uploaded control v50 and counter v15, use with play v134, play stops at the end now and sounds better with counter back in use.
noticed a mistake in the load assembly code, uploaded mqnloader v30 and mqncontrol v51, surprising lift in detail and definition.
am now working on an sse2 version, but would urge people to upgrade as it makes a big difference.
MQN
Re: MQN
sbgk wrote:uploaded control v50 and counter v15, use with play v134, play stops at the end now and sounds better with counter back in use.
noticed a mistake in the load assembly code, uploaded mqnloader v30 and mqncontrol v51, surprising lift in detail and definition.
am now working on an sse2 version, but would urge people to upgrade as it makes a big difference.
Further improvement Gordon
Mqn counter stops when track finish but cpu and ram usage still climbing up
Need to be stop manually to prevent pc from locking
WMI wasn't running that is why I had that problem
It is getting better and better
How far this can go??
Thank you
Last edited by elaprince on Sat Nov 07, 2015 9:34 pm, edited 1 time in total.
Re: MQN
With mqnplay v134, control v51, counter v15 and loader v30 music play fine, and
at end of play all these processes quickly close down by themself. I see no rise in Cpu for mqnplay (just 3% at closedown).
SQ - I have not listen much yet, but it sounds really good, a clear improvement over 134/49.
sbgk - thank you so very much.
Cheers
Jan H.
at end of play all these processes quickly close down by themself. I see no rise in Cpu for mqnplay (just 3% at closedown).
SQ - I have not listen much yet, but it sounds really good, a clear improvement over 134/49.
sbgk - thank you so very much.
Cheers
Jan H.
Gigabyte H97M-D3H with PPA OCXO module. i7-4790T, 800MHz. 8GB Ram, 800MHz.
PPA 2-rail LPSU & Pico. JCAT battery for OS-SSD and PPA v3 USBcard.
Server 2012 R2, AO 1.40. APL HiFi DAC-S, upd. Only use 1644 .wav
PPA 2-rail LPSU & Pico. JCAT battery for OS-SSD and PPA v3 USBcard.
Server 2012 R2, AO 1.40. APL HiFi DAC-S, upd. Only use 1644 .wav
Re: MQN
uploaded 10 ms versions play v135 and counter v16. Think 10ms seems to have the best balance and seems glitch free.
while there are only wait and io instructions in the render loop there are also instructions that don't do anything that are used for alignment, sometimes it is better to use a longer instruction rather than a shorter one with the padding instructions.
uploaded play v136 which does this by changing a 7 byte pad instruction to a 2 byte one by using a longer instruction, seems to make a difference.
And that's about as good as you can get. Counter in assembly will improve things further.
The whole digital music implementation is flawed because it doesn't take the noise generated by electrical interactions into account, in fact those most knowledgeable take great pride in denying any effect.
Should the wav file format be reversed, in that currently the most dynamic sound requires the most data to be transferred, generating the most noise. dsd get's round this to an extent, but sounds digital to me.
MS supposed low latency kernel streaming is flawed in that it requires kernel mode/user mode interaction for every buffer, which they admit has a big effect on latency ie SQ.
So we are stuck until someone integrates the player into the kernel, could do it in linux as I've said before. MS is a no.
while there are only wait and io instructions in the render loop there are also instructions that don't do anything that are used for alignment, sometimes it is better to use a longer instruction rather than a shorter one with the padding instructions.
uploaded play v136 which does this by changing a 7 byte pad instruction to a 2 byte one by using a longer instruction, seems to make a difference.
And that's about as good as you can get. Counter in assembly will improve things further.
The whole digital music implementation is flawed because it doesn't take the noise generated by electrical interactions into account, in fact those most knowledgeable take great pride in denying any effect.
Should the wav file format be reversed, in that currently the most dynamic sound requires the most data to be transferred, generating the most noise. dsd get's round this to an extent, but sounds digital to me.
MS supposed low latency kernel streaming is flawed in that it requires kernel mode/user mode interaction for every buffer, which they admit has a big effect on latency ie SQ.
So we are stuck until someone integrates the player into the kernel, could do it in linux as I've said before. MS is a no.
Last edited by sbgk on Sun Nov 08, 2015 12:23 am, edited 1 time in total.
Re: MQN
uploaded optimised counter v16 with code in assembly, works on win 10, if someone can test it on win 8.1 or win 2012 r2.
Anyway that's it for sq changes for now. Shall do a non haswell version when I have time.
If someone has heard a better player then let me know, all I know is it makes my Van Halen tracks recorded in the 80's sound like some sort of Mozart symphony with the amount of detail that's being retrieved, remarkable.
Anyway that's it for sq changes for now. Shall do a non haswell version when I have time.
If someone has heard a better player then let me know, all I know is it makes my Van Halen tracks recorded in the 80's sound like some sort of Mozart symphony with the amount of detail that's being retrieved, remarkable.
-
lukivision
- Posts: 27
- Joined: Sat May 10, 2014 10:10 pm
Re: MQN
Gordon, if you want to make the next big step: go to linux. I`d love it.
Luki ;-:
Luki ;-:
Re: MQN
sbgk wrote:uploaded optimised counter v16 with code in assembly, works on win 10, if someone can test it on win 8.1 or win 2012 r2.
Anyway that's it for sq changes for now. Shall do a non haswell version when I have time.
If someone has heard a better player then let me know, all I know is it makes my Van Halen tracks recorded in the 80's sound like some sort of Mozart symphony with the amount of detail that's being retrieved, remarkable.
Counter v17 works beautifully on win 2012r2
HDPLEX;picoPSU;ASUS Q87M;i7-4770T;PH SR7EHD;Server2012R2;Thesycon 2.24;
JCAT USB;Sonicweld DiverterHR2;Naim DC1;Chord Hugo;Morrow Audio MA6;Naim NAC-282,SuperCapDR;NAP-300;
AQ Cinnamon;GISO GB;Netgear Pro+XM21X;Cisco SG300;NAS-ZFS.
JCAT USB;Sonicweld DiverterHR2;Naim DC1;Chord Hugo;Morrow Audio MA6;Naim NAC-282,SuperCapDR;NAP-300;
AQ Cinnamon;GISO GB;Netgear Pro+XM21X;Cisco SG300;NAS-ZFS.
Re: MQN
Gordon everything working perfect no glitches
Sound superb on my setup hirez would be a bonus
Thank you
Someone mention Linux few post back
There is a possibility that well design and modified unit based on Linux with kernel player (I think you already know that Gordon) will outperform any pc based music setup
I have Aries (streamer) in mind
With right knowledge it could be turn into player/streamer that no pc will touch IMHO
Sound superb on my setup hirez would be a bonus
Thank you
Someone mention Linux few post back
There is a possibility that well design and modified unit based on Linux with kernel player (I think you already know that Gordon) will outperform any pc based music setup
I have Aries (streamer) in mind
With right knowledge it could be turn into player/streamer that no pc will touch IMHO