Tír Na HiFi

There is no server 10 yet
Win10 it is not better then server 2012 r2 only cheaper
So people who dont want to pay high price for server have a choice to buy something cheaper
New win server will be released in 2016
Thanks

m5lig wrote:Would someone be kind enough to post how the latest MQN files should be loaded / modified /configured to get a system going from scratch?

I'm back in an 8 month technological time warp , and parked my system in the "old" 4 file, download and rename , MQN system.

I now read of mqnload , hexadecimal strings , identification of DACs and have to admit to not being clear on where to start .

get device name using mqndevinfo.exe

copy and paste it into mqnparam.txt

download latest copies of mqnplay, mqnloader and mqncontrol and files mentioned below.

put mqnparam.txt, paste.exe, mqnplay.exe, mqnloader.exe, mqncontrol.exe and mqn.bat into a folder eg c:\mqnplayer (default is c:\musicplayer)

edit mqn.bat so that it points to c:\mqnplayer ie change cd c:\musicplayer line near top of mqn.bat

copy required wav file to the clipboard ie ctrl-c and double click on mqn.bat

I have set my exe's to be run as administrator and created a shortcut for mqn.bat and made it run as administrator, this then allows the kill mqnplay command to work in mqn.bat

there is a haswell and an sse version of the exe's and a win 10 and win 2012 r2/win 8.1 version of the haswell exe's

elaprince wrote:There is no server 10 yet
Win10 it is not better then server 2012 r2 only cheaper
So people who dont want to pay high price for server have a choice to buy something cheaper
New win server will be released in 2016
Thanks

someone on CA was saying he'd created a great machine based on win 8 embedded, first I'd heard of someone doing that.

So after all that the question is is there a correlation between the no of instructions and the sq ?

what do we know

wasapi sounds different to ks
minimalist c program sounds different to c++
assembly sounds different to c++
assembly instructions sound different to each other
calling direct to ntdll sounds different to kernel32
alignment of instructions affects the sq
order of instructions affects the sq
priority of program affects sq
number of instructions between wait and deviceio call affects sq

What I was trying to get at is is v48 the best, it has all the above optimised, so has a good chance of being so. What raised the question is that versions with seemingly minor changes to the assembly code could still sound like a dog.

The annoying thing is that the code still affects sq to the extent it does. I still maintain the best you can get is going to be a linux os with optimised kernel that reads the data from ram without any user app interference, this will require the driver to be modified as well. Not a small task, why has no one tried it ? Are they all deaf ?

The other way is a hardware fpga method, but they seem to have their own issues of noise.

There's also the issue of what is going on that the way the data is loaded into ram affects sq, that is a question that should be fundamental to anyone trying to get good sq out, but only seems to have been taken up by a few hobbyists.

sbgk wrote:someone on CA was saying he'd created a great machine based on win 8 embedded, first I'd heard of someone doing that.

I've used Windows 7 Embedded when it came out. Was good, small and I liked it. Got a fully working installation with around 900MB, booting extremely fast from mSATA.

When Windows Embedded Standard 8 came out I'd tried that version as well. But MS changed the possibilies of modularization. Now the smallest possible installation was around 3GB. At the same time the community started with testing Windows Server. And I decided to use the latter, as it seemed to me that its core mode is even more compact then the smallest Embedded installation. Additionally there were optimization tools for Windows Server, but not for Windows Embedded Standard.
Plus, Server with AO sounded better to me.

This was more than two years ago. Since then never had an eye on it again.

Christoph

sbgk wrote: There's also the issue of what is going on that the way the data is loaded into ram affects sq, that is a question that should be fundamental to anyone trying to get good sq out, but only seems to have been taken up by a few hobbyists.

Exactly, I have found mqnloader to be making big difference. I believe the a lot of difference between LDN15 and new versions comes from MQNloader v7 actually. At least my tests seems to be pointing to MQNloader as the big factor.

sbgk wrote:So after all that the question is is there a correlation between the no of instructions and the sq ?

what do we know

wasapi sounds different to ks
minimalist c program sounds different to c++
assembly sounds different to c++
assembly instructions sound different to each other
calling direct to ntdll sounds different to kernel32
alignment of instructions affects the sq
order of instructions affects the sq
priority of program affects sq
number of instructions between wait and deviceio call affects sq

What I was trying to get at is is v48 the best, it has all the above optimised, so has a good chance of being so. What raised the question is that versions with seemingly minor changes to the assembly code could still sound like a dog.

The annoying thing is that the code still affects sq to the extent it does. I still maintain the best you can get is going to be a linux os with optimised kernel that reads the data from ram without any user app interference, this will require the driver to be modified as well. Not a small task, why has no one tried it ? Are they all deaf ?

The other way is a hardware fpga method, but they seem to have their own issues of noise.

There's also the issue of what is going on that the way the data is loaded into ram affects sq, that is a question that should be fundamental to anyone trying to get good sq out, but only seems to have been taken up by a few hobbyists.

Well the grass always looks greener on the other side, but sofar nobody has ever produced even a decent sounding Linux audioplayer, and MQn tops everything else out there.

Aleg wrote: Well the grass always looks greener on the other side, but sofar nobody has ever produced even a decent sounding Linux audioplayer, and MQn tops everything else out there.

Very true.

elaprince wrote:v48 play I there is new king on the block
Sounds amazing in one word
Thanks

+1

My dedicated listening room is very small and has booming bass. I experimented with RoomEqWizard and found that Foobar with convolution sounds better than MQn in my room.
So now I'm attempting to add convolution to MQn. In MQn.bat I execute convolverCMD but MQn cannot play the convolved file.
The mqncontrol window opens and closes faster than usual and mqnplay does not execute.
Foobar plays the convolved file normally. The only difference apart from the sound quality is that the file is identified as PCM (floating-point) whilst the original file is just PCM.
Does anything prevent MQn from playing PCM floating-point?