|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
66.241.132.98
In Reply to: help with components for music server... posted by Ron on September 29, 2006 at 05:24:12:
Ron,You have not mentioned what operation system you will be using. If you plan to use Windows 2K or XP, I strongly advise against Epia mobo and Crusoe 1GHz fanless - there is not enough horsepower.
I have built a couple of music-only PCs and here is what I have found.
Windows 2K and XP are resource hogs. On Windows platform the speed of processor DOES count for quality of music reproduction. AMD Athlon64 X2 3800 sounds signifficantly better than AMD Athlon 64 3200.
I have shut down all of Windows unecessary processes and services (more than 40). Even then, when only music player is active. on average first processor is 5-6% active and second around 25%.
I am planning to UNDERclock processor by 25% and thus cut on heat emission. Less heat, less fan noise.
Damping a case, installing additional internal bracing and using a butcher's maple block can cut a lot of noise.
I have seen a teeny weeny HP media desktop in Best Buy for Canadian $600. It is pretty silent and you'll be hard pressed to build it yourself for that money.If you want to use Epia, there are solutions with DOS and DOS based players or Linux where 1GHz is enough. I have no experience with this and do not know it first hand.
Follow Ups:
no comment.
***On Windows platform the speed of processor DOES count for quality of music reproduction. AMD Athlon64 X2 3800 sounds signifficantly better than AMD Athlon 64 3200...I am planning to UNDERclock processor by 25% and thus cut on heat emission. Less heat, less fan noise.***bobp -
This is very interesting. I have read about the notion that the CPU speed (as well as memory speed/latency) affects the sound quality, and have tried a few experiments of my own (inconclusive so far). But I'm confused by what you say. You say the faster processor sounds better, so why are you going to underclock it? I mean I understand what you said about heat and fan noise, but if it affects sound quality, why do it?
Edward, I have read somewhere that clock of 2MHz and over is causing a lot of EMI noise, and 1.8GH should be sweet spot. It translates to
slowing down the CPU by 10-12%.
x2 3800 reduced by 10-20% will be still signifficantly more powerful tnan one ply 3200.When the cold weather arrives I will start playing to find the best ratio. Also, I will try to find X2 4800 to see how it will perform.
This is uncharted territory, and I would appreciate any information I can get. Also, I will share my findings.
***I have read somewhere that clock of 2MHz and over is causing a lot of EMI noise, and 1.8GH should be sweet spot.***I assume you meant 2GHz (not MHz) and yes I did read this also. I have an Intel 3.2GHz processor and have experimented with speed reduction. I can't change the multiplier with my BIOS, but I can change the clock refquency (FSB). I reduced it from 200MHz to 166 and also 100 (which changes the CPU speed to 2.6GHz and 1.6) I run foobar with SRC (at 24/96) and 1.6 was too slow to play without stuttering. When reducing the clock to 166MHz I can also change the timings of my memory. The SPD of my memory (DDR400) applies latency timings of 2.5-3-3-8 (when run at 200MHz), but when lowered to 166 I can change the timings to 2-2-2-6. I think this makes a difference in sound quality also.
Not many people believe these things affect sound quality, and I'm not sure yet myself. But I'm trying to keep an open mind and don't mind experimenting.
Presto once asked the question; "If all this emi is such a big deal, why doesn't it affect video?"Any ideas as to why sound would be impacted and not video too?
***Presto once asked the question; "If all this emi is such a big deal, why doesn't it affect video?"
Any ideas as to why sound would be impacted and not video too?***Dawnrazor -
I'm afraid that's way beyond the scope of my technological knowledge, so I'll have to defer to someone with more expertise. I'm just an enthusiast who's experimenting. But allow me to speculate if you will.The sound differences I've experienced so far are like comparing music with a SNR of 80db to music with 90db SNR. (I didn't actually make any measurements, I'm just throwing out numbers for the sake of example) I just assumed EMI was at work here, and I was just able to lower the noise floor. And of course, this is just based on my system. Someone with completely different computer parts may hear no difference whatsoever.
And who says it doesn't affect video. I mean assuming, of course, we are talking about the SNR. Of course SNR affects video. I had to call the cable guy one time because my video signal was all jumbled up. He checked the SNR and it was down to 26db. There was a problem with the transformer down the street and he said I was getting electrical interference and, after he fixed it, the SNR was back to a normal 34db with an acceptable picture. I have no idea how all this works or if it has any correlation to what you are asking, but maybe we just assume EMI is not affecting video playback on our computers because the SNR is high enough for us to not notice a difference. Maybe we are more sensitive to hearing these kinds of differences.
But truly, maybe I don't know what the hell I'm talking about and this is all pure speculation.
It would sure be nice to get a definitive and measurable response about what exactly does affect the sound when playing from a computer. I mean first we are presented with over a dozen different ASIO foobar plugins (all of which sound different than the last) and now have you read what LessLoss is saying:
Ed,YOu raise some interesting points here and so does the article.
I don't doubt for a minute that many of this stuff IS audible, especially different settings with in the computer.
Maybe the video is affected, but if it is no one ever notices it or talks about it, and it since my soundcard sound great hanging out in all this emi and powersupply hash, it makes me wonder if it is as big an issue as everyone claims, or if there are ways to effectively treat it.
From the sounds of it, I am glad I am not messing around with an external dac and usb.
Dear god! What music player are you running that generates 5-6% activity on one processor and 25% on another? Seems outrageous to me. What are you using to measure activity?My little SFF PC is running XP Pro on an EPIA SP8000 board, and does music fine, even when its supplying stuff to remote squeezeboxes. I'm using a NAS RAID for the music library and a USB audio output device, but that shouldn't make that much of a difference either way.
I am running JRiver MC11 and apple lossless. It seems that the more resources you have, the more window uses. On the same Mobo, same settings, but AMD 64 3200 plugged in, MC11 uses only 5-6%. what the second CPU in X2 3800 is churning is beyond me.
B,THis has not been my experience at all.
I am running an epia 1ghz with windows XP pro dedicated to music only, it is not even hooked up to the web.
It works great. THe only problems I have had in terms of it being underpowered have to do with using the Secret Rabbit upsampler in Foobar (other upsamplers work fine) and crossover slopes beyond 24 db.
But all other normal audio only functions work fine. Maybe it is the fact that I am not sending signal out to a dac, but using a Lynx 2b card, but I have had no problems whatsoever. And I haven't really stopped all the processes I can, so there is room for improvement.
FInally, in my case it was more important to have it be fanless as any reduction in performance (if indeed that happens at all) would be more than offset by the noise created by a fan.
Downrazor,MY PC is not networked and does only music playback.
MC11 works great on AMD 64 3200, but when I replace it with X2 3800 it works much better. The audible difference is comparable to switching cables or using two different CD players, not subtle.
Same machine, same settings, only different processor.I have compared it with Apia and the sound is hands down better. This is not simpkle comparrison because I am comparing two different PCs, two different hardware and driver environments.
Just my experience, with my equipment, my AC, my EMI/RF environment, my furniture, my room. I do not claim this is a general rule.
Dawnrazor,What exactly is upsampling, and does it have anything to do with just playing music using foobar(connected to my home stereo system)?
Also, I would probably end up getting a DAC in the future. Do you think the DAC would affect anything?
thanks,
SRC (=secret rabbit coder or something) is an optional upsampler in foobar that is supposed to be pretty good. Redbook CD format uses a 16/44.1 format--16 bits of information to encode the music, with samples taken 44,100 times per second.You would have a much more accurate digital rendition of the analog signal if you had better resolution for encoding the music (e.g., 24 bits) and took samples more often (e.g., 192,000 times per second). There are a fair number of pieces of pro gear that run, in fact, at 24/192 or at 24/96.
But, upsampling doesn't get you all the way there, because you are working with the 16/44.1 copy of the original, not the original. Still, good algorithms for upsamplers can (to my humble ears) make a difference. I haven't used SRC in foobar, since I've got a standalone dCS Purcell. An upsampler is going to require a heftier processor, however, since you are doing a lot of real-time data manipulation.
A good DAC will help the sound. I run a number of systems using both soundcard (E-MU 1212m)-> digital coax-> DAC (Theta) and USB based audio (PC-> USB-> Waveterminal U24-> digital coax-> DAC (dCS). The sound is tremendous, and I'm a great believer in buying a good DAC. The goal of PC audio then becomes reducing the amount of jitter on the otherwise bit-perfect output of your computer.
e-desilva,Most of what you said is Greek to me. I don't understand the 16/44.1 or 24/192 thing.
Ron,I bet it is not all that important.
But I'll try to put it in terms that you might have a better grasp of.
DO you know how if you have more pixels, video has more resolution?
It is kind of like that with audio. The more bits and the faster the sampling frequency, the better the resolution. SACD and DVD audio can have better sound quality due to this fact.
A computer can upsample (adding pixels say) to get better resolution.
Upscalers in video are similiar, adding more lines of resolution. BUt to Ed's point, those lines are fabricated and while they look good, it would look even better if the TV show was RECORDED that way.
SOme like what upsampling a 16 bit 44.1k cd to say 24 bit 96k does to the sound. Others think that one can't improve on the original recording. I think everyone agrees that something recorded at 24/96 will sound better than the same recording at 16/44.1.
So, all I was saying is that with the epia, regular upsamplers work fine, but the Secret Rabbit code one is too processor intensive to pull off higher rates.
Excellent explanation, thank you. So here's another question. Are origianl cd's recorded in different formats, or only 16/44.1?Does anything change when you rip a cd?(All my music is in .flac)
thanks again for the excellent explanation.
16/44.1 is the redbook standard.So all cds are 16/44.1.
If one rips properly, there is no difference.
FWIW, with the stock upsampler in Foobar, I go from a 2-7% usage to 3-13% CPU usage. S0, 32/96 is possible with the epia if your hard ware supports it. Just not with the secret rabbit.
That said, unless you plan to do upsampling with the secret rabbit upsampler or 60db crossovers, there is nothing n my experience to say that it won't work on the epia you are looking at especially if you are dedicating the computer to audio.
ok i'm just curious now. Am I understanding right then that the master copy(the one they use to make the printings) is actually 24/96? It seems wierd that if the master is 24/96 that they are recorded in 16/44.1. Like I said, just curious....what case did you go with? I'm looking at the morex 2699 case. Seems the cheapest one I can find..
thanks again,
Nah, when they say "mastering," they are talking about the studio process of mucking around with the originally recorded tracks, downmixing to stereo from multitrack, etc. While digital studios may keep most of that in the 24/96 or 24/192 format, it all gets downsampled to 16/44.1 before getting dumped on a CD. So, there isn't a "master copy" of a CD in a higher resolution format... When you find those CDs that say, for example "Superbit 22 bit format", its saying that the CD was mastered using 22 bit "words." You aren't getting some odd 22/44.1 format on what you are buying--its still 16/44.1.Dawnrazor-What kind of DACs deal with 32 bit formats? Are you sure you aren't talking about upsampling to 24bits? 24/96 and 24/192 are fairly common, but I haven't seen 32/96.
ED,The lynx 2 is one of those "prosumer cards". It supports up to 32/200. Although pricey (why is it that audiophiles will easily spend $2-3K or more on a cdp, but get cheap as hell when it comes to PC audio balking at a $1k sound card...and expect it to sound better tahn their 3k CDP?) I figured that it handled high enough sampling rates and frequencies, to be future proof for a long while. I can always get another computer if I need to go to those levels, or a new format comes along.
BUt The max I can do is 32/96 since I haven't found an upsampler that goes higher. Well, I CAN create files with db power amp that are 24/192 and then the computer doesn't have to do the heavy lifting realtime, but that is a ton of space, and I am not sure it is worth the time effort and money to do so.
Foobar has a setting to output 32 bits, and I usually have it set to that.
Do you know of an upsampler that goes to 192, or 176?
So are you using the analog or digital outputs of the Lynx? Guess I was wondering what DAC could chew a 32 bit input. I've got a hardware upsample (dCS Purcell), so I haven't messed with the software upsampling in foobar...
BOTH.Analog to the amps, and for head phone listening I run out to a Birdland Dac...which apparently has no problem with 32 bits.
I am assuming it is putting out 32 bits as foobar reports it, and the lynx mixer reports this as well, but I don't know for certain.
***Do you know of an upsampler that goes to 192, or 176?***Dawnrazor -
I thought you said you used "Secret Rabbit" (SRC) resampler in foobar. I'm pretty sure you can change it to 192 or 176. I know the preset numbers only go up to 96, but you have to manually type in the value if you are going higher. (192000 or 176400) Just be sure you stop any music that's playing before you do this. It freaks out if you do it while it's playing.Give it a try. Let us know how it works. It's going to take a lot of CPU power though.
It doesn't work, as you rightly say, it requires a ton of processing power. My little 1ghz machine can't do it at those rates. I think it got up to 88 with some stuttering using the SRC upsampler.I did just try playing a 24/192 file that I made on another computer using db poweramp. Foobar refuses to play it on version 09.3, but 08 will play it, there is just no sound.
My lynx software reports 192 for the clock rate on 08, there is just no sound.
SO maybe Foobar doesn't really support 192???
R,I just bought the Hush from logicsupply.com (unfortunately, that no longer seems to be available)...mostly because it was fanless, and looks like a hifi piece. Very pricey, but not to me...I thought it was worth it.
The files are 16/44.1. Foobar upsamples to 32/96, if I have it set to do so.
This post is made possible by the generous support of people like you and our sponsors: