A toast to PGGB, a heady brew of math and magic

[Zaphod Beeblebrox]

4 minutes ago, MemoryPlayer said:

So, playing 64/768 sounds the same as 32/768?

No most likely not, when playing 64/768, your player is most likely either going to use dither (or worse truncate) going from 64bit doubles to 32bit integers that your DAC accepts, this will degrade the sound quality. it is better to stay at 32/768 if you do not intend to do any further processing (including digital volume control, EQ etc.) You will also be reducing the size by half.

[Zaphod Beeblebrox]

@MemoryPlayer I should further qualify what I said, if your player does not process the 64bit files in any way and reads and keeps them as 64bits, then it could losslessly play them back into a DAC that accepts 32bits. In this case there will be no change in quality (but just an unecessary doubling of size). But any simple operation the player does would affect the quality like I mentioned previously. The only reason to keep them in 64bits in Audirvana would be to do EQ outside of PGGB. But if Audriana reads them as 32bit floats, then that would lead to truncation. 

[MemoryPlayer]

8 hours ago, Zaphod Beeblebrox said:

@MemoryPlayer ...The only reason to keep them in 64bits in Audirvana would be to do EQ outside of PGGB. But if Audriana reads them as 32bit floats, then that would lead to truncation. 

If I want EQ outside with HQPlayer or upsample to DSD256 with Audirvana or HQPlayer, is better use 64its, am I right?

[Zaphod Beeblebrox]

7 minutes ago, MemoryPlayer said:

If I want EQ outside with HQPlayer or upsample to DSD256 with Audirvana or HQPlayer, is better use 64its, am I right?

Yes, that is correct.

[Atriya]

Will soon acquire a powerful machine with a 32-core CPU, and initially 64 GB but support for up to 2 TB (!!) of DDR5 RAM. This machine will have my PGGB license for sure. :)

[Zaphod Beeblebrox]

30 minutes ago, Atriya said:

Will soon acquire a powerful machine with a 32-core CPU, and initially 64 GB but support for up to 2 TB (!!) of DDR5 RAM. This machine will have my PGGB license for sure. :)

Do share the build details too, as it may help others.

[Atriya]

23 minutes ago, Zaphod Beeblebrox said:

Do share the build details too, as it may help others.

 

Here it is:

 

https://www.dell.com/en-us/shop/workstations-isv-certified/new-precision-7875-tower-workstation/spd/precision-t7875-workstation/xctopt7875mtus_vp?configurationid=3546e7d1-5d8d-4ced-9298-d591b527b3c7

[Zaphod Beeblebrox]

53 minutes ago, Atriya said:

 

Here it is:

 

https://www.dell.com/en-us/shop/workstations-isv-certified/new-precision-7875-tower-workstation/spd/precision-t7875-workstation/xctopt7875mtus_vp?configurationid=3546e7d1-5d8d-4ced-9298-d591b527b3c7

What DAC are you using now?

[Atriya]

2 minutes ago, Zaphod Beeblebrox said:

What DAC are you using now?

 

Weiss DAC501-Mk2... which will stay for the foreseeable future...

 

I do 8fS/32-bit into the DAC, with noise shaping.

[kennyb123]

17 minutes ago, Atriya said:

 

Weiss DAC501-Mk2... which will stay for the foreseeable future...

 

I do 8fS/32-bit into the DAC, with noise shaping.

How much of an improvement does PGGB bring to that DAC?

[Atriya]

4 minutes ago, kennyb123 said:

How much of an improvement does PGGB bring to that DAC?

 

A subtle but noticeable improvement in overall realism that I was able to successfully identify in A/B.

[rayon]

@Zaphod Beeblebrox could you please comment on certain fields in PGGB_user_config_256.json:

 

• mode
  • What is this field and what are the options?
• dither
  • What are the options? Especially as some guides talk about switching off dither somehow. Is it still possible?
• pmode
  • What is this and what are the options?
• as_is
  • Seems to be boolean, but what does it control?
• rbits
  • What is this and what are the options?

I didn't find any explanation for these anywhere, but they are still available in configuration, so it would be nice to know what these do as I may like to try things out.

[Zaphod Beeblebrox]

42 minutes ago, rayon said:

@Zaphod Beeblebrox could you please comment on certain fields in PGGB_user_config_256.json:

 

• mode
  • What is this field and what are the options?
• dither
  • What are the options? Especially as some guides talk about switching off dither somehow. Is it still possible?
• pmode
  • What is this and what are the options?
• as_is
  • Seems to be boolean, but what does it control?
• rbits
  • What is this and what are the options?

I didn't find any explanation for these anywhere, but they are still available in configuration, so it would be nice to know what these do as I may like to try things out.

It is just saving the state of UI, best not to mess with it as it could lead to unpredictable behavior.

[rayon]

9 hours ago, Zaphod Beeblebrox said:

It is just saving the state of UI, best not to mess with it as it could lead to unpredictable behavior.

Ok, thank you.

[Atriya]

@Zaphod Beeblebrox, have you ever considered making a PGGB license usable on 2 machines instead of just 1? For those of us with multiple machines but varying degrees of access to them, this would be a great convenience. It's not like my actual usage of PGGB would double - it would remain about the same - but it would provide some flexibility.

[Zaphod Beeblebrox]

35 minutes ago, Atriya said:

@Zaphod Beeblebrox, have you ever considered making a PGGB license usable on 2 machines instead of just 1? For those of us with multiple machines but varying degrees of access to them, this would be a great convenience. It's not like my actual usage of PGGB would double - it would remain about the same - but it would provide some flexibility.

I am considering issuing a license through USB dongle to make this easier, no ETA yet.

[Atriya]

Just now, Zaphod Beeblebrox said:

I am considering issuing a license through USB dongle to make this easier, no ETA yet.

 

Yes, that approach would be perfect!

[PaperBoat]

@Zaphod Beeblebrox

PGGB-RT is not showing in the Foobar context menu after the fresh installation of Windows 11 but was working fine before on the same system. Please help.

[taipan254]

Hi everyone:

I will be attending CanJam NYC; if anyone on this thread is attending as well, please let me know as I'd love to meet you!

[Xoverman]

Hello everyone from German,

first of all i would like to congratulate you to this great product !

Im just starting to evaluate the sound quality on my extremely resolving system.

But i already have two things that i would really appreciate:

1 - A precheck of the installation process if the Visual C++ Redistributable" is already installed or not !

2 - Support for .W64 input file format  :-)

 

Cheers

[Zaphod Beeblebrox]

36 minutes ago, Xoverman said:

Hello everyone from German,

first of all i would like to congratulate you to this great product !

Im just starting to evaluate the sound quality on my extremely resolving system.

But i already have two things that i would really appreciate:

1 - A precheck of the installation process if the Visual C++ Redistributable" is already installed or not !

2 - Support for .W64 input file format  :-)

 

Cheers

Thank you.

 

While all of the processing logic of PGGB is in C++, I am using Matlab runtime for some of its functionalities like graphing, encode/decode audio files. this puts some constrains on what I can do:

 

1. Matlab runtime typically installs the VC++ redistributable if it needs them, but sometimes fails, and the installer does not provide me with an option to pre-check and warn the user.
2. .W64 format is not yet supported by the native audio decoders Matlab runtime provides as it is a less common audio format.

However, PGGB-IT! which uses the same C++ processing as PGGB, supports '.W64' format, and uses a lighter .Net interface.

[Zaphod Beeblebrox]

🍿Something new is coming to a DAC near you....

 

[austinpop]

If my early impressions are any indication, this will be worth the wait! 

[taipan254]

Very cool. 
 

I'm excited to hear how folks like DSD vs PCM on ESS and AKM chip based DACs. And how DSD on those DACs compare to PCM on chord DACs. 

[seeteeyou]

So far the biggest hurdle for both DSD512 and (especially) DSD1024 should be getting the right hardware that could handle the workload, while NOT taking THAT time to get things done that is.

 

 

While stuff like this wouldn't be exactly feasible for the average household unless we've got our own basement, having 96 cores / 192 threads would totally smoke a single 13th Gen Intel Core i9 thanks to the power of combining 4 "monsters"

 

https://valid.x86.fr/bench/d8kzm4/32

 

Due to the sheer number of cores, that's still performing pretty darn well when compared to previous generation of Threadripper PRO that's like $5,000 at the moment

 

https://valid.x86.fr/bench/d8kzm4/33

 

Since Broadwell (released back in 2016) should be considered a bit long in the tooth already, used pricing of each 24-core CPU turned out to be as little as 30 bucks a pop at the moment

 

$300 for the barebone

 

https://www.ebay.com/itm/375282366811

 

$649 for 48 cores / 128GB of RAM

 

https://www.ebay.com/itm/154477919929

 

$1,649 for 96 cores / 256 GB of RAM but obviously the bulk of the costs should be coming from adding quite a bit of DDR4, the rest of the hardware wouldn't be worth all that much after all these years

 

https://www.ebay.com/itm/155888750827

 

We're talking about 4U servers here, therefore they're like super heavy for shipping. Cooling fans are gonna be LOUD while all that noise would be quite annoying unless we've got a garage or basement etc. Most importantly, the electricity bill will add up pretty quickly when we're having a blast with PGGB 24 hours a day.

 

Other than that, we aren't really sure if PGGB were able to utilize so many cores (close to 100) while benchmarking might not always reflect the actual performance of running a real-world application.

 

---

 

To get the "best" performance available, we might also borrow a little something from this guy

 

https://www.whatsbestforum.com/threads/introducing-olympus-olympus-i-o-a-new-perspective-on-modern-music-playback.37939/page-52#post-936875

Quote

- The Olympus has 1 AMD Epyc 4th gen processor with 4*8 physical CPU cores for a total of 32 physical CPU cores (64 virtual), 128Gb DDR5-4800 Memory, PCIe gen5 interfaces, uses MCIO (PCIe) gen5 U.2/U.3 music storage drives and runs Windows 11

 

Benchmarks looked pretty darn well for both 8534P and 8534PN, though most likely Taiko Audio Olympus should have 8324PN or else passive cooling might be a bit too challenging

 

https://www.phoronix.com/review/amd-epyc-8534p-benchmarks/6

 

 

64 cores (200W TDP) for roughly $3,300, that ain't too bad at all since we're talking about $4,950 MSRP

 

https://www.ebay.com/str/alinctechnologies

https://www.resellerratings.com/store/Alinc_com

https://alinc.com/amd-siena-8534p-64c-2-3g-200w-03ld835/

 

ATX motherboard for a nice price

 

https://www.taobao.com/list/item/742819554881.htm

https://www.gigabyte.com/Enterprise/Server-Motherboard/ME03-CE0-rev-10

https://www.alibaba.com/product-detail/Ga-EPYC-8004-ME03-ME33-ME03_1600998173388.html

 

Costing quite a bit more from US, though

 

https://www.provantage.com/gigabyte-technology-ms03-ce0~7GIGB1C9.htm

https://www.provantage.com/gigabyte-technology-me03-ce0~7GIGB1E4.htm

 

The costs of DDR5 RDIMM with ECC looked fine and dandy to me, whether it's 64GB or 48GB

 

https://www.ebay.com/itm/225705329877

https://www.ebay.com/itm/325805936725

 

48GB UDIMM without ECC would cost about $27 cheaper from the same seller, not really all that much of a difference while 64GB UDIMM might be somewhat expensive at the very beginning

 

https://www.ebay.com/itm/204683132592

 

With 1 DIMM per channel, we'll get 100% speed at 4,800MHz. And then 75% speed @ 3,600MHz with 2 DIMMs per channel.

 

---

 

Just a "regular" ATX motherboard with 64-core processor @ 200W TDP, I really don't think the cooling system would be all that different then?

 

And then we've got 96 CPU-direct PCIe lanes with EPYC 8004, one day maybe adding GPU could be quite helpful if PGGB were able to support that down the road. Having everything (i.e. quite a few SSDs for caching purposes plus the GPU) connected to the CPU directly wouldn't be such a bad idea.

 

Not to mention the fact that in general AMD seemed to be "just a little bit more" realistic when it comes to the estimation of peak power, I wouldn't even wanna know what to do about cooling something that would go totally nuts

 

https://www.anandtech.com/show/21124/amd-ryzen-threadripper-7980x-and-7970x-review/6

 

Most importantly, we might never know the longevity of consumer-grade hardware if we're running PGGB non-stop. 24/7 operations at full blast might not be suitable for something that wasn't designed to be used that way in the long run.