Tír Na HiFi

Aleg


5ab4 is 23220 I think, which was the recommended setting for 1024 versions. 46440 was for 2048 versions.

So what is the correct setting for 448 versions? Logic would suggest 10158 using the same ratio, but I wonder if logic has a part to play here!

Perhaps SBGK can put us out of our misery?

ok just tried a few versions

agree with aleg
448 seems better to me with 23220

jrling wrote:Aleg


5ab4 is 23220 I think, which was the recommended setting for 1024 versions. 46440 was for 2048 versions.

So what is the correct setting for 448 versions? Logic would suggest 10158 using the same ratio, but I wonder if logic has a part to play here!

Perhaps SBGK can put us out of our misery?

I believe that there is no a "correct" clock settings, only recommended, or preferred!
Like I mentioned before, the last 2.73 448 has a nice relaxed sound with 10158, but I never like it the bass.
With clock at 23220 It's a bit more rough, but with the best punchy bass I ever heard so far. Even better than the "2.71 Intel rb 1024 sw", what was my "roll-model" for bass.

The new "2.73 1024 sq 0x8 raw background" reminds me more to 448's with 10158 clock settings. Beautiful middle range and nice, soft highs, but with bass to forward and to mellow, to soft, not punchy enough.

Still prefer the 2.73 448 raw background the best. Second will be the 2.73 1024 no raw background. Both with 23220 clock settings.

clockrate is DAC dependent, also needs to round to nearest timerresolution, also needs to match software player buffer. otherwise, there will also be negative effect and you never be satisfied. i would say 10158 is only ok rounded. you will hear issues sooner or later.

sgbk, why don't you try 10ms buffer with
435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445...448

different DAC handles 10ms buffer differently.
my DAC apparently likes <=444 as 10ms
every DAC is different.

on paper, my DAC can handle 40ms buffer 441*4/44100=40ms exactly. in real listening, not good.
think that bigger the buffer, the less is the error to having exact buffer for everyone.

another example, my DAC can handle 1ms buffer. in exact number, it needs 48samples instead of 41samples theoretical best. as you can see, the error is huge.

jesuscheung wrote:sgbk, why don't you try 10ms buffer with
435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445...448

different DAC handles 10ms buffer differently.
my DAC apparently likes <=444 as 10ms
every DAC is different.

on paper, my DAC can handle 40ms buffer 441*4/44100=40ms exactly. in real listening, not good.
think that bigger the buffer, the less is the error to having exact buffer for everyone.

another example, my DAC can handle 1ms buffer. in exact number, it needs 48samples instead of 41samples theoretical best. as you can see, the error is huge.

I do not understand all the facts, but I am wondering if some of these applies on my case. I can't play any of 448 versions, even 2.73 1024 versions cause some issues, I'm still not sure why.
Some albums play without any problem, some would start to crackle after certain track. Some of my CD's were HDCD, and I have to carefully monitor if those causing this issue with 1024 or it's random.
Bottom line, I am not a happy user of these new versions. 2.73 1024 sq 0x8 raw background sound really good (probably the best sounding so far, I was able to test), but it showed these weaknesses too, at least once on Patricia Barber Verse album.

2.73 1024 sq 0x8 raw background is really good in my system, easily tops 448 movqda which was the best so far ;)

jesuscheung wrote:sgbk, why don't you try 10ms buffer with
435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445...448

different DAC handles 10ms buffer differently.
my DAC apparently likes <=444 as 10ms
every DAC is different.

on paper, my DAC can handle 40ms buffer 441*4/44100=40ms exactly. in real listening, not good.
think that bigger the buffer, the less is the error to having exact buffer for everyone.

another example, my DAC can handle 1ms buffer. in exact number, it needs 48samples instead of 41samples theoretical best. as you can see, the error is huge.

hdaudio has to be 16bit aligned, you can request a different buffer size, but for 10ms wasapi will give you 448.

sbgk wrote:

jesuscheung wrote:sgbk, why don't you try 10ms buffer with
435, 436, 437, 438, 439, 440, 441, 442, 443, 444, 445...448

different DAC handles 10ms buffer differently.
my DAC apparently likes <=444 as 10ms
every DAC is different.

on paper, my DAC can handle 40ms buffer 441*4/44100=40ms exactly. in real listening, not good.
think that bigger the buffer, the less is the error to having exact buffer for everyone.

another example, my DAC can handle 1ms buffer. in exact number, it needs 48samples instead of 41samples theoretical best. as you can see, the error is huge.

hdaudio has to be 16bit aligned, you can request a different buffer size, but for 10ms wasapi will give you 448.

i see! how about 448-16=432? (though 448 is closer to 10ms, but over it too)

actually why not try 20ms?

448*2-16=882
441*2=880

only differ in 2samples of error.

with 10ms
448-441=7 samples of error.