Jump to content
IGNORED

How much cpu power is required to convert PCM to DSD on the fly?


Recommended Posts

I'm basically looking to buy a new/used laptop to feed my Marantz HD Dac 1 with a DSD signal via J River. I have seen that a Zuma can do it, that CPU ranks over 9,000 on the CPU benchmark tests. I've also seen that a 2012 i5 Mac Mini cannot. That processor ranks around 4,000. As of right now it looks like a 15" Quad Core MBP is about the only thing comparable to the Zuma in terms of horsepower. Yes I want a laptop and yes I know that a headless desktop is probably better sounding. The laptop can either run Windows or OSX, I really have no preference as J River works on both.

Link to comment

Not to mess up your path but I believe that HQPlayer's "-2" series of filters will allow you to get away with less horsepower than what JRiver needs. Hopefully this will open up some laptop options for you.

Analog: Koetsu Rosewood > VPI Aries 3 w/SDS > EAR 834P > EAR 834L: Audiodesk cleaner

Digital Fun: DAS > CAPS v3 w/LPS (JRMC) SOtM USB > Lynx Hilo > EAR 834L

Digital Serious: DAS > CAPS v3 w/LPS (HQPlayer) Ethernet > SMS-100 NAA > Lampi DSD L4 G5 > EAR 834L

Digital Disc: Oppo BDP 95 > EAR 834L

Output: EAR 834L > Xilica XP4080 DSP > Odessey Stratos Mono Extreme > Legacy Aeris

Phones: EAR 834L > Little Dot Mk ii > Senheiser HD 800

Link to comment
Not to mess up your path but I believe that HQPlayer's "-2" series of filters will allow you to get away with less horsepower than what JRiver needs. Hopefully this will open up some laptop options for you.

 

On my MacBook Pro (2011 2.5 GHz i7, 16GB RAM, JRiver benchmark ~3,300), JRiver upsampling to 2xDSD uses 10% CPU. No stuttering, can do other tasks at the same time, etc.

 

HQPlayer upsampling to 2xDSD (poly-sinc-2s, ADSM7) uses 40% CPU and if I switch away from the HQPlayer window (e.g., to surf the web, etc.), starts stuttering pretty quickly.

John Walker - IT Executive

Headphone - SonicTransporter i9 running Roon Server > Netgear Orbi > Blue Jeans Cable Ethernet > mRendu Roon endpoint > Topping D90 > Topping A90d > Dan Clark Expanse / HiFiMan H6SE v2 / HiFiman Arya Stealth

Home Theater / Music -SonicTransporter i9 running Roon Server > Netgear Orbi > Blue Jeans Cable HDMI > Denon X3700h > Anthem Amp for front channels > Revel F208-based 5.2.4 Atmos speaker system

Link to comment
On my MacBook Pro (2011 2.5 GHz i7, 16GB RAM, JRiver benchmark ~3,300), JRiver upsampling to 2xDSD uses 10% CPU. No stuttering, can do other tasks at the same time, etc.

 

HQPlayer upsampling to 2xDSD (poly-sinc-2s, ADSM7) uses 40% CPU and if I switch away from the HQPlayer window (e.g., to surf the web, etc.), starts stuttering pretty quickly.

 

Interesting, I had the exact opposite experience. Maybe my JRiver install isn't as clean as yours (using v19 and ASIO drivers). And then my HQP runs at 10% for any of the -2s filters, I don't get to higher numbers (and then it's only 30%) until I abandon the short filters (non -2s variants). I am up sampling to 2xDSD (DAC limit, which I am happy with).

 

Other specs (knowing you are a computer guy), CAPS v3 Zuma, Win Pro 8.1, Intel i7-4790S CPU running at 3.2 GHz, and 16 GB RAM. Typical HQP settings: poly-sync-short, ASDM7, 5644800. You can probably make more sense out of our computing environments than I can (I'm an older computer person who rarely touches the keyboard anymore).

Analog: Koetsu Rosewood > VPI Aries 3 w/SDS > EAR 834P > EAR 834L: Audiodesk cleaner

Digital Fun: DAS > CAPS v3 w/LPS (JRMC) SOtM USB > Lynx Hilo > EAR 834L

Digital Serious: DAS > CAPS v3 w/LPS (HQPlayer) Ethernet > SMS-100 NAA > Lampi DSD L4 G5 > EAR 834L

Digital Disc: Oppo BDP 95 > EAR 834L

Output: EAR 834L > Xilica XP4080 DSP > Odessey Stratos Mono Extreme > Legacy Aeris

Phones: EAR 834L > Little Dot Mk ii > Senheiser HD 800

Link to comment

On my mid-2011 i5-based Mac mini (8GB RAM), in HQPlayer 3.12, I can do poly-sinc-short-mp and DSD7, upsampling everything to DSD128 (the max upsampling I can get my DAC to do) with no problems. CPU usage is 30%-35% most of the time with these settings.

 

I recently noticed that if I try to do poly-sinc-hb (with other settings the same), that will initiate stuttering; however, poly-sinc-hb-2s works fine.

 

--David

Listening Room: Mac mini (Roon Core) > iMac (HQP) > exaSound PlayPoint (as NAA) > exaSound e32 > W4S STP-SE > Benchmark AHB2 > Wilson Sophia Series 2 (Details)

Office: Mac Pro >  AudioQuest DragonFly Red > JBL LSR305

Mobile: iPhone 6S > AudioQuest DragonFly Black > JH Audio JH5

Link to comment
Interesting, I had the exact opposite experience. Maybe my JRiver install isn't as clean as yours (using v19 and ASIO drivers). And then my HQP runs at 10% for any of the -2s filters, I don't get to higher numbers (and then it's only 30%) until I abandon the short filters (non -2s variants). I am up sampling to 2xDSD (DAC limit, which I am happy with).

 

Other specs (knowing you are a computer guy), CAPS v3 Zuma, Win Pro 8.1, Intel i7-4790S CPU running at 3.2 GHz, and 16 GB RAM. Typical HQP settings: poly-sync-short, ASDM7, 5644800. You can probably make more sense out of our computing environments than I can (I'm an older computer person who rarely touches the keyboard anymore).

 

I always find it interesting how computers with the *same* / similar hardware can run so differently given different software ;)

 

It could be the Windows version of HQPlayer is more efficient *running under Windows* than is the same / similar code running under OS X. Or it could be that my OS X platform is not running optimally for HQPlayer, or that your Windows install is really "clean", etc.

 

In any case, I would really love to see from Miska a definitive reference for each of the oversampling mode and each modulator, showing:

 

 

  1. expected CPU utilization (not as a %, but ranked in order)
  2. which works with which (e.g., I think I know some modulators will not work with some oversamplers, while some modulators are only good for the 44.1k variants vs. 48k variants, but don't know which is which)
  3. additional comments on expected audible result
  4. etc.

 

I'm as much of a hobbyist as the next guy, and I love futzing about with HQPlayer (and reading the existing documentation included in the download), but I feel as if I'm pretty much poking about at random right now LOL

John Walker - IT Executive

Headphone - SonicTransporter i9 running Roon Server > Netgear Orbi > Blue Jeans Cable Ethernet > mRendu Roon endpoint > Topping D90 > Topping A90d > Dan Clark Expanse / HiFiMan H6SE v2 / HiFiman Arya Stealth

Home Theater / Music -SonicTransporter i9 running Roon Server > Netgear Orbi > Blue Jeans Cable HDMI > Denon X3700h > Anthem Amp for front channels > Revel F208-based 5.2.4 Atmos speaker system

Link to comment
I'm as much of a hobbyist as the next guy, and I love futzing about with HQPlayer (and reading the existing documentation included in the download), but I feel as if I'm pretty much poking about at random right now LOL

 

I feel the same. All the information you want to see from Miska would be great to have, all in one place. (I know documentation's a PITA, but good documentation should reduce the number of support requests or at least make many of them easier to respond to.)

 

--David

Listening Room: Mac mini (Roon Core) > iMac (HQP) > exaSound PlayPoint (as NAA) > exaSound e32 > W4S STP-SE > Benchmark AHB2 > Wilson Sophia Series 2 (Details)

Office: Mac Pro >  AudioQuest DragonFly Red > JBL LSR305

Mobile: iPhone 6S > AudioQuest DragonFly Black > JH Audio JH5

Link to comment
I always find it interesting how computers with the *same* / similar hardware can run so differently given different software ;)

 

It could be the Windows version of HQPlayer is more efficient *running under Windows* than is the same / similar code running under OS X. Or it could be that my OS X platform is not running optimally for HQPlayer, or that your Windows install is really "clean", etc.

 

In any case, I would really love to see from Miska a definitive reference for each of the oversampling mode and each modulator, showing:

 

 

  1. expected CPU utilization (not as a %, but ranked in order)
  2. which works with which (e.g., I think I know some modulators will not work with some oversamplers, while some modulators are only good for the 44.1k variants vs. 48k variants, but don't know which is which)
  3. additional comments on expected audible result
  4. etc.

 

I'm as much of a hobbyist as the next guy, and I love futzing about with HQPlayer (and reading the existing documentation included in the download), but I feel as if I'm pretty much poking about at random right now LOL

 

Guys, let's think about this. Any one of us could do items 1 and 2 if we wanted to take the time, try all of the combinations, and then write it all up. It would be interesting to see if we all came up with the same ranked order, it would also be interesting to see if the type of music had any influence (massive orchestra versus a few violins, could do the same with rock, full band versus a single voice). Jussi has hinted from time to time regarding item 3, particularly when he adds a new filter or modulator.

Analog: Koetsu Rosewood > VPI Aries 3 w/SDS > EAR 834P > EAR 834L: Audiodesk cleaner

Digital Fun: DAS > CAPS v3 w/LPS (JRMC) SOtM USB > Lynx Hilo > EAR 834L

Digital Serious: DAS > CAPS v3 w/LPS (HQPlayer) Ethernet > SMS-100 NAA > Lampi DSD L4 G5 > EAR 834L

Digital Disc: Oppo BDP 95 > EAR 834L

Output: EAR 834L > Xilica XP4080 DSP > Odessey Stratos Mono Extreme > Legacy Aeris

Phones: EAR 834L > Little Dot Mk ii > Senheiser HD 800

Link to comment

IMO too much different computer configurations exist. Only information like i5 does not yet give full information about computer performance. So I doubt if it is realistic to fully reach what you wish. But IMO it can be useful to collect information from HQPlayer users about limits of their system - which filter combinations yet run OK and which are unusable.

 

My notebook is Fujitsu LIFEBOOK E754 [email protected] running Win10 Pro 64bit

I am using my own "Foobar playlist to HQPlayer" solution, HQPlayer 3.12 plays from RAM disk.

My XMOS & Sabre 9018 based DAC supports PCM max. 384k / DSD128 (DoP) max. 6.144 Mhz

Because of 44.1k PCM to DSD seems to be the most resource consuming, I am providing you information for this type of usage. I am providing information only about filter combinations which IMO sound best on my system.

 

What I can play with this setup without stuttering - 44.1k to DSD128

poly-sinc + ASDM7 to DSD128 5.6MHz *

poly-sinc + DSD7 to DSD128 5.6MHz *

poly-sinc-mp + ASDM7 to DSD128 5.6MHz *

poly-sinc-mp + DSD7 to DSD128 5.6MHz *

poly-sinc-short + ASDM7 to DSD128 6.1MHz

poly-sinc-short + DSD7 to DSD128 6.1MHz

poly-sinc-short-mp + ASDM7 to DSD128 6.1MHz

poly-sinc-short-mp + DSD7 to DSD128 6.1MHz

 

Plays with stuttering (unusable) - 44.1k to DSD128:

poly-sinc + ASDM7 to DSD128 6.1 MHz

poly-sinc + DSD7 to DSD128 6.1 MHz

poly-sinc-mp + ASDM7 to DSD128 6.1 MHz

poly-sinc-mp + DSD7 to DSD128 6.1 MHz

 

IMO I can reach the best sound on my system with poly-sinc + ASDM7 to DSD128 5.6MHz.

 

Lines marked with * mean running on limits of my system. Those filter combinations play well only when I don't use other programs. Setting HQPlayer process to Realtime priority helps with this - I can then browse web, write mails and playback is still OK.

 

HQPlayer 3 versions before May 2015 (I think 3.0.7 and earlier) were less resource consuming. With those older versions I could run 44.1k to DSD256 11.2 MHz poly-sinc + DSD7. I tested it with Oppo HA-1 and Yulong DA8, filter initialization took a minute.

i7 11850H + RTX A2000 Win11 HQPlayer ► Topping HS02 ► 2x iFi iSilencer ► SMSL D300 ► DIY headamp DHA1 ► HiFiMan HE-500
Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...