Visual studio 2012 c++ and wasapi minimalist player

[jesuscheung]

sound was a bit closed in, uploaded 2.53 sse4 intel full loop which brings it back to life

 

2.53 sse4 intel full loop

-previous versions' brightness has converted into mellow sound. convert more! still a little bright. mellow sound is getting close to become rax no zi.

-need more depth, longer duration of impact.

-everything else is good and better. no complain.

[LowOrbit]

Just had a quick listen to a couple of the new 2.53 versions. Slight changes in tonal balance don't bother me as I EQ using DSP anyway. I focus more on detail on the rationale that the more detail is coming through cleanly the closer other aspects of the sound will be to the mastering engineer's intentions.

 

I'm inclined to agree with Goon-Heaven - 2.53 sse4 intel add eax new loop mov r9 is very good, deep solid bass, great timing and clarity.

[jesuscheung]

Just had a quick listen to a couple of the new 2.53 versions. Slight changes in tonal balance don't bother me as I EQ using DSP anyway. I focus more on detail on the rationale that the more detail is coming through cleanly the closer other aspects of the sound will be to the mastering engineer's intentions.

 

I'm inclined to agree with Goon-Heaven - 2.53 sse4 intel add eax new loop mov r9 is very good, deep solid bass, great timing and clarity.

 

agree all eax versions in 2.53 have good timing.

[taggart]

I do have the later version which does allow for firmware updates via the Thesycon control window. I updated to the 384k/32 bit driver without any problem. Lucian's daughter board is highly anticipated, but given the improvements that applications like MQn bring in delivering a cleaner data stream I am not sure how much we have to gain from what the new board will deliver.

 

Same here. But seems to me that the 32/384 driver is only 32bit at that frequency. 44.1 is 16bit and all other HiRes Frequencies are awaited in 24bit.

 

The 24bit Version of mqncontrol.exe works with the Thesycon driver for WaveIO board, but only for 24/96 files. I couldn't play 16/44.1 files with that version, but maybe that could be fixed later. Thanks so far sbkg!

[sbgk]

Same here. But seems to me that the 32/384 driver is only 32bit at that frequency. 44.1 is 16bit and all other HiRes Frequencies are awaited in 24bit.

 

The 24bit Version of mqncontrol.exe works with the Thesycon driver for WaveIO board, but only for 24/96 files. I couldn't play 16/44.1 files with that version, but maybe that could be fixed later. Thanks so far sbkg!

 

ok, shall get on and do the 24 bit version.

 

24 bit is not ideal as it is 3 Bytes (3 x 8 Bits) , CPUs deal with cache lines of 64 bytes, 3 doesn't go into 64 and the data is loaded into cache in 16 Byte chunks, so with 16/32 bit the data can be loaded as 8 x 16 Bytes = 128 Bytes at a time and with 24 bit it can is loaded in 9 x 16 Bytes = 144 Bytes, this means it loads 2 whole cache lines and 16 Bytes of a third cache line, this is called a cache line split as the whole of the 3rd cache line is loaded even though it isn't used and can lead to extra cpu cycles and more noise.

 

Whether this will affect the SQ I don't know yet, but it is a reason why I think 24 bit is not a particularly smart format, because extra noise will be generated and it won't be a regular noise. Think AMD cpus are less affected by cache line splits and haswell isn't affected so much.

[lekt]

sound was a bit closed in, uploaded 2.53 sse4 intel full loop which brings it back to life

 

3 last 2.53 versions: soundstage was more than norm, all detail sounds are become near at ears. i think reference to v1.2 for soundstage/3Dwidth, just only enough, then keep vocal focus and have correct/good stereo effect. thanks, sbgk.

[jesuscheung]

....

I'm inclined to agree with Goon-Heaven - 2.53 sse4 intel add eax new loop mov r9 is very good, deep solid bass, great timing and clarity.

 

did you mean 2.53 sse4 intel eax new loop mov r9? no "add" seems to have better or the fullest, deepest bass of all versions. i am not good with bass. someone confirm pls.

 

actually, 2.53 sse4 intel full loop sometimes has better bass. even 1.2 too.

[nige2000]

Same here. But seems to me that the 32/384 driver is only 32bit at that frequency. 44.1 is 16bit and all other HiRes Frequencies are awaited in 24bit.

 

The 24bit Version of mqncontrol.exe works with the Thesycon driver for WaveIO board, but only for 24/96 files. I couldn't play 16/44.1 files with that version, but maybe that could be fixed later. Thanks so far sbkg!

 

Funny that I had assumed that the rpac having the xmos chipset and thesyscon drivers same the wavio that with the latest firmware and drivers would be working the same as the rpac

 

all bitrates and frequencies so far seem to work perfect with mqn 2.53 xxxxxxxxx in the rpac

[sbgk]

did you mean 2.53 sse4 intel eax new loop mov r9? no "add" seems to have better or the fullest, deepest bass of all versions. i am not good with bass. someone confirm pls.

 

actually, 2.53 sse4 intel full loop sometimes has better bass. even 1.2 too.

 

the new loop versions were trying a trick to load the cache line, but think it just read the first 16 Bytes so there were cache misses due to the missing data not loaded into the cache, the full loop is equivalent to add eax new loop mov r9, but with full cahce lines loaded into cache - does it sound any better now you know that ?

[jesuscheung]

the new loop versions were trying a trick to load the cache line, but think it just read the first 16 Bytes so there were cache misses due to the missing data not loaded into the cache, the full loop is equivalent to add eax new loop mov r9, but with full cahce lines loaded into cache - does it sound any better now you know that ?

 

i am talking about the version without the 'add'. was listening to Metallica with 2.53 sse4 intel eax new loop mov r9, forgot which song, the drum was super deep and full, no version comes close to it, and i thought it has the best bass. and then listening more, drum that has shorter duration of vibration, full loop sounds better, more open. by the way, full loop's bass is not full enough.

[goon-heaven]

the new loop versions were trying a trick to load the cache line, but think it just read the first 16 Bytes so there were cache misses due to the missing data not loaded into the cache, the full loop is equivalent to add eax new loop mov r9, but with full cahce lines loaded into cache - does it sound any better now you know that ?

 

Would that not have resulted in stutter? Please explain

[sbgk]

Would that not have resulted in stutter? Please explain

 

No, the cpu checks to see if the data is in cache, if not it has to go to ram which takes hundreds of cycles which presumably generates noise which can be heard as loss of detail or some times sibilance.

[goon-heaven]

No, the cpu checks to see if the data is in cache, if not it has to go to ram which takes hundreds of cycles which presumably generates noise which can be heard as loss of detail or some times sibilance.

 

So, if by some remotest instance, I heard more detail and cleaner SQ on 2.53 sse4 intel add eax new loop mov r9 (after mqnplay.exe 2.53 sse4 intel 8 16 16 4 no tlb), would that suggest mqnplay here wernt suffering from cache misses?

[John Dot]

2.53 sse4 intel eax new loop I think this could be the final version for me. Just perfect at all.

SBGK great idea to start the blog!

 

BTW is there difference between mqncontrol.exe "core affinity" and process lasso?

[sbgk]

So, if by some remotest instance, I heard more detail and cleaner SQ on 2.53 sse4 intel add eax new loop mov r9 (after mqnplay.exe 2.53 sse4 intel 8 16 16 4 no tlb), would that suggest mqnplay here wernt suffering from cache misses?

 

I thought it sounded funny, so thought it wasn't happy about something.

[sbgk]

2.53 sse4 intel eax new loop I think this could be the final version for me. Just perfect at all.

SBGK great idea to start the blog!

 

BTW is there difference between mqncontrol.exe "core affinity" and process lasso?

 

all core affinity does is change the process affinities, process lasso can set the priorities as well etc, but it's an extra process running on the box and dynamically monitors/adjusts the processes, haven't used it in a long time.

[sligolad]

Hi sbgk, first post here regarding MQN and all I can say is....stunning work! It is a long way from the years posting regarding the TT updates to the Touch from Klaus on the Slimdevices forum which I followed for a long time. I have been a JPlay convert for the past few years but I have to say in all honesty your efforts here have surpassed anything I have heard in my journey with JPlay. I will be getting up to speed soon with your work here but so far I am blown away with what I am hearing. If you have a donate link I will be happy to give a little back for what is an excellent endeavour you have here. Continued success and now I am off to try some more recent versions. Cheers, Pearse.

[tonerei]

2.53 sse4 intel eax new loop I think this could be the final version for me. Just perfect at all.

SBGK great idea to start the blog!

 

BTW is there difference between mqncontrol.exe "core affinity" and process lasso?

 

This man speaks does not speak in a forked tongue. If Sbjk decides to go on strike it matters none as this version means one does not notice as 2.53 see4 intel eax new loop (what a mad name) means one just keeps playing full albums till it is time to retire for the night.

 

BTW It would still be nice to be able to play hires stuff just to see if there is something beyond heaven.

[tonerei]

Hi sbgk, first post here regarding MQN and all I can say is....stunning work! It is a long way from the years posting regarding the TT updates to the Touch from Klaus on the Slimdevices forum which I followed for a long time. I have been a JPlay convert for the past few years but I have to say in all honesty your efforts here have surpassed anything I have heard in my journey with JPlay. I will be getting up to speed soon with your work here but so far I am blown away with what I am hearing. If you have a donate link I will be happy to give a little back for what is an excellent endeavour you have here. Continued success and now I am off to try some more recent versions. Cheers, Pearse.

 

 

Hard to believe Pearse that one could buy the meicord one week to bring Jplay to the next level and the following week it is all thrown away for MQn. For me at least it is only the meicord you have a cirrus now looking for a new home!

[sbgk]

Hi sbgk, first post here regarding MQN and all I can say is....stunning work! It is a long way from the years posting regarding the TT updates to the Touch from Klaus on the Slimdevices forum which I followed for a long time. I have been a JPlay convert for the past few years but I have to say in all honesty your efforts here have surpassed anything I have heard in my journey with JPlay. I will be getting up to speed soon with your work here but so far I am blown away with what I am hearing. If you have a donate link I will be happy to give a little back for what is an excellent endeavour you have here. Continued success and now I am off to try some more recent versions. Cheers, Pearse.

 

Hi, glad you're enjoying the experience, hope there are some more improvements still to come.

[John Dot]

This man speaks does not speak in a forked tongue. If Sbjk decides to go on strike it matters none as this version means one does not notice as 2.53 see4 intel eax new loop (what a mad name) means one just keeps playing full albums till it is time to retire for the night.

 

BTW It would still be nice to be able to play hires stuff just to see if there is something beyond heaven.

 

It was just an opinion. Don't get it too serious.

Cheers

[tonerei]

It was just opinion. Don't get it too serious.

Cheers

 

No not taking it too serious just agreeing and acknowledging the work done here.

[sligolad]

Brilliant, thanks. Shall update the files in the upload site.

 

suppose the second

taskkill /fi "PID gt 0" /IM mqnplay.exe /F >nul

 

isn't required.

 

Tried the bat file for FLAC conversion but no luck and also tried to code from Adolfo with no luck, has MQN moved on past working for FLAC conversion file at this point with 2.53? Cheers, Pearse.

[sligolad]

This man speaks does not speak in a forked tongue. If Sbjk decides to go on strike it matters none as this version means one does not notice as 2.53 see4 intel eax new loop (what a mad name) means one just keeps playing full albums till it is time to retire for the night.

 

BTW It would still be nice to be able to play hires stuff just to see if there is something beyond heaven.

 

Been playing 24 96 files this evening on 2.53 new loop with no issues! Cheers, Pearse.

[sbgk]

Hard to believe Pearse that one could buy the meicord one week to bring Jplay to the next level and the following week it is all thrown away for MQn. For me at least it is only the meicord you have a cirrus now looking for a new home!

 

don't know how good MQn is or whether it just sounds good because it's new, I think it is still far from perfect, still it shows people what's important in the SQ chain.