by a specific motherboard’s voltage regulator configuration in thermally-constrained scenarios.

 

- Electrical Design Current (“EDC”): The maximum current (amps) that can be delivered by a specific motherboard’s voltage regulator configuration in a peak (“spike”) condition for a short period of time.

i don’t know the intel equivalents of the above but that could fix the issue

 

And I agree hyper threading should be disabled (for sound quality) if you’re not upsampling  but I don’t think it’s the root cause

 

I couldn’t boot in euphony (maybe like you if you use it too) and I could see on the Keces P8 the CPU pulling much more than 8A (which is the max for P8), probably boosting the CPU during boot. When I manually locked the voltage to 1.25V, the issue went away.

 

while you’re at it, disable spread spectrum as well since it measurably creates jitter. 

Thanks I’ll have a look! Expect a few private messages! 😂😂! Cheers appreciated

[Dev]

3 hours ago, ASRMichael said:

Hi

 

I’m using Euphony worked no problems. Using onboard LAN didn’t work but I use JCAT net anyway. 
 

I’m just glad back up & running. So I’ll try 10 cores again in few weeks. 
 

Actually playing music for a while (no upsampling) my temps are 65c at 5.3ghz. What are yours? 

 

46-48C

[Topk]

3 hours ago, Dev said:

 

46-48C

43C as we speak on the MB, and drawing 2.0 Amps on the 12V Keces P8, playing music with Euphony 

 

the PCIe card on the photo with red Wima caps is an El Fidelity PCIE power filter, a must have for an audio PC (but that’s for another post...)

 

[Nenon]

19 hours ago, genvirt said:

Great news @Nenon, if that so, can You please share final dimensions (H:D:W) and would be the holes the same as on HDPlex?

 

On 1/13/2021 at 10:22 AM, Nenon said:

Dimensions are approximately 170 mm x 78 mm. 

Depth is around 20 mm but you need clearance for the Molex connectors. Holes match the HDplex 400W DC to DC ATX. 

 

You can find more specs in my post:

On 1/13/2021 at 10:22 AM, Nenon said:

Taiko Audio DC-DC ATX

 

Input Voltage:
- 16VDC min
- 48VDC max

- Input connector: Molex 469920610 (same as on the HdPlex)

 

Output rails:
- 12VDC - 40A
- 5VDC - 15A
- 3.3VDC - 15A
- 5VDC SB - 4A (this is a also available on the SATA connector as an always on supplyrail for a USB card or something similar)

 

Dimensions are approximately 170 mm x 78 mm. 

 

Other notes:
- A little larger than the Hdplex 400W DC to DC ATX but it has the same mounting holes, which makes it a drop in replacement.
- Amps are continuous ratings. 

- Optimal input seems to be in the 19VDC - 35VDC range, but that may change with more experiments.

 

Here are some photos but please keep in mind this is a modified prototype on which several components have been exchanged, hence the residual manual solder marks, wires, scratched copper plate, etc. The production units would be much cleaner / better looking.

 

Connectors from left to right: 6-pin DC input, 8-pin EPS connector output, 24-pin ATX connector output, 8-pin EPS connector output, 4-pin SATA connector output (with 5VDC SB rail as mentioned above and a shared 12V rail).

 

 

Price: 1250 Euro + shipping

First batch is expected in March 2021 end of April.

 

Pre-order form:

https://forms.office.com/Pages/ResponsePage.aspx?id=wAyHaKEU5EWxi4f_4gJE6LY6WJsC4qJJgufOWga3mOJUNlZSMDhBOEs5NVBVUDFKRENINlQzOVlZOCQlQCN0PWcu

 

 

 

[guiltyboxswapper]

23 hours ago, Topk said:

Since I was able to make 3900x (12 cores) work in H5, I then decided to go for 3950x (16 cores but the same 105w TDP) and use same BIOS techniques, assuming more cores, more processing power = better/bigger sound.

 

I tried this a year ago with HQPlayer on a 3900x - setting CPU to 1.0v and it destroyed the SQ in this case, but yes massive power and heat savings.  This was with fan cooling in both cases, NH-D15.  I wonder if the big SQ improvements in your case come from running the CPU much closer to its thermal limits, where as in stock it would easily be closer to its thermal cutoff points?

 

Interestingly I also swapped in my Ryzen 5800x in place of my 3900x.  5000 series has big AVX2 improvements which meant the 5800x could do the same tasks with less power/compute which meant HQPlayer was running with the lowest cpu usage i'd seen in years.  SQ wise?  To my surprise I could barely tell them apart, maybe the 5800x has a slight advantage but beyond that I'd struggle.

 

The other thing worth mentioning is the Solarflare X2522 is worth getting hold of.  It's a clear refinement of the 8522 cards, especially when it comes to transient attacks and obvious lower noise floor.  Just be warned its not as widely compatible with SFP+ being SFP28, but some work no problem.

[Topk]

29 minutes ago, guiltyboxswapper said:

I tried this a year ago with HQPlayer on a 3900x - setting CPU to 1.0v and it destroyed the SQ in this case, but yes massive power and heat savings.

 

Interestingly I also swapped in my Ryzen 5800x in place of my 3900x.  5000 series has big AVX2 improvements which meant the 5800x could do the same tasks with less power/compute which meant HQPlayer was running with the lowest cpu usage i'd seen in years.  SQ wise?  To my surprise I could barely tell them apart, maybe the 5800x has a slight advantage but beyond that I'd struggle.

 

The other thing worth mentioning is the Solarflare X2522 is worth getting hold of.  It's a clear refinement of the 8522 cards, especially when it comes to transient attacks and obvious lower noise floor.  Just be warned its not as widely compatible with SFP+ being SFP28, but some work no problem.

1V is very low. When you went to 1V, did you lock the CPU core ratio back then? This is key otherwise the bios would make dynamic adjustments impacting performance and sound quality indeed. You can even manually lock and run the core above stock speed since there’s very little heat generated. 

 

On the 5800x, I didn’t compare the 3 series with the 5 series but the 5800x has 8 cores vs 12 and 36mo cache vs 70mo. On the other hand the 5800x has the latest architecture. Your finding seems to indicate the 5900x and 5950x might be even better and that the 5 series is better sounding than the 3 series (which absolutely makes sense because the 5 series has better performance than the 3 series). Could also be system dependent and O/S dependent.

 

On the solarflare, did you compare with Jcat Net cards (Femto or XE)? Did you notice any harshness? The reason I ask is that a couple of users including Emile from Taiko noticed some analytical nature or harshness with Solarflare cards.

[guiltyboxswapper]

25 minutes ago, Topk said:

1V is very low. When you went to 1V, did you lock the CPU core ratio back then and leave PBO on? This is key otherwise the bios would dramatically decrease the CPU speed to accommodate for the low voltage, impacting sound quality indeed. You can even manually lock and run the core above stock speed since there’s very little heat generated. 

 

It did indeed lock the CPU cores to 3.3ghz which was more than enough to run the workloads, I also tried 1.1v resulting in 3.6ghz core speeds and got similar results.  My guess is there's a combination of cooling and power supply pressure going between different systems, hence the different outcomes, as well as running a much harder workload than simply streaming files.

 

25 minutes ago, Topk said:

On the 5800x, I didn’t compare the 3 series with the 5 series but the 5800x has 8 cores vs 12 and 36mo cache vs 70mo. On the other hand the 5800x has the latest architecture. Your finding seems to indicate the 5900x and 5950x might be even better and that the 5 series is better than the 3 series. Could also be system dependent.

 

I'm not sure tbh.  This breaks the whole more cores = better thing IMHO, as I was expecting a downgrade tbh.  Points more towards manufacturing processes and product binning.  It is interesting how the Extreme has stuck to a certain 10-core CPU as for some reason "it sounds better than the others".

 

25 minutes ago, Topk said:

On the solarflare, did you compare with Jcat Net cards (Femto or XE)? Did you notice any harshness? The reason I ask is that a couple of users including Emile from Taiko noticed some analytical nature or harshness with this card. 

Nope, but as mentioned the x2522 is definitely the smoother on transient attacks though not enough to make the 8522 obsolete by any stretch.  With the wrong SFPs however it can easily be made to sound aggressive / robotic.  And I use Onload which takes the performance to another level entirely quite frankly (something that can't be done in Windows). 

 

Someone else on here has compared the Femto NET powered externally and still preferred his 8522 Solarfare.  I should mention the whole network chain is powered by 5x LT3045s, with an EtherRegen and OpticalModule in the path before it arrives at the SFP which does majorly improve the SQ.  You can instantly hear the difference on the network chain with the Solarflare cards, even changing 1 DC cable to the OpticalModule showed up no problem (!).

[Topk]

37 minutes ago, guiltyboxswapper said:

It did indeed lock the CPU cores to 3.3ghz which was more than enough to run the workloads, I also tried 1.1v resulting in 3.6ghz core speeds and got similar results.  My guess is there's a combination of cooling and power supply pressure going between different systems, hence the different outcomes, as well as running a much harder workload than simply streaming files.

 

I'm not sure tbh.  This breaks the whole more cores = better thing IMHO, as I was expecting a downgrade tbh.  Points more towards manufacturing processes and product binning.  It is interesting how the Extreme has stuck to a certain 10-core CPU as for some reason "it sounds better than the others".

 

Nope, but as mentioned the x2522 is definitely the smoother on transient attacks.  With the wrong SFPs however it can easily be made to sound aggressive / robotic.  And I use Onload which takes the performance to another level entirely quite frankly (something that can't be done in Windows). 

 

Someone else on here has compared the Femto NET powered externally and still preferred his 8522 Solarfare.  I should mention the whole network chain is powered by 5x LT3045s, with an EtherRegen and OpticalModule in the path before it arrives at the SFP which does majorly improve the SQ.  

 

37 minutes ago, guiltyboxswapper said:

It did indeed lock the CPU cores to 3.3ghz which was more than enough to run the workloads, I also tried 1.1v resulting in 3.6ghz core speeds and got similar results.  My guess is there's a combination of cooling and power supply pressure going between different systems, hence the different outcomes, as well as running a much harder workload than simply streaming files.

 

I'm not sure tbh.  This breaks the whole more cores = better thing IMHO, as I was expecting a downgrade tbh.  Points more towards manufacturing processes and product binning.  It is interesting how the Extreme has stuck to a certain 10-core CPU as for some reason "it sounds better than the others".

 

Nope, but as mentioned the x2522 is definitely the smoother on transient attacks.  With the wrong SFPs however it can easily be made to sound aggressive / robotic.  And I use Onload which takes the performance to another level entirely quite frankly (something that can't be done in Windows). 

 

Someone else on here has compared the Femto NET powered externally and still preferred his 8522 Solarfare.  I should mention the whole network chain is powered by 5x LT3045s, with an EtherRegen and OpticalModule in the path before it arrives at the SFP which does majorly improve the SQ.  

CPU sound: I think we all agree that each CPU range sound different and some definitely better, and yes Emile favors the silver dual Xeon due to the way it sounds

 

CPU cores: the Extreme does have 2x10=20 cores in total, split between core and endpoint and dedicated as such. This, in conjunction with the sound of the CPU itself, in my view - is part of the recipe for good sound.

So it appears part of the recipe is not only the CPU sound but the number of cores and the core isolation, in conjunction with the MB itself (and many other factors). In other words, it’s unlikely that the single CPU Xeon version of the same Asus Sage MB would sound as good since it has only 10 cores.
 

The next test to validate the core/power theory would be to move up in cores within the 5 series and assess impact on SQ. But it does seem to indicate the inherent CPU sound of the 5 series as a CPU range might sound better than the 3 series. 
 

I also want to mention that my test was with euphony (which allows for core isolation), not HQ player, and without upsampling, on a single PC setup. It would be interesting to know the rest of your system for context.

 

In other words having 16 cores and dedicating some to the Core and some to the player (gstp) in euphony in a single PC/CPU setup is some kind of replication of the taiko setup but on a smaller scale and without ram isolation.