Tír Na HiFi

shoom wrote:6001

Tested with W8.1, WaveIO.

Portaudio v30 normal, mqnplay 7.16 sse, control.exe 3.64

Listening through buffalo DAC into Ventus discrete HA into K701 phones ATM.

Sounding awesome.

Love the extra clarity and focus.

yup, haven't heard anything like it before, bit of a groundbreaker. Why can't ms make wasapi sound like that.

sbgk wrote:

shoom wrote:6001

Tested with W8.1, WaveIO.

Portaudio v30 normal, mqnplay 7.16 sse, control.exe 3.64

Listening through buffalo DAC into Ventus discrete HA into K701 phones ATM.

Sounding awesome.

Love the extra clarity and focus.

yup, haven't heard anything like it before, bit of a groundbreaker. Why can't ms make wasapi sound like that.

Yes some prefer WASAPI but in my set ups I’ve always found it a little too flat (laid back) and lacking in top end detail (sparkle?).

With Jplay I’ve always preferred KS and then ASIO over WASAPI and it seems to be confirmed again here. I liked your WASAPI player which really opened my ears a little to the possibilities but the new KS seems to be better balanced and more detailed to my ears.

The KS player also seems to be more dynamic as I think Massimo commented re the loudness thing.

It’s way out in front of the other players that I’ve heard.

Sounding superb.

Aleg wrote:

nige2000 wrote:https://www.dropbox.com/s/54xx1fjs9rll2 ... eaming.reg


thats the old reg file for ks in core

IIUC it is not just a matter of reg-changes but also needing to copy ksthunk and ksuser drivers AND change registry settings.

yes ksthunk is needed. and apparently portaudio needs more...

if i disable "Windows Audio" service, and then MQn will not make sound even in GUI mode.

just tested this combo:
portaudio_x64.dll v30 normal
mqnplay1644.exe 24 bit 7.15 wdmks sse2
and using latest mqncontrol


have to say, starting to enjoy portaudio!

think still need to improve the agility and tempo and body with musical notes

jesuscheung wrote:just tested this combo:
portaudio_x64.dll v30 normal
mqnplay1644.exe 24 bit 7.15 wdmks sse2
and using latest mqncontrol


have to say, starting to enjoy portaudio!

think still need to improve the agility and tempo and body with musical notes

7.16 has more body

expect there to be another improvement when write direct to the device and then see if can do without pa

sbgk wrote:
expect there to be another improvement when write direct to the device and then see if can do without pa

Now you're talking! That's extremely interesting and potentially a whole step up in SQ. Let's hope so.

Do you think however that we will be entering a potential minefield of the interaction with the various vendors' drivers being able/unable to accept direct writing? Suppose there is only one way to test that. Thesycon would be a good one to try as it is used by several vendors.

Jonathan

Shouldn't have any effect on the driver. Currently there is an intermediate buffer which is used to split the interleaved wav file into separate channels before writing to the output pin. The plan is to load the wav file into ram already split out into channels ie interleaved at 1792 bytes instead of 2 bytes and then the load function can just load straight from ram to the required channel.

in pa it also does the conversion from 24 to 32 etc at this stage so not the smartest thing to do for sq.

Hmmm
Sbgk has a amanero so...
maybe he meant more like mqn wasapi without pa
Writing a mqn ks driver sounds cool for a usb i2s device would it be too much work?

Unfortunately 7.16 sse2 is not working on any of the server 2012 versions as described below. Looks like mqnplay stops before playing starts, at least in my setup.

Tested on Server 2012 R2 with AudioPhil's Optimizer, single PC

Mqncontrol 3.64, portaudio 30, mqnplay 1644 7.15 wdmks sse2
GUI: ok GUI full AO optimized: ok
Minimal server: ok Minimal server full AO optimized: ok
Core full AO optimized: not working

I am still amazed listening to 7.15 ;)
Thomas

sbgk wrote:Shouldn't have any effect on the driver. Currently there is an intermediate buffer which is used to split the interleaved wav file into separate channels before writing to the output pin. The plan is to load the wav file into ram already split out into channels ie interleaved at 1792 bytes instead of 2 bytes and then the load function can just load straight from ram to the required channel.

in pa it also does the conversion from 24 to 32 etc at this stage so not the smartest thing to do for sq.

Sounds to me that you are well immersed in KS - and may be actually enjoying it!
Everything you say confirms to me what you always said - KS has to be 'better' than WASAPI as it cuts out a layer of unnecessary stuff getting in the way of direct communication with output device (paraphrasing badly probably).

Good luck.

Jonathan