Jump to content


Photo

Newbie question about buffer size and sound quality


  • Please log in to reply
6 replies to this topic

#1 ted_b

ted_b

    Die Hard

  • Members
  • PipPipPip
  • 345 posts

Posted 14 April 2013 - 01:23 PM

I stumbled across this thread from last Winter, in the Computer Audio forum section, which basically mirrors what i am finding in day three of my Jplay trial evaluations.  Ultrastream to Meitner MA-1 on Caps V3 Lagoon (low power 1.86ghz) with 4GB memory (thinking seriously of going to 8GB for all of $60); Hynes linear power supply.

http://jplay.eu/foru........ect-on-sq/

 

That is, that on my mission to get to directlink (or as close as possible) for all three categories (redbook, 96k and 192k, which i assume also handles DoP issues with DSD) I have come to an early conclusion that hiccups/static are not the prime features to flag whether buffer size is the right one.  It seems on many occasions already i have found that lower buffer (say 4, for 192k) will bring out more detail and a nice lack of hiccups or static, yet it will present the music in an almost thin or digital way...with a lack of musical bass and a bleaching of timbres.  Add a few samples (and I mean a few, like maybe just the next click over) and voila the magic is there.  Am I going insane or does anyone else experience this....click free playback that still needs a few more samples to kind of "round out" (but not round off or sound muted in any way)?

If this is true, will this kind of subjective analysis be required to futz around with US as well?  I am at 180 or so, as anything lower also exhibits some lightness and/or gltiches.

And my mention about adding RAM...does amount of RAM help both buffer and US size?  Does brand or speed/quality of RAM matter?
Controlpc:  CAPS V3 Lagoon (aka my CAPS V2+), 8 GB, WS 2012 GUI, JRiver/jplay ultrastream dl/dl/dl, Exasound multichannel driver for mobo USB/ifi iUSB, dual ethernet (LAN NIC and Jplay via mobo adapter) , LAN connected to switch and Synology DS1513+ NAS, jplay direct connect to audiopc, Splashtop VNC (from ipad or home office desktop), Remote desktop shortcut into audiopc, AudiophileOptimizer, sound sig 4C, Powered by Parts Express 12V linear (temporary), SSD powered by Anker E4 Audiopc:  Asus P8H77-I based CAPS Zuma (i7), 16 GB, WS 2012 GUI-to-core, jplay, Meitner and Chord drivers, PPA PCIe USB card (powered by 5V Red Wine modded linear ac-based Acopian ps) or SOtM PCIe card (powered by SOtM dual battery), AudiophileOptimizer, sound sig 1B, Powered by Paul Hynes SR7-EHD (dialed to 12V),SSD powered by iFi 5V ps, Acoustic Revive RLI-1 LAN isolator on mobo ethernet input, PPA TZ Yun III USB cable to DAC

#2 Marcin_gps

Marcin_gps

    JPLAY & JCAT Founder

  • Administrators
  • 4,036 posts

Posted 15 April 2013 - 08:10 AM

Hi Ted,    Lower buffer is more revealing imho. The same with lower UltraSize. You need a faster CPU. RAM performance is also important, but far less at least for getting lowest buffers to work. For sound quality I choose low power RAM, reduce operating frequency to 1066MHz and timings to minimum.    Cheers,  Marcin

Follow my hi-fi journey on YouTube


JPLAY FEMTO: a complete network music software player for the most demanding audiophile.

Want to squeeze more from your PC-audio? Hear the difference with JCAT
 
Join Computer Audio Enthusiasts group on facebook

#3 kingr81

kingr81

    Newbie

  • Members
  • Pip
  • 1 posts

Posted 10 May 2013 - 04:58 PM

Just wanted to post my agreement with ted_b.  The "bleaching of timbres" is exactly what I found when using directlink –>v link via optical–> rega dac.  This is especially noticeable on the acoustic guitar.  When setting the buffer higher, this problem is taken care of.  I did extensive tests comparing two high-priced cd transports feeding the dac to jplay w/ foobar feeding the dac.  While I did not experience the "bleaching of timbres" (a perfect description) problem when using the CD players, the sound as a whole seemed somewhat smeared and glassy (for lack of a better term)–I assume this is the sound of jitter from reading the discs live.  However, going back to the computer transport with jplay w/ foobar (using directlink), everything came into focus, yet lost some of its timbrel qualities.  It took ages for me to figure out to try increasing the buffer size, which took care of it.  I should add that I also turn off all of the services in my computer except for jplay and three others that always stay on automatically.  This seems to increase sound quality as well. Equally important–run the laptop on battery power.  I suppose I don't know enough about what buffer size actually does to understand why this is happening, but I can live with the 7-8 second latency between pushing play and hearing the music.  For what it's worth–as someone who spends a significant amount of time composing music on computer programs, I find the sound with the higher buffer setting to be more similar to what I hear within the actual digital audio workstations themselves.   BTW:  I haven't even begun to mess with the settings in the registry part yet.  As I spend more time messing with my audio equipment than actually enjoying music lately, the endless permutations of settings there will just have to wait until I'm ready to go down that road.  More options=less happiness--> I'm starting to think so towards the end of my first year of audiophilia.        

Edited by kingr81, 10 May 2013 - 07:14 PM.


#4 The Chaplin (Peter C)

The Chaplin (Peter C)

    Die Hard

  • Members
  • PipPipPip
  • 326 posts
  • LocationLondon

Posted 10 May 2013 - 10:51 PM

As Marcin says........ you need low latency ram and a cpu that can use it. Then try ultrastream with a small size buffer, dual pc, the most analogue sounding digital I have heard. Usb not optical.

[JPlay FemtoServer][BubbleUPnP(Andriod)][ControlPC XM21M(@16v) Shuttle xh61v i7-3770t PX-128M5M Lan x2][WinServer 2012R2][JPlay 7Dalt KS Direct DACLink 1000Hz Clock 0.5ms AO 1.40][cat8@1Gb 3m][Cisco 2960G-8TC-L][cat8@1Gb 1m][AudioPC XM21M(@12.3v) picoPSU-160-XT Streacom FC5 B75MR2.0 Intel i7-3770t@1.6GHz JCat Usb][WinServer2012 R2 AO 1.40][CustomUSBcable no+5v][Auralic Vega filter 1, exact clock][Van Damme StarQuad Gold Neutrik XLR][ATC SCM40A][vdh 1st ultimate][Stax SRM313 SR007A][LAT & Custom Power Cables][2 Separate Ring Mains Filtered]

 


#5 drez

drez

    Advanced

  • Members
  • PipPip
  • 98 posts

Posted 03 January 2014 - 02:01 PM

Zombie thread revive - but I thought I might post my findings here.  From my experience small buffer size is beneficial only in so much as it can fit on one Ram chip.  On my system I can hear the buffer walk from one chip to two - there is a noticeable non-linearity where the extra chip comes into play which adds a veil to the sound.  I try to stay just below this sound for maximum sound quality - any lower does not net much if any sound quality benefits for me.

#6 Marcin_gps

Marcin_gps

    JPLAY & JCAT Founder

  • Administrators
  • 4,036 posts

Posted 04 January 2014 - 09:09 AM

Are you referring to MiniCache?

Follow my hi-fi journey on YouTube


JPLAY FEMTO: a complete network music software player for the most demanding audiophile.

Want to squeeze more from your PC-audio? Hear the difference with JCAT
 
Join Computer Audio Enthusiasts group on facebook

#7 drez

drez

    Advanced

  • Members
  • PipPip
  • 98 posts

Posted 06 January 2014 - 12:38 PM

I probably shouldn't speculate about what is causing the nonlinearities, as I have no direct experience developing the software, but maybe you have observed something similar, where certain buffer values fall outside the general trend of "smaller buffer, more revealing".  I find there are certain buffer values with a warmer than expected sound where both increasing or decreasing results in a more revealing sound.  I found this out working backwards trying to find a buffer value that guarantees no skipping.




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users