XXHighEnd - The Ultra HighEnd Audio Player
December 22, 2014, 03:55:53 pm *
Welcome, Guest. Please login or register.
Did you miss your activation email?

Login with username, password and session length
News: Dec. 31, 2012 : XXHighEnd + Phasure NOS1 DAC receive 6moons Blue Moon Award !
** "Lonely at the very top" **
Search current board structure only !!
Home Help Search Login Register
Pages: 1 2 3 4 5 6 [7] 8 9 10
 61 
 on: December 10, 2014, 03:06:36 pm 
Started by fralippo - Last post by christoffe

Still e.g. a laptop with the normal CPU frequency (say 10 times higher) may use 30% of CPU and it takes ages to load a track in the mean time.
It just can't be compared ...


The main point since **NOS1a** is, what PC configuration produces the best SQ, and everything else (components) follows this judgement.

A) PC system with an internal PS
or
B) PC system with an external**linear** PS

Joachim

 62 
 on: December 10, 2014, 01:26:55 pm 
Started by michaeljeger - Last post by PeterSt
Hi Michael, let's try ...

The major important one you most probably already use in well fashion, but we have to begin there ... :

When a connection is inherently slow all depends on the packet size (of transfer). So think very small packets and because of the two way communication needed for this, the overhead is huge. Now FLAC ...

FLAC itself is organized so that it *creates* small transfer "packets" (blocks). Notice this can be tuned, but this has already been done and is optimized within XXHighEnd. So we have two phenomena at hand :

1. A slow connection, especially on the two-way communication (blame the Mac here);
2. Small blocks of data to transfer (say blame FLAC now).

How to solve this ?
Avoid some whatever disk subsystem layer with the same interrupt usage or whatever it is I found. How ?
Play form RAMDisk.

I must assume that you already do this, but still it is no coincidence that you exactly mention what I know from the real slow connections (like WiFi) and which plainly is solved by means of that RAMDisk. But two requirements here :

a. XXHighEnd itself must be on a RAMDisk (and thus must be started from there).
b. The Playbask Drive must be on a RAMDisk.

When this both has been arranged for, you don't notice a thing of any slowness, not even at (I forgot) something like 1Mb ... ah ... I recall ... Bluetooth. So without this solution a track could take 1-2 minutes, and with this solution maybe 5-10 seconds.

What came from this solution for the very slow connection, is that I now generally tell that the RAMDisk playback etc. is a requirement for when playing over a network (NAS will be the same). So to my own surpise this seems even faster than loading music from an internal hdd. In the end it is all about that ... the hdd stuff which is super slow. Relatively. And what I expect is that the "processing" on the hdd for FLAC (all the small bits and pieces) sort of detach the network connection (or don't let it nicely flow) and the overhead for this now is too large to be acceptable.

Now tell me if this helps ...
Peter



 63 
 on: December 10, 2014, 01:10:47 pm 
Started by Robocop - Last post by PeterSt
Hi Dave,

In addition to what Anthony (acg) said, Xeon's are not better that I can see. They are slower on the (max) frequency and btw both the i7 and the Xeon don't have on board graphics. OK, it depends a bit on the type (of i7) but now this is also related to something you may have overlooked : the CPU socket.

So if you go for the LGA2011 ... in there both go i7 and Xeon. This can be very convenient. Btw, where our PC could only focus on the X79 chipset for this, today this is X99 and don't ask me how the choices are for that at this moment (I just didn't investigate it - I think Michael did).
In the end this is all about the existence of MoBo's and for LGA2011/X79 there's one reliable MiniATX only (that I know) : ASRock. Small downside : The X79 version was taken out of production a year ago, and I assume it will be replaced by a X99 version. Or not at all ... and then it stops for small MoBos.

The only positive that I can think of with the Xeons is the number of cores. Like 20 (hyperthreaded) if you want (or even 24 - I forgot).
Thinking Xeon is a bit dangerous at this moment, because new I7's will be there with more cores than the 12 from today. Or are already there - again I didn't look. But the 12 cores (or native 6) will be 16 ( 8 ) at least because the i7 actually already was that. But it couldn't be utilized because of not enough pins on the chip possible. I do this by heart : I suppose that LGA2011-3 has the sufficient amount of pins so the already existing 16 core i7 (design) can go in against a fair price.

Now I should start digging myself I suppose ... Happy
Peter

 64 
 on: December 10, 2014, 12:55:48 pm 
Started by fralippo - Last post by PeterSt
Hard to explain in a few words when you didn't follow this all. But think over-dimensioning ... OK, now you still won't understand.

Today's better (or best) CPU's are not only faster for frequency, but they contain fast memory caches, more memory lanes and a lot more. These parts won't be divided by 10 when the frequency goes down by a factor of 10. Frequency (speed) is important, but it is by far the least important (for what we do with it). So notice :
When this is optimized like I did it for Windows 8 (see sig for Windows 8 (not the NOS1a link)) then the PC uses 48 Watts during playback of 32/750600. And that with a CPU usage of virtually 0%. Still e.g. a laptop with the normal CPU frequency (say 10 times higher) may use 30% of CPU and it takes ages to load a track in the mean time.
It just can't be compared ...

Peter

 65 
 on: December 10, 2014, 12:51:40 pm 
Started by christoffe - Last post by juanpmar
Thanks Peter, l pass the information to the technician.

Juan

 66 
 on: December 10, 2014, 12:48:13 pm 
Started by christoffe - Last post by PeterSt
Hey Juan,

If I see it correctly, that capacitor is soldered on the ground pin (of the three of the XLR terminal I mean). You can measure (Ohm) whether the terminal itself is connected to the chassis. So I mean : whether the metal of the terminal where the cap is soldered to, is connected to the chassis. That assumed :

Solder the cap to the ground of the BNC terminal (the lip on the side) and connect the other end of the cap to the chassis somewhere (extend it a little with a wire).

Clear ?

Peter

 67 
 on: December 10, 2014, 12:31:33 pm 
Started by christoffe - Last post by juanpmar
Hi Peter,

Ive found that the XLR in my speakers have a capacitor, is there some problem if I remove it to solder the wires as you told me to place there the BNC connector?

Regards,
Juan

 68 
 on: December 10, 2014, 05:04:22 am 
Started by Robocop - Last post by acg

In doing my research a question has come up for Peter - Why use the I7 over a Xeon?  The Xeon has a lower TDP, doesn't have on board graphics, more cores and supports ECC RAM.  All these things seem like pluses for a Xeon over an i7.  Certainly a response from anyone else is welcomed as well.

thanks,
dave


   

Price I believe.  The Xeons are very expensive.  I purchased mine second hand for less than the cost of the i7 otherwise it would have cost twice the i7 new.

 69 
 on: December 10, 2014, 03:44:46 am 
Started by Robocop - Last post by music33
I've been researching out different computer builds.  After looking at many parts / configurations and comparing them it becomes pretty obvious that Peter's PC is very well thought out.  michaeljeger PC although different is also very well thought out.  I also really like acg's computer that he has listed in his signature.

The v4 Pipeline seems to have every bell and whistle - separate USB card, linear PS, passive cooling case all of which makes the computer expensive.  It isn't clear from the article how the different components impact the sound.  With the NOS1a these things may not matter as much.  From reading Peter, a more powerful chip would be better.  michaeljeger PC is closer to the v4 pipeline, but much cheaper to build.  it is more like a minimalistic build of the v4 pipeline with a better CPU and picoPSU.  looking forward to hearing about it.

In doing my research a question has come up for Peter - Why use the I7 over a Xeon?  The Xeon has a lower TDP, doesn't have on board graphics, more cores and supports ECC RAM.  All these things seem like pluses for a Xeon over an i7.  Certainly a response from anyone else is welcomed as well.

thanks,
dave


   

 70 
 on: December 09, 2014, 11:12:25 pm 
Started by Nick - Last post by Nick
Hi,

Quick update. The lash up cheap module clock and LifePo4 battery are out of the PC now. It's been replaced by a spare dexa I modified to work with the motherboard.

My little one is in bed again so I've been listening at -60db and it sounds really good, full sound and disconnected from the speakers   dntknw defiantly improved on the lash up clock even at these stupid volume levels. Bass so far as I can tell is better but need to turn up the wick to be sure it might be a couple of days before I can do this. Very frustrating.....

I think I may have a  way to produce clocks for USB, and Mobo with -95db phase noise at 10hz and femto second jitter levels, at significantly lower cost than the dexas. Need to buy some parts and have a play.

Fralippo hi, sure will post some pics.

Nick.

Pages: 1 2 3 4 5 6 [7] 8 9 10
Powered by MySQL Powered by PHP Powered by SMF 1.1 RC2 | SMF © 2001-2005, Lewis Media Valid XHTML 1.0! Valid CSS!
Page created in 0.262 seconds with 15 queries.