Tír Na HiFi

I don't understand the technicalities of it but I just judge by what I hear - I will try 6,6,6,23 and 12,12,12,23 and let you know if any change - the first 3 numbers seem to make the biggest difference to me.

Clive

just be careful. timing a-a-a-b
the chances of a correct 'b' is 7/(a*7). all other b is incorrect.
e.g. 6-6-6, the chance of a correct b is 7/42. the other 42-7=35 of b are all incorrect. be very careful.

clive101
whats your cpu motherboard and ram?

jesuscheung wrote:just be careful. timing a-a-a-b
the chances of a correct 'b' is 7/(a*7). all other b is incorrect.
e.g. 6-6-6, the chance of a correct b is 7/42. the other 42-7=35 of b are all incorrect. be very careful.

Is there a formula for determining the "correct" setting for "b"?

nige2000 wrote:clive101
whats your cpu motherboard and ram?

Asus P8Z77-M and 8GB (2x4GB) Corsair DDR3 Vengeance

Clive

jesuscheung wrote:just be careful. timing a-a-a-b
the chances of a correct 'b' is 7/(a*7). all other b is incorrect.
e.g. 6-6-6, the chance of a correct b is 7/42. the other 42-7=35 of b are all incorrect. be very careful.

Will check and try tonight- if I can get any improvement in SQ I will be happy.
BTW - I like your optimisations - less services / OS threads not necessarily best SQ

Clive101 wrote:

nige2000 wrote:clive101
whats your cpu motherboard and ram?

Asus P8Z77-M and 8GB (2x4GB) Corsair DDR3 Vengeance

Clive

Forgot to mention have gone to using only 1 stick of memory - better SQ

jkeny wrote:

jesuscheung wrote:just be careful. timing a-a-a-b
the chances of a correct 'b' is 7/(a*7). all other b is incorrect.
e.g. 6-6-6, the chance of a correct b is 7/42. the other 42-7=35 of b are all incorrect. be very careful.

Is there a formula for determining the "correct" setting for "b"?

given timing a-b-c-d

yes. d is the cycle needed to push data b+c onto somewhere else. in perfect world, let say 9 9 9, d should be 18 exactly. but in real life, you need a 'delay', so, the best d is 18-1=17. you see most ram is rated at 9-9-9-24. 24=9*3-3. 3 cycles delay.

the bigger the a b c, the bigger the 'delay'. for most rams, for tight timing like 6-6-6, delay=1 is enough. so 6-6-6-11, or 6-6-6-17 are usually correct timing, the most effective use of your ram. in terms of OC, it makes no difference in benchmark. in terms of SQ, it makes HUGE difference.

What amazes me with all of this is that we get such good sound from an unsuitable platform.
I'm wondering how much more sound quality there is left to squeeze out of a PC or other processor based devices?
I have heard the squeezebox & some SD card based systems but none came up to the standard of sound that we currently have heard from some modded/tweaked PC configurations. These were dedicated audio platforms but there implementations were not optimal.

So I'm wondering, are we 70%, 80%, 90%, 99% near the top of what's possible in the sound quality stakes when using such modded/tweaked PCs? Are we ultimately being limited by the PC platform? In other words, if we changed hardware platform away from a general purpose PC to to a processor more dedicated to audio & optimally implemented, what % improvement is there to be achieved? A 1% or 10% improvement might suggest it wasn't worth the effort of looking elsewhere but 20 -30% improvement would spur on efforts in this direction.

JC, thanks for the explanation

JC strikes again, and I thought I was going off topic.

don't know what % we are at. Would need to audition something like the NAD that uses different technology.

It's the convenience of a one/two box solution that is also attractive, goodbye to all the tweaking.

I'm pretty sure the future of high end is dedicated hardware and software, there are just too many compromises when using a pc.