Tír Na HiFi

http://www.thewindowsclub.com/processor ... indows-7-8

is this how I change the value?

I just upgraded to Win 8.1, and I downloaded 8.1-R2 version.
All I can say is, WoooW!

wushuliu wrote:

jesuscheung wrote:sbgk, back then probably when it was MQn 2.3x, you asked the value for Win32PrioritySeparation. back then, i said 18 was best because it was the most revealing.

now MQn 2.5x, 2.6x has gotten so much more revealing, 18 starts to be inappropriate. i now believe 28 is best. 18 was forcing the details. 18 was good because MQn was lacking details relative to now.

18 losses some layers of bass and adds earaches and losses some musicality.

in a high digital jitter environment, meaning, OS is un-tuned, 18 is better. in a highly tuned OS+2.6x/2.5x MQn, 28 is better.

of course there is 14 recommended by jplay. all it does is make stage sounds 'better' in exchange for musicality and many details. i dislike it.

Hi, I have just installed MQN on an AMD PC. I've been using Jplay for a longtime and have to agree - after changing Win32 Separation to 28 MQN soundsMUCH better. Very obvious change in presentation. MQN is amazing!

which version ?

erin wrote:http://www.thewindowsclub.com/processor ... indows-7-8

is this how I change the value?

yes, via registry

wushuliu wrote:

jesuscheung wrote:sbgk, back then probably when it was MQn 2.3x, you asked the value for Win32PrioritySeparation. back then, i said 18 was best because it was the most revealing.

now MQn 2.5x, 2.6x has gotten so much more revealing, 18 starts to be inappropriate. i now believe 28 is best. 18 was forcing the details. 18 was good because MQn was lacking details relative to now.

18 losses some layers of bass and adds earaches and losses some musicality.

in a high digital jitter environment, meaning, OS is un-tuned, 18 is better. in a highly tuned OS+2.6x/2.5x MQn, 28 is better.

of course there is 14 recommended by jplay. all it does is make stage sounds 'better' in exchange for musicality and many details. i dislike it.

Hi, I have just installed MQN on an AMD PC. I've been using Jplay for a longtime and have to agree - after changing Win32 Separation to 28 MQN soundsMUCH better. Very obvious change in presentation. MQN is amazing!

been playing around with clockrate + Win32PrioritySeparation + timer resolution all week. i have finalized the best combination for MQn, XA and lekt under 44100. the answer is:

clockrate = 5805 or 5804 or 5806 (only one is correct. if all 3 doesn't work, this is not for you)
Win32PrioritySeparation = 28
timer resolution = 1 or 2

1 clockrate rules all!

might not work for everyone. DAC may or may not be a variable too. not sure.

this is not the most 'hard working setting'. the hardest working settings would have a 6 digits clockrate that is too accurate, with Win32PrioritySeparation=18 to tell OS to try harder, and timer resolution=0.5 to tell the OS to try even harder. my machine can't handle such 'pressure', SQ is not musical using the most 'pushing' settings.

jesuscheung wrote:

wushuliu wrote:

jesuscheung wrote:sbgk, back then probably when it was MQn 2.3x, you asked the value for Win32PrioritySeparation. back then, i said 18 was best because it was the most revealing.

now MQn 2.5x, 2.6x has gotten so much more revealing, 18 starts to be inappropriate. i now believe 28 is best. 18 was forcing the details. 18 was good because MQn was lacking details relative to now.

18 losses some layers of bass and adds earaches and losses some musicality.

in a high digital jitter environment, meaning, OS is un-tuned, 18 is better. in a highly tuned OS+2.6x/2.5x MQn, 28 is better.

of course there is 14 recommended by jplay. all it does is make stage sounds 'better' in exchange for musicality and many details. i dislike it.

Hi, I have just installed MQN on an AMD PC. I've been using Jplay for a longtime and have to agree - after changing Win32 Separation to 28 MQN soundsMUCH better. Very obvious change in presentation. MQN is amazing!

been playing around with clockrate + Win32PrioritySeparation + timer resolution all week. i have finalized the best combination for MQn, XA and lekt under 44100. the answer is:

clockrate = 5805
Win32PrioritySeparation = 28
timer resolution = 1 or 2

1 clockrate rules all!

might not work for everyone. DAC may or may not be a variable too. not sure.

this is not the most 'hard working setting'. the hardest working settings would have a 6 digits clockrate that is too accurate, with Win32PrioritySeparation=18 to tell OS to try harder, and timer resolution=0.5 to tell the OS to try even harder. my machine can't handle such 'pressure', SQ is not musical using the most 'pushing' settings.

18/46440 for me, what is timer resolution ?

have uploaded an 8 no cat and 8 other version, they both have the raw setting. think 8 other sounds best - sort of makes sense as it says to use this one for capture and loopback apps ie presume it gets max priority or throughput. Thought the original 8.1-r2 was a bit anemic

What is 'XA'? and where do I change Clock Rate?

jesuscheung wrote:this is not the most 'hard working setting'. the hardest working settings would have a 6 digits clockrate that is too accurate, with Win32PrioritySeparation=18 to tell OS to try harder, and timer resolution=0.5 to tell the OS to try even harder. my machine can't handle such 'pressure', SQ is not musical using the most 'pushing' settings.

win 8 has a tickless kernel and the timer resolution can be set just for one process without affecting the whole system, shall investigate if it makes a difference for MQn. There is evidence that setting it to 0.5 or 1 etc actually makes the system run slower as the timer irq is firing all the time.

sbgk wrote:

jesuscheung wrote:this is not the most 'hard working setting'. the hardest working settings would have a 6 digits clockrate that is too accurate, with Win32PrioritySeparation=18 to tell OS to try harder, and timer resolution=0.5 to tell the OS to try even harder. my machine can't handle such 'pressure', SQ is not musical using the most 'pushing' settings.

win 8 has a tickless kernel and the timer resolution can be set just for one process without affecting the whole system, shall investigate if it makes a difference for MQn. There is evidence that setting it to 0.5 or 1 etc actually makes the system run slower as the timer irq is firing all the time.

yes 0.5 can be bad. can be too aggressive. it can crash my mySQL server when flooded with connections at the same time.

0.5 can be good when OS is doing really nothing. like a music server.

recently, i realize that timer resolution is related to buffer size.

for example, buffer = 640 samples = 14.5 msec. if timer resolution = 0.5msec, then 14.5 is divisible by 0.5, hence, theoretically good for SQ.

in the same way, buffer= 352 samples = 8 msec. using ws2012, timer resolution is default at 1 or 2 msec. again, this buffer is good for SQ.

i have clock rate=5805 because it seems to sound good for both MQn's buffer size and XA's buffer size and also lekt's.

256/44100 = 5804988. best rounding is 5805. i used to have 58050 thinking 5 digits increase accuracy. since the rounding is so good, 5805 seems better.

256*2*2=1024=MQn's buffer size. so, 5805 is usable according to my ears.

5805 is usable for 640 samples too because 640*4/44100=5805

how do you change the timer resolution ?