Jump to content
IGNORED

Visual studio 2012 c++ and wasapi minimalist player


sbgk

Recommended Posts

2.49 no buff have bass micro-vibration is a bit better than rax, and also drum/beat/knock sound bouncy better (even though it is non-musical/dumb sound, but it can also vibrate). very difficult to detect.

what we now can do more, jesuscheung.

sbgk, could you make me and people 1.xx or 3.xx versions similar to 2.49 no buff, for onboard soundcard, HDMI. with 2.xx i just can test only, not for listening. on my laptop mqnplay 2.xx sound have click/pops.., 2.xx not compatible. thanks.

 

after these versions I need to do some more experimentation to find out what is going on, am thinking I don't even need to read from the source in the first loop, I can prefetch straight into the L2 cache and then stream it out in the second loop, also found I could be using a better prefetch command.

 

Still don't understand why 2 loops sounds better than 1, 1 has fewer instructions, the data is in the cache and should sound better in theory, but 2 loops gives much better timing, might be the noise associated with reading and writing at the same time in the 1 loop version.

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
just reporting a bug. when i execute MQn.bat at the same time i browse with chrome, my extension mouse gesture can crash. don't always happen. just occasionally. also MQn will stuck with no sound.

 

Think I said a couple of posts back that the assembly code has been minimised to such an extent that it may get affected by other processes, hopefully this will be fixed when/if I can put the whole render loop in assembly, it hasn't caused any issues apart from MQn not starting or shutting down when there has been other activity, although I have had one occasion where the machine shut down, not sure if that was related.

If I can't fix it, I'll make a safe version as well, it does affect the SQ a bit.

 

What's supposed to happen is that 7 XMM registers will have their state saved and restored before/after each memcpy call, I have removed this and thus if an outside process uses one of these registers then there will be a problem. If the render loop was in assembly then the save/restore can take place without affecting the SQ as it would only happen once for each play instead of for every memcpy call. I know it's not what you're supposed to do, but it's an experiment.

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

very busy here today, need to be up early in the morning so i dont miss all this excitement

 

tried the few earlier liked the rax, tried no buff just now and it is exceptional

sse2 has the least noise but that was it for me

2.49 no buff is thee best theres ever been

 

Hi res for 24 in 24?

i know im a pita

Link to comment
after these versions I need to do some more experimentation to find out what is going on, am thinking I don't even need to read from the source in the first loop, I can prefetch straight into the L2 cache and then stream it out in the second loop, also found I could be using a better prefetch command.

 

Still don't understand why 2 loops sounds better than 1, 1 has fewer instructions, the data is in the cache and should sound better in theory, but 2 loops gives much better timing, might be the noise associated with reading and writing at the same time in the 1 loop version.

 

alright. after some careful listening, i think that 2.49 all of them can improve on the stage. coz i can still find many cases that 1.2 is even more engaging than 2.49. coz i think 2.49 needs a little stronger main music or a little weaker background music, not sure which.

 

other aspects regarding eax, rax and no buffer, they have many small differences, it is difficult to decide which version is best.

 

-eax has a little more 3D.

-no buf has mellower tunes, more emotions, but i think something is lost.

-rax is more balance?

Link to comment
alright. after some careful listening, i think that 2.49 all of them can improve on the stage. coz i can still find many cases that 1.2 is even more engaging than 2.49. coz i think 2.49 needs a little stronger main music or a little weaker background music, not sure which.

 

other aspects regarding eax, rax and no buffer, they have many small differences, it is difficult to decide which version is best.

 

-eax has a little more 3D.

-no buf has mellower tunes, more emotions, but i think something is lost.

-rax is more balance?

 

think the 1.2 version sounds ok, but slightly sibilant as Nige said not in the same league as 2.48/9.

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
Hi sgbk:

'Found an installation in one of mine computer from 03.27.13 with Mahler - Simphonie 5 - Adagietto.

That bass!!

I second lekt´s request for 1.49 no buff version.

Some computers can´t play 2.xx. well

Possibly the"1" matters??

 

Thanks you very much in advance

 

Juan

 

wondered why people weren't asking for the different versions.

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
alright. after some careful listening, i think that 2.49 all of them can improve on the stage. coz i can still find many cases that 1.2 is even more engaging than 2.49. coz i think 2.49 needs a little stronger main music or a little weaker background music, not sure which.

 

other aspects regarding eax, rax and no buffer, they have many small differences, it is difficult to decide which version is best.

 

-eax has a little more 3D.

-no buf has mellower tunes, more emotions, but i think something is lost.

-rax is more balance?

 

To quote Van Morisson -

 

"Enlightenment says the world is nothing

Nothing but a dream, everything's an illusion

And nothing is real.

 

you're just listening to the cpu do it's stuff - it's not music

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
i also have a PCIe souncard, i can use 2.x.

 

readme

https://docs.google.com/file/d/0B3vvH5WBfg8PWHdTSXY2eHJzcXc/edit

 

Thanks JC & SBGK for reminding me where I read it. I had previously discarded the info as it appears 1 & 3 releases wernt being actively updated.

 

Does WASAPI have fixed buffer size?

 

2.4* works excellent here.. as does 1.2 - I thought I'd better ask.

 

2.49 rax: still listening... sounds technically very good, but seems a little quieter and distant after what WoWed me yesterday afternoon...

 

.. but unable to repeat that magic headphone holographic experience of 2.48 ntdqdqnt r8 from a direct mqncontrol call I experienced yesterday afternoon. Maybe it was my neighbour pulling her hairdryer from the supply?

Link to comment
i also have a PCIe souncard, i can use 2.x.

 

readme

https://docs.google.com/file/d/0B3vvH5WBfg8PWHdTSXY2eHJzcXc/edit

 

Thanks JC & SBGK for reminding me where I read it. I had previously discarded the info as it appears 1 & 3 releases wernt being actively updated.

 

Does WASAPI have fixed buffer size?

 

2.4* works excellent here.. as does 1.2 - I thought I'd better ask.

 

2.49 rax: still listening... sounds technically very good, but seems a little quieter and distant after what WoWed me yesterday afternoon...

 

.. but unable to repeat that magic headphone holographic experience of 2.48 ntdqdqnt r8 from a direct mqncontrol call I experienced yesterday afternoon. Maybe it was my neighbour pulling her hairdryer from the supply?

 

Holophony is the audio equivalent of holography, didn't know that

 

That quiet sound is the lack of noise.

 

MQn is fixed and uses 8192 Byte buffer size = 2048 samples for 16 bit and 1024 samples for 24. Wasapi is not fixed - smallest device setting seems to be 3ms, default is 10ms. MQn for example use around 40ms for 16/44.1 and reduces as the resolution goes up, it has to be 128 bit aligned.

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

Hi again

 

Downloaded the SSE 4 versions of 2.49. Listened to a couple of them and they are really very good. I cannot get 24bit files to play at all, yet they play fine in other players. Not an issue - I'd forego hi-res with 16bit sounding this good.

 

I tried to run the Wasapi test but it starts then closes very quickly.

 

I want to keep playing more music now, but work is getting in the way!

 

Thanks sbgk - brilliant stuff.

Link to comment
Hi again

 

Downloaded the SSE 4 versions of 2.49. Listened to a couple of them and they are really very good. I cannot get 24bit files to play at all, yet they play fine in other players...

 

i can play 24/44100, but not 96k or above. 16bits already sounds like vinyl. i cannot imagine what actual 96k files sound like. heaven.

Link to comment

Tested 2.49 sse2 atom. Big step forward compared to 2.48.

For the first time, I felt like listening to my old Linn LP12 - Naim system.

Just listening to Free's Fire and Water album, Paul Kossoff's guitar never sounded so sweet, stunning bass and vocals. A private rock concert in my room.

And all that with flac and Foobar interface.

I have already bought the parts for my next PC audio player (S1200kpr mobo and Xeon e3-1265L V2) but now I wonder if this will actually be an improvement...

Link to comment
alright. after some careful listening, i think that 2.49 all of them can improve on the stage. coz i can still find many cases that 1.2 is even more engaging than 2.49. coz i think 2.49 needs a little stronger main music or a little weaker background music, not sure which.

 

other aspects regarding eax, rax and no buffer, they have many small differences, it is difficult to decide which version is best.

 

-eax has a little more 3D.

-no buf has mellower tunes, more emotions, but i think something is lost.

-rax is more balance?

 

do you want to make crystal to diamond!

i'm very happy today, jesuscheung.

 

sse2 intel, sse2 atom, sse4 eax, sse4 rax, no buff

all these 5 versions have soundstage/wight a bit less than v1.2, 2.79, JEP, JPlay. but just a bit of bit. if it is forced expand by bad method may lose something precious, e.g vocal focus. very dangerous, be careful. there everything is perfect: drum, bass, guitar, keyboard, vocal, treble... and piano sound in 2.49 no buff, no buff give best vibration, micro-micro. it make edges of all sounds be come very good, even drum sound.

no buff stage is narrowest, but vibration is best. i don't know why.

Link to comment

.. but unable to repeat that magic headphone holographic experience of 2.48 ntdqdqnt r8 from a direct mqncontrol call I experienced yesterday afternoon. Maybe it was my neighbour pulling her hairdryer from the supply?

 

I can repeat this. It happens when the music is replayed using direct mqncontrol call. Somehow more vivid than 1st play - and damn lot more distracting from work.

Also I note music(double CD) starts much quicker than 1st time.

OS and Music on single USB2 stick.

Any ideas?

Link to comment
Whats you pc setup goon ?

just trying to gather up different ideas

 

i5-2500T, 8Gb, umpteen USB sticks with different shades of Server2012 & optimisation to suit different players.

RME HDSPe soundcard locked to EmmLabs DCC-2SE.

Switch mode PS - umpteen Maplin PSs waiting for me to come out trance.

What have I missed?

Link to comment
do you want to make crystal to diamond!

i'm very happy today, jesuscheung.

 

sse2 intel, sse2 atom, sse4 eax, sse4 rax, no buff

all these 5 versions have soundstage/wight a bit less than v1.2, 2.79, JEP, JPlay. but just a bit of bit. if it is forced expand by bad method may lose something precious, e.g vocal focus. very dangerous, be careful. there everything is perfect: drum, bass, guitar, keyboard, vocal, treble... and piano sound in 2.49 no buff, no buff give best vibration, micro-micro. it make edges of all sounds be come very good, even drum sound.

no buff stage is narrowest, but vibration is best. i don't know why.

 

"2 loops gives much better timing"

 

maybe 3 loops you be happier.

 

i like all 3 rax, no buff, eax. i just feel that 'vocal focus' is about 1% off from perfection due to the stage. yes, no buff is great as you say. something bothers me, think that might be muddiness, not sure. no buf is my new favorite for vocal, but it does lacks some vocal details.

Link to comment
I can repeat this. It happens when the music is replayed using direct mqncontrol call. Somehow more vivid than 1st play - and damn lot more distracting from work.

Also I note music(double CD) starts much quicker than 1st time.

OS and Music on single USB2 stick.

Any ideas?

Is it possible to call mqncontrol that just loads music into memory, but does not run mqnplay?

Does this question make any sense?

Link to comment
Is it possible to call mqncontrol that just loads music into memory, but does not run mqnplay?

Does this question make any sense?

 

why would you do that ?

 

The reason it is quicker is that the data is in cache after the first play, so if you want to impress someone then load the albums into cache first, the only way at the moment is to stop playing after the data is loaded.

 

For hirez it takes the same time because it needs to convert to 24 in 32, I suppose that's one advantage of 24 in 24.

 

When you say starting MQncontrol directly are you typing it into a dos prompt, not sure what you mean.

 

There is a utility called MQnplay.bat which allow different versions of MQn to be played, a good trick is to stick the same version on twice and see if you can spot that it is the same version, very difficult because the mind wanders etc, but over time it is possible to identify different versions.

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
"2 loops gives much better timing"

 

maybe 3 loops you be happier.

 

i like all 3 rax, no buff, eax. i just feel that 'vocal focus' is about 1% off from perfection due to the stage. yes, no buff is great as you say. something bothers me, think that might be muddiness, not sure. no buf is my new favorite for vocal, but it does lacks some vocal details.

 

i suspect that 2.22_atom vocal is better than no buff vocal. i said about its bass, is special bass vibration. and now i heard it in 2.49 rax and no buff. i will test for compare vocal of 2.22_atom & no buff. interesting.

 

you say "no buf is my new favorite for vocal, but it does lacks some vocal details". when the vibration reaches the threshold of speakers, instability would occur. maybe too low frequency bass affect to vocal? reduce bass by tuning ampliffer.

Link to comment

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 account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...