 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
88.110.47.8
Several posts over recent months, especially on the cMP2 threads, have discussed RAM quality and settings and how these affect playback.I found many claims to be contradictory and remained unconvinced: I stuck in a decent-quality one GB Corsair stick into my audio PC and left motherboard settings at SPD.
However, I recently bought a couple of the much-praised Kingston Hyper X modules (KHX6400D2K2/1G) and set timings at the recommended 3-3-3-8.
After installation, the system sounded about as attractive as a cat swallowing broken glass. Reverting to the Corsair stick restored quality.
It demonstrated to me, though not quite how I'd hoped, that RAM does seem to matter.
Can anyone offer suggestions as to what RAM to choose (and why) and what settings to use (and why). I have the impression that I'm far from alone in floundering here but would like to be proved wrong.
Best
Dave
Edits: 01/21/09Follow Ups:
i did a brief calibration listen, then moved my 2 gig 5-5-5-18 corsair ram stick from one slot to the other (new position was closer to the 24 pin power plug on the mobo -- and the sound opened up, got more dynamic, seemd to flesh out the top and bottom, and brought out the upper mids lower highs a bit. it did sound very slightly grainier though.
then i swapped the corsair ram for the 1 gig hyperex LL version ram stick, in the "better sounding" slot.
all of the improvements were still there, but the grain was gone, and a lucid, coherent sound took its place, with all of the gains in dynamics, immediacy, and open-ness intact.
next step is to finally implement the cics bios tweeks, since i got the cics recommended mobo in the first place. (i have an extra one, because i thought an ebay seller was jerking me around-- maybe he was, it only shipped once i demanded they cancel my order, and ordered another one from a different source.....)
happy happy joy joy!!!!!!!
i've been using a single 2 gig stick of corsair ram, spec-ed at 5-5-5-18. it's a tight fit with my new essence stx sound card, so i was going to move it to the other memory slot in my cics gigabyte mobo, but have not gotten around to doing it yet. i have not, btw, done any of the cics bios optimizations yet.
being impulsive and impressionable I tried to find some of the UL kingston hyperex ram, to no avail, but i did find a stick of the LL (1 gig), which has 4-4-4-12 specs.
i will move the 2 gig ram stick first, and see if the slots make any difference, then i'll put the LL into the slot which affords my xonar more space. i will post the results when i figure them out.
I can just imagine the wheels turning in some entrepreneur's mind... "Hmmm... audiophile grade RAM!"
LOL
We tested many kinds of RAM when we were designing the Revealtion PC-based systems. We heard lot's of differences between RAM, but it was not related to price or quality. Changing timing does make a difference, but results were not reliable enough to come to any conclusions. Lower latency in some cases sounded better.
The manufacture of RAM changes so drastically -- the same model RAM can change the source of it's chips and not have any change in part number. So even if you find a stick of RAM that sounds good, if you buy another a month later with the same part number, it might sound different.
That's why it's so hard to pin down what RAM reliably sounds good.
Did you use any type of scientific method to determine how you would change latencies on different RAM products?
Also its my understanding that "some" motherboard manufacturers are optimized to work with specific companies RAM products or more importantly just don't work well with other companies RAM.
Like with everything else in audio, there is no true "silver bullet" solution to obtain perfect audio, rather some synergistic homogenization between different variables that must come together in order to get good sound. Generally speaking I find that its always best to leave the house of cards alone for the most part and only try to adjust certain key features, otherwise you are going to have a delicate balancing act on your hands.
Karma Means Never Having To Say You're Sorry
Dynaudio is right, I build custom computers and the mainboards manual will list recommended ram company names. Not that one brand not listed wont work. I use 800Mhz with duo-core CPU's, 1600Mhz with Tri or Quad core builds. I personally use Corsair 800 Mhz in my Main Rig. Alan
Vista Ultimate 32bit/Diamond XS Dac/ Sterovox coaxial line in to Insignia Amp/Cambridge SoundWorks& Infinity RS 1001 Speakers
Well, since only certain combinations of latency works with any given motherboard and memory combos, it is tough to do it with rigour across different brands.
We discovered this 5 years ago when we were trying to build our Revelation computers with consistently good sound.
As computer builders, we found it very hard to build them consistently good. This is because the computer hardware changes so fast. In a three month period, either a motherboard manufacturer or memory manufacturer, etc,. will change their products. The turn over of new products kept us from being able to build with consistency over time. So we had to listen to every machine and be sure it met the sonic standards of our reference. This meant swapping in a lot of different parts, listening, and modding parts, a very long and costly procedure.
Ask Vincent about the tens of thousands of dollars of working but sonically inferior and now outdated computer parts he has in his garage!
This is one of the reasons we are happy to move to a monolithic platform that we can have consistency of sonics - a Mac and Amarra.
And on the Apple platform, each model of Mac has a different sonic signature as well - the sum of all it's parts.
I have taken to building my own Macs, or Hackintoshes, to test the differences parts make to them. And it's the same story as the PC.
So, the unpredictability of PC audio is finally being admitted after I first said sosome 2 years ago. Then nobody would openly say so, except stress the superiority of PC audio over boxes.
The world does turn around.
fmak,
Unpredictability does not equal inferiority. Even if I made a box with all the 'wrong' choices, it still bested my $10,000 CD/Transport rig. That is my opinion anyway.
But just as a specialist tube designer does not sell an amp without listening and tuning to a specific build of a type of tube, we never made a system without listening and tuning it to meet our reference standards.
We never had a goal for mass production, partly because we knew that it would be impossible with current equipment standards.
Its only a matter of time before some manufacturer can make enough money on a PC platform to have the strength to get a Motherboard, RAM etc. manufacture to make a MB, RAM or other components to spec and not be forced to purchase off shelf ever changing products.
If company A were able to nail down a certain combo of parts that gave them the sound they wanted it would behoove them to get those parts and only those parts for future builds. As it stands, it seems, that PC audio manufacturers do not have the muscle or the financial numbers to make them take on such a debt as to buy X amount of products from MB, RAM and other manufacturers that would have them make their parts to the PC audio companies specs. This would ensure consistency in quality and sound, ie the end product that we enjoy.
Karma Means Never Having To Say You're Sorry
Some years ago, we approached manufactures of computer parts to find out how much we needed to buy directly from them vs. their distributors. What was told to us, was that we would have to buy in a quantity that we could not even look upon, as there would be no way to sell that much product in a months time. You have to sign a contract for one year to buy x amount of product and let me be the first to tell you, that the amount is not in the hundreds! I have so many computer parts in my garage and storage, that I will have to donate them to a school, as the parts are brand new, but are either outdated, or sound quality is inferior.
Vincent
Which is why an optimized computer for the means of audio [at least for PC] is definitely not in the near future.
BUT, how about a few PC audio manufacturers teaming up with a PC manufacturer to lock down a specific MB etc. just for audio. Would a company like Dell for example have enough weight to team up with some Audiophile companies to get them to manufacture MB's to spec for the purpose of audio? Even IF the Audiophile companies would agree to share a certain amount of technology to the PC company?
The merging of Audio/Video/PC is inevitable, it would be in the best interest of even the PC companies to tap into the knowledge base of Audiophile companies to make better products for the entertainment sector. Of course this is long term and not considering the huge consumer base that already thinks MP3 sounds wonderful. Maybe this whole idea is years ahead of the game but as the 3 merge together it will most likely be advances in sound reproduction that will set one PC company apart from the others. Because only that GUI's are cheap, and initially it will be ease of use and GUI's that attract customers. But once that settles in, it will be up to concrete audible improvements that will set one manufacturer apart from others. Hoping.
Karma Means Never Having To Say You're Sorry
But bad RAM - which you would expect to be binary in that it either works or doesn't work - can cause a wide range of very hard to diagnose and categorize malfunctions on Macs. The malfunctions caused by bad RAM can look like software problems, can be intermittent, can look like video card problems, even like motherboard battery (partial, ongoing) failures.
Adjust your RAM settings and see if you can hear a difference.
Take some RAM out and see if you can hear a difference.
My bet is some will and some won't, just like everything else in audio....and those who don't hear a difference will say those who do ONLY do because of the placebo affect.
Perception is 9/10ths of reality...
Karma Means Never Having To Say You're Sorry
But it turns out there are shades of "working."
As for choosing RAM based on any perceived sound quality, I am afraid my commitment to good sound just doesn't extend that far if I'm paying for it. The enjoyment would be overwhelmed by concern about ever more esoteric variables. Cable elevation is already pushing it, when you consider the fact that most systems have a rat's nest of different cables all coagulating on the back of the preamp.
This is why digital audio is not down to 0 and 1. It is to do with software and hardware reading 0 and 1 in the right places at the right time.
> Cable elevation is already pushing it. . .
What, you mean you haven't hired a Feng Shui specialist
to align your audio system with the vibrational energy
fields in your neighborhood?
Or had an exorcism performed to expunge the jealous
spirits of gramophones past?
Full of weed to cope with all this.
> [J]ust like everything else in audio....and those who
> don't hear a difference will say those who do ONLY do
> because of the placebo effect.
>
> Perception is 9/10ths of reality...
Once upon a time, 15 years ago now, I did a fairly major DIY
project (the most elaborate one I'd ever attempted until I
made my "gainclones" in 2000). It involved:
1) Using S/PDIF out from a Radio Shack "Optimus" CD-4300 portable
CD player, powered by a 6-volt motorcycle battery into
2) an Audio Alchemy DTI powered by a 12-volt wheelchair battery,
into
3) a second DTI powered by a second 12 V. battery, into
4) a Theta DSPro Prime that I had modified myself so that its **digital**
section was also powered by a third 12 v. battery (through
an external voltage regulator chip mounted on a heatsink
that dropped 12V down to 5V). The analog section of the Pro Prime had
been completely bypassed so that the digital section would
feed the (6DJ8 tube) analog box of a two-box Melos MAX-1 two-box D/A converter.
(I got that modification through a Melos dealer in New Jersey,
who also arranged for Melos to sell me half of a MAX-1; both
the Melos and the Theta Pro Prime were based on the Philips 7350 bitstream DAC,
so the modification was practical.)
It sounded pretty good. (The speakers back then were my Quad ESL-63s,
run full range without a subwoofer, and powered by a Counterpoint
SA-4 OTL tube amp.)
But the thing that's apropos of your "placebo" remark
is that, during the course of my experiments with that set-up,
I tried to cheeze out (or economize) by putting both DTIs
on the same 12V battery (and having the ground return happen
through the S/PDIF ground connection between them).
The result? The "magic" completely disappeared. When I
went back to using separate batteries, the "magic" came
back.
Now, in spite of my own rampant audiophilia I was half convinced
back then that any improvements I thought I was hearing
were in my imagination. But that effect was beyond placebo,
especially since I wasn't expecting it. The "magic" I'm
referring to here was mostly in the bass -- the ambient
"bubble" of room tone in a recording -- that just went away
with the "inferior" power (or at least **grounding**) arrangement --
and came back, unmistakably, when I went back
to the separate batteries (with separate returns for the grounds,
which may have been the key difference rather than the two
separate batteries per se.)
It really made a big impression on me -- not just the sound,
but that I was able to hear the difference (and a fairly
subtle one at that) so clearly.
> I did a fairly major DIY project. . .
The riskiest thing about it (apart from having hydrogen
gas bubbling out of the motorcycle battery into the
air of my living room during its recharge cycle,
and presumably out of the gel cells too,
though at least with those I couldn't see and hear the fizzy
bubbles ;-> ) was that there could be a nasty transient
through the system if the Pro Prime's digital section
lost its DC power. That happened once because the
connection to one of the pins of the voltage regulator
came loose.
I didn't use two DTIs just for the sake of using two
DTIs. Because of the way the equipment was disposed,
there was a long distance between the transport and
the DAC, so I had one DTI right next to the transport and
one DTI right next to the DAC, with a fairly long coax
cable between them (the cable was long enough that it
couldn't be anything "audiophile approved" -- it had to
be an RG-59 RF cable with RCA adapters on the ends).
Wow Jim you really are a true Audio hobbiest. Have you put together any NOS DACs???
Do you know anything about the Transcendent Grounded Grid preamp???
Karma Means Never Having To Say You're Sorry
> Wow Jim you really are a true Audio hobbiest.
As far as the soldering iron is concerned, compared to
the folks who post in the DIY forums I'm strictly a
novice dilettante.
Same, I suppose, compared to the folks with the money
and connections to hear the latest truly cost-no-object
gear. Like, oh, Srajan Ebaen of 6moons; or Stephæn Harrell
of 6moons, The Absolute Sound, and who also is (or was)
a member of "The Bored" around here.
Speaking of The Bored, BTW, while we goobers chat merrily away
with each other, it's wise never **ever** to forget on whose
behalf and for whose benefit this place really exists (hint: who's
paying for the servers and the administration? It ain't
you, Mr. Consumer, and it ain't me neither.)
And if you are ever unfortunate enough to get into a serious
dispute with a manufacturer or a dealer, you will find no
audience for your problems here -- any such posts will
be disappeared at the first growl from said manufacturer
or dealer. Count on it.
> Have you put together any NOS DACs???
I have an Audio Note DAC Kit 1.1 that I threw together
back in 2000.
I've been trying to get a damned CS8414 input receiver
chip pre-soldered to a DIP adapter so I can pop out the
socketed CS8412 and pop in the CS8414. I know such a thing
exists, because the latest Audio Note 1x DACs come with
them.
But so far, nobody's answered my e-mails. :-(
(I do have a Scott Nixon NOS DAC with a CS8414 and
a TDA-1543.)
So which is better or what are the attributes associated with the CS84xx vs. TDA15xx??I was looking at a NOS DAC with the TDA chip and a DIRxxx receiver chip, it also has Blackgates throughout and big power supply...in fact I think it uses 8 chips in parallel
Karma Means Never Having To Say You're Sorry
Edits: 01/21/09
> what are the attributes associated with the CS84xx vs. TDA15xx??
CS84xx are S/PDIF input receiver chips; TDA15xx are Philips
16-bit ladder DACs. You need one from Column A and one from
Column B for a balanced meal. ;->
I want to ramp up to 96kHz input with my NOS DACs
(via CS8414) because I want to use my 24/192 upsampled
stuff (even if I have to downsample it again to 96kHz
and dither/noise-shape 24 bits to 16 bits). Listening
to unadulterated CD on a NOS DAC isn't really in the
cards for me anymore.
That said, all the NOS DACs I have are tubed, and I always
love the color and tonality of the tubes when I swap
'em back into the system.
Modern operating systems (and the applications that use them) generally perform better with a lot of memory at their disposal. My minimum build these days has 4GB, and my main audio computer has 8GB.Oh and make sure the RAM has been cryogenically treated! Just kidding (seriously...even the mere mention of something like that around here will turn it into a must-have feature, it seems).
Edits: 01/21/09
> Modern operating systems (and the applications that use them)
> generally perform better with a lot of memory at their disposal.
> My minimum build these days has 4GB, and my main audio computer
> has 8GB.
PC folklore claims that Windows XP can't handle (or handle "well",
or "efficiently" or something) more than 2 GB of RAM, but
that Vista performs well with 4 GB or more.
You're using Vista (or Mac OS-X, or Linux), I presume.
I believe the usable limit for 32-bit Windows operating systems is 3GB.
Although some users report that the whole 4G does not show up as available.
In mine it does....:-)
My other machine also Vista has only 2G, both seem to work fine...though my processes and everything else is at the bare minimum.
Karma Means Never Having To Say You're Sorry
In theory a 32-bit processor should be able to support up to 4GB, but there is some strange limitation in Windows that prevents it from using more than 3GB. Here's a link to an article that attempts to explain the situation:
http://www.start64.com/index.php?option=com_content&task=view&id=2140&Itemid=104
Windows is happily reporting that you have 4GB installed, but that doesn't mean it can actually use all of that!
Windows is happily reporting that you have 4GB installed, but that doesn't mean it can actually use all of that!Yeah I guess its really a mute subject seeing that my memory usage never goes further than 20% anyways, so even IF Vista were only using 3G of RAM I never use more than .6G of RAM anyways....:-)
During normal listening the machine is hovering around 14% memory usage...keep in my this is 'after' it loads the entire CD to RAM.
I guess that would explain why there is no difference between the 4G machine and the 2G machine running Vista.
So what do you do, that requires 8G of RAM???? Do you use your machine for gaming???
Karma Means Never Having To Say You're Sorry
Edits: 01/23/09
Yes, I play and create 3D games, do software development in Visual Studio, use spreadsheets, surf the web, and play audio in Foobar2000. All at the same time! With no difference in audio quality between running all that and running just Foobar2000. I continue to be mystified by all the attempts to minimize system usage and resources in order to get "high quality" audio around here...audio playback (assuming no resampling or other DSP-like manipulation of the data by the audio player) is a trivial task for any modern computer (much easier than, say, surfing web pages with pictures on them). My Foobar2000 CPU-usage is usually 0-1% of CPU resources (according to the Task Manager).
Of course, a quad-core extreme edition Intel processor w/ 8GB of fast RAM, fast hard drives, a nice motherboard, 1GB video card, etc. doesn't hurt. :-)
Scrith wrote:
> > "I continue to be mystified by all the attempts to minimize system usage and resources in order to get "high quality" audio around here - audio playback . . . is a trivial task for any modern computer."
Audio playback is indeed a trivial task for a modern computer but that is far from the whole story. The optimal presentation of the processed data to the outside world is much less trivial.
As we all know, hearing is sensitive to remarkably small timing errors in the digital reproduction of music or, in the jargon, to “jitter”. (That the term is as much abused as used in audiophile circles is unfortunate but it is nevertheless clearly defined and well understood by audio engineers.)
Judging by posts on this list, it seems that many do not understand that the final stage of PC audio reproduction is a 'real-time' process which, absent design measures to tackle it, is especially prone to “jitter”.
The process cannot be optimised by throwing computer power at it. On the contrary, theory and experience suggest that minimising, not maximising, data processing power is a more fruitful route.
The designer (amateur or professional) is attempting to handle signals that require the rigour of instrumentation electronics in the worst possible environment - inside a desktop computer.
A major cause of the degradation of the audio signal is the electrical noise in the computer. One obvious way to address this is to select quality components and to configure or modify them to minimise the noise they generate.
A related approach, a feature of the cMP2 project, is to dedicate a small and relatively low-cost computer exclusively to playing music and to ‘underclock’ it to the slowest speed at which it can perform satisfactorily. This reduces power draw and thus noise pollution.
It is difficult for amateurs to quantify results but there are enough subjective reports to give both approaches credibility.
Another major source of timing errors is (for lack of a better term though one is certainly needed) “operating system jitter”. This reflects the fact that, while audtion is a real-time process, computer operating systems are not.
This is obvious at the macro level: at the drop of a hat, Windows does things when you least want it to: e.g., when an AV update slows everything down.
In extreme cases, the listener experiences “drop outs” - detectable breaks in the audio stream. No one would dispute the need either to have enough computing power to avoid the problem (as Scrith recommends) or to reallocate resources to tackle it.
Optimisation procedures similar to those recommended for cMP2 (though less thorough) are described on the pro-audio web sites and are commonplace when running any processor-intensive software.
Increasing computing power may avoid breaks in the data stream but it is inevitably accompanied by increased noise pollution and thus tends to be self-defeating: it improves the availability of music data but not necessarily its integrity.
For “high-end” reproduction, a more fruitful approach is to tackle OS “jitter” where it is less obvious but probably more critical - at the micro level.
The problem is well expressed in the documentation for the useful Windows utility DPC Latency Checker:
“Processing of streaming data in real-time is a very challenging task for Windows-based applications and device drivers. This is because by design Windows is not a real-time operating system. There is no guarantee that certain (periodic) actions can be executed in a timely manner.”
(See: http://www.thesycon.de/dpclat for the full text.)
We can squabble at leisure over the relative merits of the cMP2 project and of various Linux setups but both adopt this approach and report good results.
It can be demonstrated (and, to a limited degree, quantified) using DPC Latency Checker whilst successively performing cMP2’s optimisation steps.
The documentation for cMP2 reports jitter levels measurements which, while incomplete and thus inconclusive, nevertheless offer strong support for the approach.
In summary, the argument is that suppressing electronic noise and reducing the response time of an operating system to its core task both significantly improve the quality of computer-based music reproduction but that throwing computing power at it is less fruitful.
If the user's aim is simply to keep Foobar busy while doing other things, then fair enough.
But if the aim is to achieve the best possible reproduction of music from a computer, it is possible to do a great deal better with a slightly unorthodox approach.
Critics are free to deny the extent of the claims made by proponents of the techniques described here or to question the merit of some of the measures they propose.
However, there is nothing especially esoteric and even less that is mystifying about what they are doing. True, it needs good software and some innovative thinking (who'd think of using a Fit-PC to play music?) but the rest is basic electronics. No more.
Best
Dave
Of course, a quad-core extreme edition Intel processor w/ 8GB of fast RAM, fast hard drives, a nice motherboard, 1GB video card, etc. doesn't hurt. :-)
Wow!!!
You got that right!!!!!
Karma Means Never Having To Say You're Sorry
My new rig only uses 2 gigs of Corsair 800 Mhz ram currently. I'll be upgrading to Vista Ultimate 64 bit soon. I will purchase two more matched pairs of Corsair 800Mhz and install before the transition. I'll report back as to sonics,Alan
Vista Ultimate 32bit/Diamond XS Dac/ Sterovox coaxial line in to Insignia Amp/Cambridge SoundWorks& Infinity RS 1001 Speakers
Or it all sounds the same, as usual? Which, BTW, wouldn't be surprising - the less computer is optimized (which includes huge quantities of memory), the more differences are masked by all kinds of interference.
does the 2 gig version sound as good, or at least close? how about the LL versions, vs. the UL.
why is this sh%t always so hard to find?????
1. HyperX you mentioned is not the best HyperX for our purposes - the one that really shines in my system is ultra-low latency 1Gb stick, spec'd by manufacturer at 3-3-3-10. It was compared to Crucial, Corsair XMS and some others - head and shoulders above everything.2. MOST IMPORTANT - 3-3-3-8 is just plain wrong, recommended or otherwise, especially for your memory, spec'd at 5-5-5-15 (see link below).
So, what you really want to do to improve on Corsair is to buy a stick of KHX6400D2UL/1G, and set it to 3-3-3-10 (sounds best in my system).
Edits: 01/21/09
Although the Kingston webpage lists specs for the UL memory, it was discontinued around February of 2008.I chased EVERYWHERE on the net to find it, and eventually talked with the sales people at Kingston... sales volume was to low, and the UL versions of HyperX were no longer offered for sale.
Now if you can find some... somewhere... you're a better Sherlock than I.
Hope this helps.
Cheers,
GrantThat's not a Toy... IT'S A TOOL !!
Edits: 01/22/09
Now that the Kingston 256MB ValueRAM has been crowned by cics, I think the best way to obtain a Kingston UL would be to acquire a used one from the inmates who made the switch. Start asking!
If there are just 10 varaiables in an audio PC to set for optimal performance, then there are 10 factorial ie 10x9x8-----1 combinations.
In reality there are probably 50 factorial or more combinations.
Who has the energy to try?
Magic "audiophile approved" memory DIMMs (at HUGE price premiums) from the usual snake oil vendors. ;-)
t
usual snake oil vendors. ;-)
Like the ultra high end CD player? Yes of course your option is done to 1 factorial!
Care to name a few?
Karma Means Never Having To Say You're Sorry
Care to name a few?
No. I'll refrain as these posts are ALWAYS controversial, often lead to emotionally charged threads, and some people being censored or banned.
.
Karma Means Never Having To Say You're Sorry
t
Post a Followup:
| FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
||||||||||||||
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
