MQN

[jrling]

v86 > v84. liitle bit better.

uploaded v86 normal no wt version, think I prefer it

yes. v86 no wt > v86. better agility.

bass control is great now. about the same as wasapi.
plus portaudio has additional ease + organic.

+1

Best taut bass I have heard. Lovely detail but also relaxing to listen to. Played all my test tracks all the way through. My new reference.

It may be me, but I did detect the slightest issue with timing on some tracks, as though the various performers on the stage were not coordinated.

Could be:
Me!
v86 no wt
The performers are being heard as they really were recorded

On one piano track I thought the notes were blurring into each other. Just a little but detectable. Perhaps that indicates a little overhang at the end of notes?

Anyone else found this?

[nige2000]

I still find it all a little blurry yet though it's improving all the time
No wt seems best

[sbgk]

I still find it all a little blurry yet though it's improving all the time
No wt seems best

uploaded v87 avx2, uses timestamps, seems to be better. p, didn't have them for some reason and little documentation on how the driver uses them, or rather pa had different ones which I don't understand. v87 rocks.

shall have to think about the control/white noise issue.

uploaded v88 avx2, better way of doing the timestamp.

[nige2000]

I still find it all a little blurry yet though it's improving all the time
No wt seems best

uploaded v87 avx2, uses timestamps, seems to be better. p, didn't have them for some reason and little documentation on how the driver uses them, or rather pa had different ones which I don't understand. v87 rocks.

shall have to think about the control/white noise issue.

yea its a little noiser too
pa is a mass of code
can it be reduced much in the hope of better efficiency?

i remember when you got mqnplay reduced to 7kb

[sbgk]

I still find it all a little blurry yet though it's improving all the time
No wt seems best

uploaded v87 avx2, uses timestamps, seems to be better. p, didn't have them for some reason and little documentation on how the driver uses them, or rather pa had different ones which I don't understand. v87 rocks.

shall have to think about the control/white noise issue.

yea its a little noiser too
pa is a mass of code
can it be reduced much in the hope of better efficiency?

i remember when you got mqnplay reduced to 7kb

only using a tiny part of pa, so should be able to junk it.

try v88

7kb, it's now 2kb.

[nige2000]

only using a tiny part of pa, so should be able to junk it.

cool is that far off yet?

try v88

ill try it in the morning along with v89:)

7kb, it's now 2kb.

i know but 7kb was the big drop with impact

[sbgk]

uploaded v89 avx2, think it's got the correct calculation for time increment

framesinbuffer*2*1/samplerate*1000000 to get buffer time in 100 nanosecond units

now got full range of atmospherics in whole lotta love and the bass/drums really kick in, which is nice.

saw robert plant on the one show and at glastonbury, sad. these old rockers need to know when to give up. think rock music requires a certain amount of energy and these old guys just don't have it. mind you saw bb king in his 80's and he still had it.

the time thing is funny because it doesn't fail if it's wrong, but sounds good when it's correct.

uploaded v89 normal for those in the cheap seats.

there is an avx setting I could try as well, but avx2 is definitely better than normal.

[nige2000]

Seen ac dc few years ago
They did pretty ok
But Yea the old guys need to bring it rather than failing and going through the motions for whatever reason as theres Some gigs you wish you didn't see or hear as it tarnishes the memories of used to be

Fair play to those guys that can:)

[sbgk]

The presentation time of the filter pin is specified as a KSTIME structure whose interpretation depends on the interface used. For the standard streaming interface, time is specified in 100-nanosecond increments (unless the numerator and denominator specify otherwise) representing the presentation position of the stream the filter is currently processing or is seeking to process. If this is a rendering filter, this position represents the data that is currently being rendered. This positioning information is synchronized with the master clock's presentation time. Presentation time typically begins at zero and may represent a time offset into file data. The numerator and denominator can be used to specify block alignment that the interface enforces.

http://msdn.microsoft.com/en-us/library ... s.85).aspx

that make sense to anybody ? J Keny ? I mean the synchronisation with master clock bit, I understand the rest.

[jkeny]

Interesting, Gordon - this might also explain the benefit that Nige gets when he changes the clocks on his mobo along with the PS - just more accurate & better timing of the stream?

Bas & lower midrange seem to be the frequencies where ITD (interaural Time Differences) is mostly used by our psychoacoustic brain processing so this might be what you are hearing with the better bass definition?

I know someone doing experiments with huge amounts of capacitance at the pins of DACs (& other devices) - amounts in farads at very low esr. His reports of the audible effects that he hears are a bass ambience that he never knew was there "The improvement in SQ is rather hard to describe in words - its lower colouration to the LF which might be described as improved 'weight' or 'authority'. Its quite surprising to me how much colouration has been removed which I was previously totally unaware of - I'd just adapted to the normal sound of electronics in the LF region."

It seems that we may be more sensitive to correct phase than was/is mainstream thinking?

Keep up the good work, Gordon - you're playing a blinder!