jabbr Posted April 7, 2015 Share Posted April 7, 2015 Working with some lower power but higher core CPUs. Does HQPlayer max out at 6 cores/threads? I'm trying to get an HQPlayer linux VM working on an older AMD 12 core machine (Athlon 64 x2, 2200 series, 2.0 ghz) which outputs DSD128 just fine but hiccups at DSD256 ... I can't seem to get the 8 core VM fully utilizing and the entire server maxes out at 36% CPU utilization. Other systems which should be interesting are the Intel Compute Stick (4 x Atom) and more particularly the upcoming D-1540 based systems which I have very high hopes for. Thoughts on where I'm going to run up against limitations? Custom room treatments for headphone users. Link to comment
jabbr Posted April 7, 2015 Share Posted April 7, 2015 Does anyone have a setup that can upsample to DSD512 reliably? Yes, my OSX 10.9 i7 4830k (6 core) machine does using 2.5 - 3 CPUs and pipeline SDM. Sent from my iPhone using Tapatalk Custom room treatments for headphone users. Link to comment
jabbr Posted April 7, 2015 Share Posted April 7, 2015 Stupid question: how do you send DSD512 from OS X? I thought that CoreAudio was unable to stream native DSD and that DoP was limited to DSD256. I go through an NAA. i.e. OS X -> NAA -> iDSD Micro Custom room treatments for headphone users. Link to comment
jabbr Posted April 9, 2015 Share Posted April 9, 2015 A Hackintosh? Its a multiboot machine, os x, win7, linux ... of course the H220 pump just died so its out of commission until I replace the watercooling system I'm waiting for the D-1540 boards to get released though and it would be really good to know if a 2.0 Ghz 8 core machine would do DSD512. It should if the threading scales up. Custom room treatments for headphone users. Link to comment
jabbr Posted April 12, 2015 Share Posted April 12, 2015 My ramdisk experiment is over. The sound was strident, flat, artificial sounding, with one note bass. That is oversimplified but not worth the effort on my system. YMMV, obviously. There is no question this whole area can be tricky. FWIW: the biggest and most dramatic improvement I've seen is by switching from direct-attached DAC (e.g. via USB) to NAA. Of course by whole workstation computer is horribly noisy with a high powered GPU and 4K monitor. I'm now upsampling to DSD512 on a mac and then listening through the low power CAPS as NAA. Custom room treatments for headphone users. Link to comment
jabbr Posted April 14, 2015 Share Posted April 14, 2015 Thanks but it apeares it's stacked deck against me trying it. So enjoy your HQ I'll delete mine , what's wrong with you people , it's odd how some people think that have it all figured out for every one else. Goodnight guys. Al Nothing is stacked against you. Like everything in engineering and systems design there are different subunits and different options which each have different strengths and weaknesses. The strength of HQPlayer is that it uses general purpose CPU resources to make less work for DAC. You need to decide whether its worth investing the time to test out in your system given the other components in your system. At end of the day it may or may not be a good fit and you'd need to determine a) is it worth trying? b) does it make improvement to your own ear? The advice you get helps you with (a Custom room treatments for headphone users. Link to comment
jabbr Posted April 14, 2015 Share Posted April 14, 2015 A little counterintuitively, I've found it seems to be easier to upsample 44.1 to DSD than upsampling starting with higher resolutions. Given enough coffee, I can upsample 44.1 to DSD in my head, and the resultant hand tremor is the readout ... only problem is that the filter gets a bit lax... Custom room treatments for headphone users. Link to comment
jabbr Posted April 14, 2015 Share Posted April 14, 2015 I'm working on setting up the scripting interface with HQPlayer embedded. Looks like this works with rygel. Goal is to control hqplayer via iPhone/iPad. Has anyone done this? Would the sony DLNA controller app work? Custom room treatments for headphone users. Link to comment
jabbr Posted April 17, 2015 Share Posted April 17, 2015 I'm not aware of the Sony controller app. But so far I've had best success on iOS with media : connect : premium app. It is not as fancy as BubbleUPnP on android, but at least it works. Seems to work. It lets me queue playlists and/or songs from playlists to be rendered by hqplayerd. HQPlayerd get them. I'm not sure how to or if I can queue albums or individual songs to be rendered ... but they can play on my iPhone. I'm not sure how to tell hqplayerd to switch to SDM output. I'm not sure how to tell hqplayerd which bitrate I want. Custom room treatments for headphone users. Link to comment
jabbr Posted April 17, 2015 Share Posted April 17, 2015 Thanks... I was looking in /usr/share/doc/hqplayer rather than /usr/share/doc/hqplayerd (duh!) Custom room treatments for headphone users. Link to comment
jabbr Posted April 19, 2015 Share Posted April 19, 2015 ... But it *is* a pain to have to convert everything to AIFF just to listen to it. I've taken to maintaining a completely separate / much smaller subset of my music just to use HQPlayer - my main 80k track collection (almost 2TB) is just too large to convert to AIFF without a substantial investment in disk space $ zfs set compression=on tank/Music Custom room treatments for headphone users. Link to comment
jabbr Posted April 20, 2015 Share Posted April 20, 2015 It is a last minute addition and starting point for possibility to externally control the desktop version. There's now a small command line program "hqp-control" included that can be used to initially add things to the playlist and clear the playlist. Main intention being to support Geoff's efforts to queue Tidal content from Kodi to HQPlayer. Control protocol is XML over TCP socket, so it works over network. I will add more things in future there, such as playback control. XML over TCP socket is good, there are those of us who could term in using an iPhone to remote control Custom room treatments for headphone users. Link to comment
jabbr Posted April 21, 2015 Share Posted April 21, 2015 It is a last minute addition and starting point for possibility to externally control the desktop version. There's now a small command line program "hqp-control" included that can be used to initially add things to the playlist and clear the playlist. Main intention being to support Geoff's efforts to queue Tidal content from Kodi to HQPlayer. Control protocol is XML over TCP socket, so it works over network. I will add more things in future there, such as playback control. Available on OS X? Custom room treatments for headphone users. Link to comment
jabbr Posted April 21, 2015 Share Posted April 21, 2015 Yes, there's a command line controller "hqp-control" included for all supported OS (/Applications/HQPlayerDesktop.app/Contents/MacOS/hqp-control). There was a mistake in OS X packaging of the utility, so I rebuilt the DMG earlier today. Please re-download if you'd like to try it out. I should have looked there before asking The general idea is that you can expose various types of scripting interfaces. This link describes the desired functionality although I'd use a different implementation ... the idea is that we could use voice recognition + scripting as a music/album/playlist selection interface. e.g. Brian's Bits - Remote Control Your Mac With Siri Custom room treatments for headphone users. Link to comment
jabbr Posted April 28, 2015 Share Posted April 28, 2015 Ok I have been reading scanning this thread for days in hopes of not asking questions that have been answered hundreds of times. But it seems like lots of questions have been asked and answered hundreds of times. I'm upsampling from my 2011 2.3 mac mini, running Mountain lion. I have a Lampizator big 7, I think it is DSD 128 so i want to have my the 5644800 option chosen? I picked the filter ASDM7 and poly sinc mp-2s. I chose the 2s because without it i was taxing my mini and getting drop outs. Now i'm using 30-40% of the cpu. I have the pipeline sdm checked because, I don't know why. I control the volume with my preamp thinking thats the best sound quality option? My volume max is -3. I have the direct sdm box checked which means if i do play a DSD it is passing through untouched? Also have the 6db gain box checked. Does this set up seem right? Could i be doing something better? I'm not getting any popping or cracking other than the normal Lampizator pops when starting or ending a listening session but not in between tracks. It sounds good, great, whatever. I'm I drag and drop listener so i haven't messed with the library function yet. Thanks in advance for any thoughts or suggestions. guf Uncheck Direct SDM and then you will up sample to DSD 128 (or whatever you ask ) Custom room treatments for headphone users. Link to comment
jabbr Posted May 4, 2015 Share Posted May 4, 2015 By "DoP file" do you mean DoP encapsulated in FLAC? HQPlayer can encapsulate DSD in PCM over USB using the DoP protocol but reads the DSD from either DSF or DFF files. Where did you get these DoP files? Do you have access to DSF versions? Custom room treatments for headphone users. Link to comment
jabbr Posted May 13, 2015 Share Posted May 13, 2015 My i7-4930k uses roughly 3 cores for octo so i7-5820k should be good. What I'd like to know is whether I can drop down to 2 ghz but use more cores e.g. D-1540 Custom room treatments for headphone users. Link to comment
jabbr Posted May 15, 2015 Share Posted May 15, 2015 Thanks for the info. Your i7-4930 still cannot handle up sampling to DSD512 with polysinc? Jabbr seems to suggest that his i7-4930k system is more than capable of doing that. Maybe he is over-clocking and he is using a different filters configuration. Just checked on cpubenchmark.net, the 4930k is actually scoring a bit higher than the 5820k, interesting. If you search back you'll see that I had to check PipelineSDM (and Miska enabled this for OS X). I've used ifferent verions of polysinc ... usually polysinc-mp-2s. Various filters no problem ADSM7. Overclocking is not necessary. Again this uses ~50% of my CPU. I'm also reading files on a NAS and outputting to NAA ... gotta love those Ethernet switches. Custom room treatments for headphone users. Link to comment
jabbr Posted May 15, 2015 Share Posted May 15, 2015 ] I would like to have the possibility to test NAA without the need to buy it, but in the case of NAA that doesn't seem to be possible. NAA could be a raspberry-pi type box (i.e. ARM) which can cost ~$35. It also could be a celeron or atom board. Does not need to be expensive. In any case it does require more infrastructure but allows the audio signal to be better separated from the CPU horsepower needed to upsample to DSD512 Custom room treatments for headphone users. Link to comment
jabbr Posted May 15, 2015 Share Posted May 15, 2015 Not at the moment, because DoP is needed on OS X. Hmm... this might answer several questions above: What about running NAA as a VMWare linux image? (OS X host, linux guest) Custom room treatments for headphone users. Link to comment
jabbr Posted May 15, 2015 Share Posted May 15, 2015 Just a friendly reminder that DoP is native DSD, too. It's not raw or direct but it is native (i.e never converted to PCM). Won't do DSD512... Custom room treatments for headphone users. Link to comment
jabbr Posted May 16, 2015 Share Posted May 16, 2015 No, due to its bulk transfer, for DSD Linux ALSA supports Amanero board through DoP only, therefore it is limited to DSD128 playback. So with DSC1, where Miska demonstrates the measurement results for DSD512: Preliminary measurement results for the DSC1 DAC - Blogs - Computer Audiophile , are you saying that DSD512 is not achievable with Linux (or Linux/NAA)? Custom room treatments for headphone users. Link to comment
jabbr Posted May 17, 2015 Share Posted May 17, 2015 These plants in the corners are home grown marijuana? Not even close.., Custom room treatments for headphone users. Link to comment
jabbr Posted May 17, 2015 Share Posted May 17, 2015 How would you be so sure? uuuhhh look at the photos ... and yes Back on topic, you can generally swap out motherboard. Windows may want you to reactivate by calling and just explain that motherboard died and had to be replaced and they will allow key to work... I would look at upping number of cores to 6 or more ... Custom room treatments for headphone users. Link to comment
jabbr Posted May 17, 2015 Share Posted May 17, 2015 I seem to recall Miska saying that with the AMD architecture and instruction set you get faster filter setup but slower computation for the realtime processing I was toying with the idea of an octocore with 2 cores assigned to NAA with realtime priority and 5 cores assigned to HQP with High Priority, leaving one core for household duties. However, I am now thinking HQP player needs the highest priority ie Realtime to get the SD Modulation timing right. I think we need to treat, nurture and respect the SD modulation like a Digital to Analog conversion step / process. After my chat with some guys from Meridian in the Hotel Lobby here in Munich last night about realtime PCM to DSD conversion, I am starting to think there is mileage to be had from a more stable clock on the motherboard . . . Sort of negates the idea of NAA if HQPlayer and networkaudiod running on same machine. It is good that Miska is able to measure and share plots including jitter for various levels of upsampling. Based on his data I'm unsure that this is generally a big problem. I have no idea of how proving the clock on my motherboard would affect SQ. regarding jitter in the DSDr/l lines themselves wouldn't the subsequent low pass filtering make this irrelevant? Custom room treatments for headphone users. Link to comment
Recommended Posts
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 accountSign in
Already have an account? Sign in here.
Sign In Now