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

[davide256]

Just summarizing what I think I've learned reading this thread

1. 32gb RAM advised for PGGB machine with ~200 gb virtual memory

2. file size increases 8x for CD rate so my library needs would go from 1TB to 8TB+

3. PGGB converted files  for CD rate are equivalent to DSD sizes. Which means 100mb port on an Etherregen is too slow... I'd already moved mine

to the 1GB port because of frustration when trying to buffer DSD

 

Also sounds like it would be wise to upgrade NAS to dual Ethernet port, but less choice on the music server end.

Giving this a trial whirl now

[davide256]

16 hours ago, Zaphod Beeblebrox said:

128GB virtual memory on an internal disk with 200GB free space

 

 That is true for 8FS conversion, if your DAC accepts 16FS 32 bits, expect about double that if you plan to convert every album in your library.

 

If you can add internal storage for PGGBd files as a Juke box, that may be preferable.

To date all direct attached media storage solutions for NUC I've tried have been deficient for SQ compared to NAS, including NVMe. 

 

I have HQPlayer embedded, am using a Denafrips IRIS DDC so actual DAC rate I can do at this point is 768 max dropped to 192 for IRIS output.

[davide256]

Just now, kennyb123 said:

Just wanted to mention that this won't give you twice the bandwidth.  What two ethernet ports on a NAS will give you is the ability for two clients to get a full gigabit of throughput.  So if your NAS is serving up music while at the same time another device in your home Is also accessing your NAS, both should get a full gigabit throughput.

 

Bonded ports do give you 2GbE if the OS supports it. While the distant end may only support 1GbE, bonding helps improve throughput efficiency vs "redlining" a single GbE

port that needs to talk to other things on your network. Used as common practice with the carrier network I worked for prior to 2016 due to the dearth/slow delivery at that time

of 10GbE circuits

[davide256]

2 hours ago, Zaphod Beeblebrox said:

There is only one way to find the answer to that question ...

 

It is hard to predict if PGGB is worth the while for someone. Using PGGB does not mean one has to abandon streaming. If it brings enough of an improvement,  a hybrid approach can be used where a subset of albums can be remastered.

I like what I'm hearing with Redbook across a 1 GbE wired ethernet LAN, buffering ~8GB is reasonably fast. But one would to have to shell out more money

for NUC RAM with Euphony.. opera recordings would likely require the max of 32GB RAM for playback buffering.

[davide256]

So far at 32 bit default settings

John Williams HDCD "Rosewood" sounded great, smoother, more detailed, better bass

Alan Silvestri 96/24 "Ready Player One" toss up, no real improvements

Howard Shore 44/24, "The Hobbit" sounds better at native rate, PGGB creates irritating artifacts

 

A Puzzle, perhaps DAC/server limitations? will try 24 bit next

[davide256]

12 hours ago, Zaphod Beeblebrox said:

Before I can suggest the bit rate, what is the highest  sample rate supported by your DAC, is it 192kHz 24 bits? If that is the case, you should use PGGB to output at 4FS rate and it is likely a R2R DAC and start with bit depth of 19 bits or 20 bits.

DDC = Denafrips IRIS which supports 768/24bits in over USB. Output is coax, max for IRIS is 192/24 PCM or DSD64 DOP

DAC's are Metrum Octave (192/24bits) or Chord Mojo (768/32bits). I prefer the Metrum for mid range integrity but the Mojo has better dynamics.

 

Didn't realize until reading the manual the IRIS only supported 24 bits so should have started there. which might also matter

for my HQPlayer configuration. Duhh.

 

If an option for flac conversion allowed sizable benefits retention  and and keep meta tagging that would be hugely important. I originally had my NAS library as wav

then gave up because of the challenges with meta tagging wav for player needs.

 

[davide256]

44 minutes ago, austinpop said:

 

Are you feeding the Mojo through the DDC, or only the Metrum? I didn't think so, but just want to clarify: to get the benefit of PGGB on the Mojo, you'll want to process files to 32/16fs as ZB said, and pass them directly to the Mojo via USB.

I'll try both. I eventually plan to upgrade to a Denafrips Pontus so that I can use I2S/768 from the IRIS. The Mojo really doesn't gain much from the IRIS

[davide256]

38 minutes ago, austinpop said:

 

Just keep in mind that you're trying to offload the digital filtering in your DAC to PGGB. With the Mojo (and Chord DACs in general), this means your goal is to bypass the WTA1 stage in the DAC. To do so, you have to present the DAC with a 16FS signal, which you can only do through USB or DBNC (not sure Mojo supports this? H2, TT2, DAVE all do).

 

Inserting a DDC that can only output 4FS (176.4/192) in the path will defeat this. So for PGGB evaluation, you really do NOT want the Iris in the path!

Doesn't defeat it completely since a well mastered HDCD (44/20) recording improved.

[davide256]

A request. Is it possible for the program to (a) allow a named default output main directory  (i.e. "PGGB Copy") (b) to create the folder name for whatever is copied under

this directory as a default for processing output ?  Its a bit clunky having to recreate your album directory on copy, would be nice if with license you could select a directory

(i.e. "DSD"), tell it to convert and come back later to find all subfolder naming had been kept, a batch conversion.

 

[davide256]

11 hours ago, Zaphod Beeblebrox said:

Make it so 🖖

 

PGGB v1.0.80 now has an option 'FLAC when possible'. The tags and album art will be retained too. By default if the output rate is set to 8FS or lower and the bit depth is set to 24 or lower, output will be as FLAC files. To override see here.

 

As I see disk space as a major concern, I am open to implementing WavPack output format for higher rate files and expect compression down to  30% so one could fit almost three times as many albums. You can try DBpoweramp with PGGB files as DBpoweramp transfers metadata, then see if it works with your setup.. If there is enough interest, I will consider it.

Tried this just now with "The Yes Album" 44/16 version, set max rate to 384/24 and got FLAC. But sound level came out much too low, like a 15db reduction. Did I miss a step?

[davide256]

17 hours ago, Zaphod Beeblebrox said:

For those with space constraints, version 1.1.10 is now available that supports FLAC and Wavpack lossless compressed output formats but improves tagging of .wv format and long tracks will no longer be split if Wavpack/FLAC compression is enabled. 

Dropbox says file was deleted?

[davide256]

I just have to ask; what is PGGB doing that makes SACD rock music sound authentic instead of British polite?

 

As a Pink Floyd fan I couldn't wait to get the SACD of "Wish You Were Here"... but it was a snooze either played on my Oppo 103 or ripped and played streaming, 

just didn't do justice to the vinyl version. Now with PGGB FLAC conversion the SACD rip has the detail, power of the original. Gonna have to budget for this,

even if its just for converting my SACD rips.

 

Also listening to this PGGB converted to FLAC at the moment, its terrific even if only HDCD for source. My amp is getting a workout.

 

 

 

 

[davide256]

14 minutes ago, ted_b said:

Davide256, all dacs have sweet spots, and yours (Metrum Octave if your link is still up to date) clearlyhas a PCM sweet spot, if we are to beleive your reactions..  This is nothing against PGGB; in fact, it is saying that PGGB does a better job of converting DSD to PCM than whatever dac (Oppo or Metrum) you'd been using.  My $.02

353/384/24 to the IRIS DDC which down converts to 192/176/24  to Chord Mojo. There's very clearly some dynamic range expansion going on as well as low bass drum

solidity thats not there in unprocessed PCM or DSD rips. If I shared a wall with neighbors, they would be knocking on my door to complain right now 😁

[davide256]

7 minutes ago, Zaphod Beeblebrox said:

why not feed Chord Mojo with 705kHz direct from PGGB?

Doesn't sound as tight and defined as using the IRIS. Part of the problem is finding a micro USB cable as good as Lush 2.

[davide256]

52 minutes ago, Nikko1960 said:

 

I'm sitting this out.  If it's worth doing, then Jussi will come up with the gold standard solution, having been researching this tech for years with many, many happy customers.

Was wondering about this the other day... seems like Jussi could add on to the existing HQPlayer an output to disk option?

[davide256]

So

1) license obtained

2) 6TB NAS red plus drive ordered to accommodate file conversion of 800GB music, mostly CD rate

3) 16GB additional memory ordered for PGGB machine to reach 32GB for better conversion speed

4) default drive used for conversion is NVMe with 128GB virtual memory and 500GB free space

 

 

Any other steps one should take for the most efficient conversion of an existing library?

 

[davide256]

22 hours ago, Zaphod Beeblebrox said:

why not feed Chord Mojo with 705kHz direct from PGGB?

I'm there now. Was listening to some converted Richard Strauss, lots of high frequency violin, very irritating with Metrum Octave or Mojo fed through the IRIS, just fine

USB direct to the Mojo. Looks like the IRIS stays shelved until I can use its I2S connection to get full data rate with another DAC.

[davide256]

Having some difficulties with FLAC conversion of downloads from Music Society of Sound, about 4 tracks so far across a dozen

albums have given this error. Track below plays fine. Thoughts?

 

"Error Message:
File could not be read due to an unexpected error. Reason: File contains data in an unknown format.
While processing: \\?\D:\Music\MusicSociety of Sound\Apparatjik\Square Peg in a Round Hole Draft 7-FLAC24\01. Time Police.flac
Total time to process file: 0.071282 secs"

 

 

[davide256]

Have had to add to my list of items for PGGB library conversion, a FLAC library error checking solution. See thread started below. Finding/fixing errors before you

start conversion reduces exasperation

 

 

 

[davide256]

5 minutes ago, taipan254 said:

I have to agree on this. I got the iris ddc and tried my pggb test files with my lps powered 2qute. I hear more of pggb's magic via an optical cable from iris ddc downspampled to 192k than hardwired usb at full sample rate.

 

I think the source still matters depite pggb magic (dell laptop vs. Iris ddc)! Will experiment more in the next few weeks. 

Kind of a trade off for me so far... using 384/24 PGGB conversion, if the content is largely below 8khz the IRIS output is more detailed and dynamic. But if there is a lot of

content at 8khz and above the treble becomes irritating vs going USB direct to Chord Mojo. Strangely enough, using HQPlayer with upsampling disabled, TPDF noise filter

enabled mitigates the treble irritation from the IRIS but thats just too many layers of alteration for me to trust.

[davide256]

Having a problem... trying to create some listening samples at higher rates, instead of getting   .wav for output file extension I'm getting .wv

and euphony doesn't recognize any sample rate for this. Suggestions?

[davide256]

On 5/26/2021 at 11:08 AM, taipan254 said:

I have to agree on this. I got the iris ddc and tried my pggb test files with my lps powered 2qute. I hear more of pggb's magic via an optical cable from iris ddc downspampled to 192k than hardwired usb at full sample rate.

 

I think the source still matters depite pggb magic (dell laptop vs. Iris ddc)! Will experiment more in the next few weeks. 

Something I'm trying now, step the sample rate down in Euphony to the output rate (176/192)  on coax/optical from IRIS. Using Reference Recordings RR-83

Ein Heldenleben track 1  as my torture test for treble irritants. Will be curious what you think. My least offensive result so far was 705/20 down-sampled by IRIS

but could hear loss of detail vs 705/32. Giving 176/32 a whirl next out of curiosity

[davide256]

Just now, Zaphod Beeblebrox said:

Yes the better test is directly converting to 176/32. I expect it to sound better than letting IRIS do the downsampling 

yea.. I'm suspecting noise creeps in when the IRIS down converts... will see if @taipan254 can corroborate.

[davide256]

22 hours ago, davide256 said:

yea.. I'm suspecting noise creeps in when the IRIS down converts... will see if @taipan254 can corroborate.

 

from Denafrips specs for IRIS

 

DSD DSD64-DoP On All Output

Up to DSD512 On USB Input & I²S Output

PCM24bits / 44.1, 48, 88.2, 96, 176.4, 192KHz On All Output

Up to 768kHz On USB Input & I²S Output

 

getting least objectionable artifacts feeding 176/24 to IRIS. The IRIS accepts 32 bit but converts on PCM output to 24,

(excluding I2s?) where like down sampling I hear added treble irritants for 32 bit source conversion on coax/optical output

 

Was hoping to stop at 176/24 FLAC size files but I really like the 705/24 wav down sampled by Stylus

to 176/24 to feed the IRIS... doesn't have the treble irritants the IRIS creates from down sampling.  Going to

see if others like me will request Zeljko add wave pack support to Stylus

[davide256]

32 minutes ago, Zaphod Beeblebrox said:

Have you tried creating 176/24 files directly using PGGB and do you still prefer downsampling done by Stylus?

Done the range of 176-353-705 rates and 24/32  bit depth in wav with my test track. Puzzles me that I can hear more with Stylus downsampling but details are

better enunciated with 705>176 than they are with native 176. I cache all music to a FEMTO clocked NVME drive + the IRIS uses a FEMTO clock. DAC tested

with was NOS Metrum Octave.... the Mojo internally up-samples PCM to 705/768 which is its own can of worms.