jrling Posted September 5, 2013 Share Posted September 5, 2013 I would like to try coding mqn for puppy Linux, but don't know if Linux is limited by alsa/jack, think windows may be moving ahead. I'm with Clive - this thread is about MQn and that means Windows. Let's keep this thread focussed. I have been seriously into MPDPup and I can tell you it comes nowhere near MQn. Clive - I would try Windows Server. There are a couple of guys on the JPlay forum who have compiled scripts to optimise it - one paid for and the other free. But the beauty of MQn and WS2012 is that opitmising really does not make so much difference to SQ which is testament to how good SBGK's code works. Cheers Jonathan Link to comment
sbgk Posted September 5, 2013 Author Share Posted September 5, 2013 I'm with Clive - this thread is about MQn and that means Windows. Let's keep this thread focussed.I have been seriously into MPDPup and I can tell you it comes nowhere near MQn. Clive - I would try Windows Server. There are a couple of guys on the JPlay forum who have compiled scripts to optimise it - one paid for and the other free. But the beauty of MQn and WS2012 is that opitmising really does not make so much difference to SQ which is testament to how good SBGK's code works. Cheers Jonathan well, I think with mqn I have tried to prove a number of things, chasing low latency is not necessary, high powered cpus are not required, more important to have everything aligned to memory, stripping down the code makes a difference, optimising the render loop is critical, optimising memcpy is most important of all, modern chips with 128 bit registers are an improvement and 256 bit should provide another step up. These techniques could be used in any player on any OS. Don't think anyone knows why different bit perfect streams could sound different, but it's interesting finding out what effects the sound. There is no harm in doubt and skepticism, for it is through these that new discoveries are made. Richard P Feynman http://mqnplayer.blogspot.co.uk/ Link to comment
sbgk Posted September 5, 2013 Author Share Posted September 5, 2013 just one version to try tonight, hopefully the final version. Can't believe I found 4 more new things to tweak in the assembly code. MQnplay.exe 2.26 sse4 intel. scarily good now, at least on my system. Thanks for all the testers who helped along the way, ended up with the 8 4 version. https://rapidshare.com/#users|45980080|0ae609ce616a35c8de7ac5fda4b6194c|11541 There is no harm in doubt and skepticism, for it is through these that new discoveries are made. Richard P Feynman http://mqnplayer.blogspot.co.uk/ Link to comment
jesuscheung Posted September 6, 2013 Share Posted September 6, 2013 possibly tune down the treble in 2.26? a bit harsh/bright. after 20mins of listening, my ears need a break. 2.26 has a new level of clarity and details! but does it reduce musicality? Link to comment
jesuscheung Posted September 6, 2013 Share Posted September 6, 2013 I have been seriously into MPDPup and I can tell you it comes nowhere near MQn. hi jrling, just wondering if MPDPup can customize IRQ or already have IRQ optimized for SQ? Link to comment
iori Posted September 6, 2013 Share Posted September 6, 2013 "that opitmising really does not make so much difference to SQ" - I share the same feeling with Jrling. So WS 2012 GUI is my current listening platform. The WS2012 core install does not come with audio support. The get/install-WindowsFeature can't give a direct help. Even with audio drivers installed, the device can't be detected. More effort is required to drill into this area though some gamers had succeed to enable audio in the Core environment. rdsu is correct, start with GUI if Core only installation is painful. Link to comment
Julf Posted September 6, 2013 Share Posted September 6, 2013 I would like to try coding mqn for puppy Linux, but don't know if Linux is limited by alsa/jack, think windows may be moving ahead. If you genuinely would like to know, I'd be happy to continue the discussion, but for now the conclusion seems to be to keep this thread restricted to talking about windows. Link to comment
jesuscheung Posted September 6, 2013 Share Posted September 6, 2013 ....But the beauty of MQn and WS2012 is that opitmising really does not make so much difference to SQ which is testament to how good SBGK's code works. i can confirm this one. i have hundreds of processes running- I/O is always busy, so is the CPU, but the newer MQn versions still perform amazingly- sweet with good clarity. The much older versions used to degrade in SQ quite badly. This just shows how efficiently MQn is able to utilize the CPU cycles. Link to comment
sbgk Posted September 6, 2013 Author Share Posted September 6, 2013 possibly tune down the treble in 2.26? a bit harsh/bright. after 20mins of listening, my ears need a break. 2.26 has a new level of clarity and details! but does it reduce musicality? 20 mins, no need to torture yourself. It is a bit strident, shouldn't do development late at night. There is no harm in doubt and skepticism, for it is through these that new discoveries are made. Richard P Feynman http://mqnplayer.blogspot.co.uk/ Link to comment
sbgk Posted September 6, 2013 Author Share Posted September 6, 2013 i can confirm this one. i have hundreds of processes running- I/O is always busy, so is the CPU, but the newer MQn versions still perform amazingly- sweet with good clarity. The much older versions used to degrade in SQ quite badly. This just shows how efficiently MQn is able to utilize the CPU cycles. those extra processes and I/O will be affecting the sound, maybe why sometimes you find it a bit bright. the processes might be on separate cores to mqn, but the irqs will still be spread across all cores unless you have changed their affinity. There is no harm in doubt and skepticism, for it is through these that new discoveries are made. Richard P Feynman http://mqnplayer.blogspot.co.uk/ Link to comment
jesuscheung Posted September 6, 2013 Share Posted September 6, 2013 those extra processes and I/O will be affecting the sound, maybe why sometimes you find it a bit bright. the processes might be on separate cores to mqn, but the irqs will still be spread across all cores unless you have changed their affinity. i usually restart the PC to a fresh state to confirm something is too bright. have most experience with 2.10, so if anything is brighter than 2.10, i call it bright, lol. have not touched the IRQ. Link to comment
jesuscheung Posted September 6, 2013 Share Posted September 6, 2013 i do suspect win2012 has brightness issues. i have a reliable source to confirm this. usually tuesday and wednesday the machine's SQ is at its top of its game. friday today is not a good day for SQ. whether is me imagining things, or coz the temperature/humidity, or coz the house has a stabler supply of power/current that day, i have no idea... Link to comment
jrling Posted September 6, 2013 Share Posted September 6, 2013 hi jrling, just wondering if MPDPup can customize IRQ or already have IRQ optimized for SQ? Hi - cannot remember but the MPDPup forum will tell you. Link to comment
sbgk Posted September 6, 2013 Author Share Posted September 6, 2013 hi jrling, just wondering if MPDPup can customize IRQ or already have IRQ optimized for SQ? think there is some basic irq tuning, look at the mpdpup forum There is no harm in doubt and skepticism, for it is through these that new discoveries are made. Richard P Feynman http://mqnplayer.blogspot.co.uk/ Link to comment
sbgk Posted September 6, 2013 Author Share Posted September 6, 2013 possibly tune down the treble in 2.26? a bit harsh/bright. after 20mins of listening, my ears need a break. 2.26 has a new level of clarity and details! but does it reduce musicality? 2.27 has the treble turned down, using mov register, register instead of mov register, 64. Also found another optimisation to allocate the temporary buffer once instead of each time gkmemcpy is called, you can hear/reach a bit further into the music now is the best way I can describe the effect. I would say this is the best version so far. There is no harm in doubt and skepticism, for it is through these that new discoveries are made. Richard P Feynman http://mqnplayer.blogspot.co.uk/ Link to comment
jesuscheung Posted September 7, 2013 Share Posted September 7, 2013 wow. 2.27 is all-round better than all previous versions. i want to bitch about something. i can't. Link to comment
jesuscheung Posted September 7, 2013 Share Posted September 7, 2013 no. wait! i think 2.27's stage is not as fun as for example 2.10? with 2.10 and some other versions, instruments can be all over the place in a fun/musical way. 2.27's stage can sound flat. and 2.27 has gotten a bit cold. 2.24 8 4 is relatively warm and sweet. Link to comment
sbgk Posted September 7, 2013 Author Share Posted September 7, 2013 no. wait! i think 2.27's stage is not as fun as for example 2.10? with 2.10 and some other versions, instruments can be all over the place in a fun/musical way. 2.27's stage can sound flat. and 2.27 has gotten a bit cold. 2.24 8 4 is relatively warm and sweet. 2.27 is such a fantastic step forward in SQ it is difficult to compare against previous versions, I'm just revelling in the detail. 2.24 had the zp4 setting which is not really applicable to MQn and made the sound closed in, I thought . 2.27 is like having your own private concert whether it is classical, jazz or other. Shall listen a bit more before deciding whether there are any issues. Live music is not always a warm sweet experience and the standard of some recordings has much to be desired, I have a collection of hard digital sounding recordings that I use to test MQn with. There is no harm in doubt and skepticism, for it is through these that new discoveries are made. Richard P Feynman http://mqnplayer.blogspot.co.uk/ Link to comment
jesuscheung Posted September 7, 2013 Share Posted September 7, 2013 2.27 is such a fantastic step forward in SQ it is difficult to compare against previous versions... you are right. i reboot into my win7 which is totally stripped-down for music only. 2.27 sounds pretty perfect there. sound is little colder (than previous versions) still very sweet. stage is good. clarity and detail are greatest so far. i have to figure out why my win2012 sounds cold and flat with 2.27. coz i did directly compare with other versions. Link to comment
jesuscheung Posted September 7, 2013 Share Posted September 7, 2013 figured it out! GPU unplugged + 2012 -> 2.27 is sweet. 2.10 is sweet. GPU plugged in + 2012 -> 2.27 is cold. 2.10 is sweet. GPU plugged in or not + win7 -> 2.27 sounds fine. GPU does share IRQ with DAC. pretty sure i prefer the stage of older versions. Link to comment
jesuscheung Posted September 7, 2013 Share Posted September 7, 2013 hi sbgk, would you mind telling me your ram frequency and timing? you can check it with cpuz. am trying to tune the bios for MQn. my bios was previously tuned for softer sounding XA, which has some negative effects on MQn. Link to comment
sbgk Posted September 7, 2013 Author Share Posted September 7, 2013 figured it out! GPU unplugged + 2012 -> 2.27 is sweet. 2.10 is sweet. GPU plugged in + 2012 -> 2.27 is cold. 2.10 is sweet. GPU plugged in or not + win7 -> 2.27 sounds fine. GPU does share IRQ with DAC. pretty sure i prefer the stage of older versions. there were 4 settings different between 2.24 and 2.27, agree 2.27 is hard sounding, so it's a matter of finding the offending setting. There is no harm in doubt and skepticism, for it is through these that new discoveries are made. Richard P Feynman http://mqnplayer.blogspot.co.uk/ Link to comment
jesuscheung Posted September 7, 2013 Share Posted September 7, 2013 still trying to figure out why i love 2.10 so much. if i just focus on the vocal, i can feel it is a 100% sound reproduction of the vocal, all emotions are there, no added color, no blurred-out, vocal can have more depth than new versions. i think that's why i keep referencing to 2.10. what clock rate of pro audio you use in the registry? i start messing with it and 2.27 is starting to come alive, especially the stage. because i think you said you use default value 10000 when you implemented 2.10. Link to comment
sbgk Posted September 7, 2013 Author Share Posted September 7, 2013 hi sbgk, would you mind telling me your ram frequency and timing? you can check it with cpuz. am trying to tune the bios for MQn. my bios was previously tuned for softer sounding XA, which has some negative effects on MQn. cpu 931 and ram 532, 7 7 7 20 that is out of the box as can't tune. the cpu frequency is set by setting the % min/max frequency in high performance power plan to 0. Surprised you say win 7 sounds ok, I always thought it didn't sound good There is no harm in doubt and skepticism, for it is through these that new discoveries are made. Richard P Feynman http://mqnplayer.blogspot.co.uk/ Link to comment
sbgk Posted September 7, 2013 Author Share Posted September 7, 2013 still trying to figure out why i love 2.10 so much. if i just focus on the vocal, i can feel it is a 100% sound reproduction of the vocal, all emotions are there, no added color, no blurred-out, vocal can have more depth than new versions. i think that's why i keep referencing to 2.10. what clock rate of pro audio you use in the registry? i start messing with it and 2.27 is starting to come alive, especially the stage. because i think you said you use default value 10000 when you implemented 2.10. 23220 There is no harm in doubt and skepticism, for it is through these that new discoveries are made. Richard P Feynman http://mqnplayer.blogspot.co.uk/ 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