Tír Na HiFi

Aleg wrote:

sbgk wrote:

Aleg wrote: Not done extensive testing but immediately noticeable is loss of control in deep bass, which becomes booming/blooming.

ok, reversed that change.

one more thing to try, have uploaded 4.81 and then can call it a day.

Still staying with 475 256avx128

4.81 sings a bit less than 4.75 256avx128.
It can sound a bit shortened sometimes, tones not finishing fully. This reduces the sense of the acoustic space. Some recordings that are 'famous' for that aspect (often from the ECM label, which is 'known' to bring a lot of the space into the recording) become more flattened. Try this with John Taylor - Rosslyn (ECM).

Control of deep bass is there, not as insightfull as the 4.75, but it is no longer blooming (opening out, escaping, no control), but it is still a more textureless 'boom' which can become booming and annoying if played at more volume. The 4.75 gives more texture and insight in this deep bass sound whereby it is not just a closed fat 'boom'-sound, but you can hear the fluctuations in the tone caused by the swinging of these large, thick and somewhat loose strings on the double bass which create these deep bass notes. This does not become irritating a higher volume as you can hear the string swing and sound varying with it, which is completely natural.

Vocals are somewhat smoother on 4.81 and have somewhat more clarity on 4.75. Tonally both natural and equally pleasant.

Shortened note also noticeable on the Haydn Sting Quartet from Alban Berg Quartet.

My impression is that 4.75 is more strict on the timing than 4.81, and is the one that is more true.

As said 4.75 256avx128 still has my vote for nr 1.

Cheers

Aleg

m.massimo wrote:Listened to latest sse2 versions (I cannot play avx).
My preferred version evolved in the last weeks from 3.56 + 1.6 -> 4.22 512 + 3.61 -> 4.48 512 + 3.61 -> 4.74 256 + 3.61. I'm listening 4.82 256 + 3.61 right now and it's stunning. I'm playing clarinet+accordion duo albums (Mirabassi + Galliano or Biondini, all from Egea) and the clarinet really sings.
Listened to Haden+Forcione - Heartplay as well and it confirms that 4.82 256 sse2 is excellent. Double bass seems to me very good.
I'd never thought just 6 months ago a 1644 could play so good. Many thanks to sbgk.

m.massimo wrote:Listened to latest sse2 versions (I cannot play avx).
My preferred version evolved in the last weeks from 3.56 + 1.6 -> 4.22 512 + 3.61 -> 4.48 512 + 3.61 -> 4.74 256 + 3.61. I'm listening 4.82 256 + 3.61 right now and it's stunning. I'm playing clarinet+accordion duo albums (Mirabassi + Galliano or Biondini, all from Egea) and the clarinet really sings.
Listened to Haden+Forcione - Heartplay as well and it confirms that 4.82 256 sse2 is excellent. Double bass seems to me very good.
I'd never thought just 6 months ago a 1644 could play so good. Many thanks to sbgk.

sbgk wrote:just can't get the bass with 2.82, uploaded 2.83 256 avx 64 nop which is better all round for me, at least.

Aleg wrote:

sbgk wrote:just can't get the bass with 2.82, uploaded 2.83 256 avx 64 nop which is better all round for me, at least.

Listened to 4.84 which while good in itself, is not as good as 4.75 256avx128.

What is less good is the space around the instruments and the way the sound is decaying, it is shortened in 4.84 while 4.75 can decay fully like in the older best versions.

Cheers

sbgk wrote:

Aleg wrote:

sbgk wrote:just can't get the bass with 2.82, uploaded 2.83 256 avx 64 nop which is better all round for me, at least.

Listened to 4.84 which while good in itself, is not as good as 4.75 256avx128.

What is less good is the space around the instruments and the way the sound is decaying, it is shortened in 4.84 while 4.75 can decay fully like in the older best versions.

Cheers

that leaves 4.82 nop to try, think it's an improvement

see jplay say their jplay mini doesn't do any data copying, haven't tried it, but sounds interesting. Beyond my skills.

think 4.82 nop is the best one, better articulation, definition than 4.82.

Aleg wrote:

sbgk wrote:

Aleg wrote:
Listened to 4.84 which while good in itself, is not as good as 4.75 256avx128.

What is less good is the space around the instruments and the way the sound is decaying, it is shortened in 4.84 while 4.75 can decay fully like in the older best versions.

Cheers

that leaves 4.82 nop to try, think it's an improvement

see jplay say their jplay mini doesn't do any data copying, haven't tried it, but sounds interesting. Beyond my skills.

think 4.82 nop is the best one, better articulation, definition than 4.82.

I have only got a 4.83 256avx64 nop and not a 4.82 nop,

I find 4.83 nop much better than 4.84.
I think it might also be somewhat better than 4.75 256avx128.

4.75 has a sense of a bigger acoustic space.
4.83 nop has also got a good decay, but in a slightly smaller space, which might be more realistic than the larger space of 4.75

Sound balance, bass control all very good.

So 4.83 256avx64 nop might take over first place.

Cheers

Aleg

sbgk wrote:
uploaded mqnplay1644.exe 24 bit 4.82 256 avx 128 nop

and

mqnplay1644.exe 24 bit 4.82 256 avx 128 nop r15d which a different way of looping