IPB

Welcome Guest ( Log In | Register )

> Upload forum rules

- No over 30 sec clips of copyrighted music. Cite properly and never more than necessary for the discussion.


- No copyrighted software without permission.


- Click here for complete Hydrogenaudio Terms of Service

2 Pages V   1 2 >  
Reply to this topicStart new topic
Busy PC vs idle PC, different RMAA results
bennetng
post Apr 19 2013, 06:04
Post #1





Group: Members
Posts: 244
Joined: 22-December 05
Member No.: 26587



Running RMAA 4 times to average the errors
Attached File(s)
Attached File  idle_vs_busy.7z ( 64.91K ) Number of downloads: 255
 
Go to the top of the page
+Quote Post
phofman
post Apr 19 2013, 07:58
Post #2





Group: Members
Posts: 332
Joined: 14-February 12
Member No.: 97162



Interesting. And that is why audiophiles strive for the smallest buffers (i.e. much higher load), jplay author claims the sonic advantage of loading the CPU core to 100% even by using empty loops, ... These measurements make sense.

Please could you specify the load conditions? Load level, how it was reached, etc. Thanks.
Go to the top of the page
+Quote Post
skamp
post Apr 19 2013, 08:17
Post #3





Group: Developer
Posts: 1454
Joined: 4-May 04
From: France
Member No.: 13875



Both waaaaay beyond audibility.


--------------------
See my profile for measurements, tools and recommendations.
Go to the top of the page
+Quote Post
phofman
post Apr 19 2013, 08:29
Post #4





Group: Members
Posts: 332
Joined: 14-February 12
Member No.: 97162



QUOTE (skamp @ Apr 19 2013, 09:17) *
Both waaaaay beyond audibility.


I know. Yet it is a measurement, hard data. Good to have.
Go to the top of the page
+Quote Post
Nessuno
post Apr 19 2013, 09:04
Post #5





Group: Members
Posts: 423
Joined: 16-December 10
From: Palermo
Member No.: 86562



QUOTE (phofman @ Apr 19 2013, 08:58) *
Interesting. And that is why audiophiles strive for the smallest buffers (i.e. much higher load), jplay author claims the sonic advantage of loading the CPU core to 100% even by using empty loops, ... These measurements make sense.

Please could you specify the load conditions? Load level, how it was reached, etc. Thanks.

Just hair splitting: has someone ever measured the influence of the increased mass storage reading and bus/USB/network etc... activity required to deal with bigger files, like in uncompressed wav against FLAC or lossy?
Normally people fail to understand that a computer is not only made of a CPU and that decoding a compact lossy file is likely to put far less stress and waste less power in a modern system considered as a whole, than fetching and moving around larger files.

Anyway, nothing that could have the slightest impact on sound quality, of course, especially if the actual D/A is not on-board.


--------------------
... I live by long distance.
Go to the top of the page
+Quote Post
Roseval
post Apr 19 2013, 09:15
Post #6





Group: Members
Posts: 496
Joined: 26-March 08
Member No.: 52303



For those who like this type of measurements: http://archimago.blogspot.ca/2013/04/measu...rvey-apple.html


--------------------
TheWellTemperedComputer.com
Go to the top of the page
+Quote Post
phofman
post Apr 19 2013, 09:44
Post #7





Group: Members
Posts: 332
Joined: 14-February 12
Member No.: 97162



QUOTE (Roseval @ Apr 19 2013, 10:15) *
For those who like this type of measurements: http://archimago.blogspot.ca/2013/04/measu...rvey-apple.html



Very interesting. Unfortunately some of the conclusions are a bit hairy based on the test conditions - USB adaptive and async used different hardware thus drawing conclusions about the USB transfer modes is a bit shaky, IMO. Had the tester used the same HW with switchable firmware (not that I know of any smile.gif ), it would have been more valid.
Go to the top of the page
+Quote Post
stv014
post Apr 19 2013, 09:45
Post #8





Group: Members
Posts: 48
Joined: 26-February 13
Member No.: 106902



QUOTE (phofman @ Apr 19 2013, 08:58) *
Interesting. And that is why audiophiles strive for the smallest buffers (i.e. much higher load)


With the onboard ALC887 codec in my desktop PC, decreasing the buffer size actually makes the sound quality worse. That is because the increased load at each buffer creates a low level click-like artifact in the output, and with shorter buffers, these become more frequent. When the buffer size is really small (only a few ms total), there is a plainly audible tone at the frequency of the audio interrupts.

However, a good quality sound card or external interface should not have such issues to an extent that is even close to being audible, unless there are ground loops or other problems. In these tests of a sound card in the same machine, there was only a minor difference between low and high system load, and well below the threshold of audibility under realistic conditions.
Go to the top of the page
+Quote Post
phofman
post Apr 19 2013, 10:03
Post #9





Group: Members
Posts: 332
Joined: 14-February 12
Member No.: 97162



QUOTE (stv014 @ Apr 19 2013, 10:45) *
QUOTE (phofman @ Apr 19 2013, 08:58) *
Interesting. And that is why audiophiles strive for the smallest buffers (i.e. much higher load)


With the onboard ALC887 codec in my desktop PC, decreasing the buffer size actually makes the sound quality worse. That is because the increased load at each buffer creates a low level click-like artifact in the output, and with shorter buffers, these become more frequent. When the buffer size is really small (only a few ms total), there is a plainly audible tone at the frequency of the audio interrupts.

However, a good quality sound card or external interface should not have such issues to an extent that is even close to being audible, unless there are ground loops or other problems. In these tests of a sound card in the same machine, there was only a minor difference between low and high system load, and well below the threshold of audibility under realistic conditions.


Yes, technically all this makes perfect sense. But just browse to "audiophile" forums and read the claims - cplay, jplay, squeezebox touch mods. Of course you will never ever see a DBT verification. But it sounds good: "low latency = low jitter", doesn't it? smile.gif
Go to the top of the page
+Quote Post
bennetng
post Apr 19 2013, 10:11
Post #10





Group: Members
Posts: 244
Joined: 22-December 05
Member No.: 26587



QUOTE (phofman @ Apr 19 2013, 14:58) *
Interesting. And that is why audiophiles strive for the smallest buffers (i.e. much higher load), jplay author claims the sonic advantage of loading the CPU core to 100% even by using empty loops, ... These measurements make sense.

Please could you specify the load conditions? Load level, how it was reached, etc. Thanks.


To make the test more practical, I changed my mind to use synthetic benchmark but using some real programs to test. I did not check if the GPU/CPU are fully loaded but should be more than 25% in average.

I was running this 3d demo
http://www.pouet.net/prod.php?which=30244

and using SVP frame interpolation (CPU/GPU intensive)
www.svp-team.com

to play a 720P video using ffdshow software H264 decoder while running RMAA.

===========
I just bought the new card several days ago because the old card (X-Fi XtremeMusic)'s headphones out is damaged and produced statics. The new card (X-Fi Titanium HD) has better specs but the most stupid thing is I need to physically unplug the headphones to unmute the RCA line out, which is very frustrating.

I also test MME (RMAA as player) vs ASIO (foobar as player) and found that ASIO is slightly noiser (around -117 to -118dB, all sample rates). I did not do the tests intentionally, just because I need to check 88.2k results but Windows mixer doesn't have this option (44/48/96 only)so I had to use ASIO.

===========
Forget about jplay, it made my PC sluggish as hell.

This post has been edited by bennetng: Apr 19 2013, 10:14
Go to the top of the page
+Quote Post
Kees de Visser
post Apr 19 2013, 11:01
Post #11





Group: Members
Posts: 737
Joined: 22-May 05
From: France
Member No.: 22220



QUOTE (phofman @ Apr 19 2013, 08:29) *
I know. Yet it is a measurement, hard data. Good to have.
I don't know RMAA very well, but since the ADC is part of the same device, isn't it influenced by idle/busy differences as well ? Not sure how important this is, but I'd prefer to see measurements with an external ADC if possible.
Go to the top of the page
+Quote Post
Brand
post Apr 19 2013, 12:46
Post #12





Group: Members
Posts: 321
Joined: 27-November 09
Member No.: 75355



QUOTE (phofman @ Apr 19 2013, 07:58) *
Interesting. And that is why audiophiles strive for the smallest buffers (i.e. much higher load), jplay author claims the sonic advantage of loading the CPU core to 100% even by using empty loops, ... These measurements make sense.

I don't know if you were perhaps being sarcastic, but idle performed better in OPs test.


My first guess is it has something to do with power/voltages fluctuating when the PSU is under stress.
Would be interesting to see if it's related only to CPU load or also GPU load etc.
Go to the top of the page
+Quote Post
db1989
post Apr 19 2013, 12:48
Post #13





Group: Super Moderator
Posts: 5275
Joined: 23-June 06
Member No.: 32180



QUOTE (phofman @ Apr 19 2013, 07:58) *
Interesting. And that is why audiophiles strive for the smallest buffers (i.e. much higher load), jplay author claims the sonic advantage of loading the CPU core to 100% even by using empty loops, ... These measurements make sense.
Although subsequent discussion debunks it, this passage could be taken to imply that any measurements, however inaudible, lend credence to madcap ideas about how to supposedly gain better quality. Such ideas are likely to be total nonsense, and they aren’t worth the space they occupy until their purveyors provide systemically reliable evidence for their existence. That still hasn’t happened yet! How bizarre. Given that many of the proponents of such concepts like to charge high premiums for the privilege of access to them, one would think they would want to maximise their revenue by providing solid evidence; then they could gain additional custom from some sections of the buying community who place more emphasis on evidence. Again, it’s so weird that no proof of all these claims has been forthcoming. rolleyes.gif

Small measurements in no way imply that implausible and often expensive methods supposedly based on said measurements have any sense whatsoever.
Go to the top of the page
+Quote Post
bennetng
post Apr 19 2013, 13:48
Post #14





Group: Members
Posts: 244
Joined: 22-December 05
Member No.: 26587



The original purpose of this post is not to test busy vs idle, I uploaded the results here because I suspected I used a wrong cable to perform RMAA tests.

http://www.hydrogenaudio.org/forums/index....howtopic=100440

My previous soundcard also produced differences under load, but the differences are only 2-3dBs (-99 vs -102), I guess soundcards with lower noise floor may produce bigger differences, just guessing.

NO audible differences absolutely. Please don't quote my posts in other placebophile-based forums. Thanks.
Go to the top of the page
+Quote Post
Wombat
post Apr 19 2013, 15:10
Post #15





Group: Members
Posts: 1121
Joined: 7-October 01
Member No.: 235



My different nvidia geraphic cards introduce noise into different mainboards since several years now when you push them. Loudest one was the old GTX260. It even produced loud mechanical whining coming from the coils of the voltage regulation on the GPU board.
I guess all modern graphic cards that need enough power and therefore have high current regulations build onto their board will introduce noise when pushed hard.
I canīt produce that noise with loading my CPU.
Go to the top of the page
+Quote Post
saratoga
post Apr 19 2013, 18:20
Post #16





Group: Members
Posts: 5167
Joined: 2-September 02
Member No.: 3264



-120 dB on a typical sound card is about 500nV.

Thats a high impedance line wrapped around some electronics. Since theres no real load, you can certainly get nanovolts worth of voltage between two wires. I can do that with my socks on a carpet too smile.gif

Put a pair of headphones in parallel to sink stray electrons and this won't happen.
Go to the top of the page
+Quote Post
bennetng
post Apr 19 2013, 18:22
Post #17





Group: Members
Posts: 244
Joined: 22-December 05
Member No.: 26587



QUOTE (Brand @ Apr 19 2013, 19:46) *
QUOTE (phofman @ Apr 19 2013, 07:58) *
Interesting. And that is why audiophiles strive for the smallest buffers (i.e. much higher load), jplay author claims the sonic advantage of loading the CPU core to 100% even by using empty loops, ... These measurements make sense.

I don't know if you were perhaps being sarcastic, but idle performed better in OPs test.


My first guess is it has something to do with power/voltages fluctuating when the PSU is under stress.
Would be interesting to see if it's related only to CPU load or also GPU load etc.


Just tested. GPU full load (using ATI tray tool) scored -110dB, CPU full load (OCCT) scored -117dB. Both full load also scored -117dB because CPU bottlenecked the GPU, the 3D renderer in ATI tray tool dropped to lower than 1 fps.

And I said ASIO scored worse in the previous post and I was wrong, the differences are caused by different recording levels between recording software (RMAA and Reaper). The lowest latency (1ms) scored same as MME and DS, indicating a good ASIO driver without crackling and dropouts.
Go to the top of the page
+Quote Post
phofman
post Apr 19 2013, 18:30
Post #18





Group: Members
Posts: 332
Joined: 14-February 12
Member No.: 97162



QUOTE (Brand @ Apr 19 2013, 13:46) *
I don't know if you were perhaps being sarcastic, but idle performed better in OPs test.


I am almost always sarcastic smile.gif The results are aligned with common sense. Higher load produces more noise on supply lines. But try to explain that to a hard-core audiophile who simply knows what he hears smile.gif
Go to the top of the page
+Quote Post
greynol
post Apr 19 2013, 18:32
Post #19





Group: Super Moderator
Posts: 10341
Joined: 1-April 04
From: San Francisco
Member No.: 13167



Those looking for evidence to support typical placebophile claims (or otherwise) should pay close attention to this latest wrinkle. Test conditions are critical.


--------------------
Your eyes cannot hear.
Go to the top of the page
+Quote Post
Wombat
post Apr 19 2013, 19:08
Post #20





Group: Members
Posts: 1121
Joined: 7-October 01
Member No.: 235



Here i recorded my EVGA GTX-660Ti kicking in at ~sec. 10 stressed with the EVGA OC Scanner tool. This is with a X-Fi Xtreme Music running onto an ASrock z77 extreme6, Superflower 550W Golden Green PSU.

Attached File(s)
Attached File  GTX660TI_noise.flac ( 1.13MB ) Number of downloads: 101
 
Go to the top of the page
+Quote Post
bennetng
post Apr 19 2013, 19:17
Post #21





Group: Members
Posts: 244
Joined: 22-December 05
Member No.: 26587



QUOTE (Wombat @ Apr 20 2013, 02:08) *
Here i recorded my EVGA GTX-660Ti kicking in at ~sec. 10 stressed with the EVGA OC Scanner tool. This is with a X-Fi Xtreme Music running onto an ASrock z77 extreme6, Superflower 550W Golden Green PSU.


My GPU is much cheaper than yours (HD5750), it is quite reasonable that higher-end GPUs produce more noise.

PS: both my CPU and GPU are overclocked.
Go to the top of the page
+Quote Post
Wombat
post Apr 19 2013, 19:24
Post #22





Group: Members
Posts: 1121
Joined: 7-October 01
Member No.: 235



Here also, i canīt remember having something running not overclocked since my 386DX-33@40 smile.gif
No matter how hard i push/clock the CPU i canīt produce any noise that influences playback quality to my ears. Pushing the GPU does as you may hear in that sample i provided. I should have mentioned that my 6660Ti is the SC version already. I only clocked the RAM on it a bit higher. One day when i am bored i have to dig around if the induced noise comes over the mainboard and produces the noise or if the graphics card itself induces the noise to the X-Fi card over air directly.
Go to the top of the page
+Quote Post
bennetng
post Apr 20 2013, 12:02
Post #23





Group: Members
Posts: 244
Joined: 22-December 05
Member No.: 26587



QUOTE (Wombat @ Apr 20 2013, 02:24) *
One day when i am bored i have to dig around if the induced noise comes over the mainboard and produces the noise or if the graphics card itself induces the noise to the X-Fi card over air directly.


(Not in English)
http://www.dearhoney.idv.tw/?p=71
This person added a metal casing on his Audigy and tested its line out with a CardDeluxe and found that it has a 2dB reduction in noise (upper results=before, lower results=after)

If the noise are transmitted from motherboard then USB/firewire/PCI interface with breakout box could be affected as well because they are all physically connected with the motherboard. It is often said external interfaces are usually better but in the same price range I often see otherwise. For example, I never see actual reports that external interfaces under $250 could have noise below -115dB in a loopback test.

15 RMAA reports including M-Audio, Echo, RME, Lynx etc:
http://arkiv.idg.se/fileArchive/studio/24/...51020154428.zip

Go to the top of the page
+Quote Post
Arnold B. Kruege...
post Apr 23 2013, 12:29
Post #24





Group: Members
Posts: 4488
Joined: 29-October 08
From: USA, 48236
Member No.: 61311



QUOTE (bennetng @ Apr 20 2013, 07:02) *
QUOTE (Wombat @ Apr 20 2013, 02:24) *
One day when i am bored i have to dig around if the induced noise comes over the mainboard and produces the noise or if the graphics card itself induces the noise to the X-Fi card over air directly.


(Not in English)
http://www.dearhoney.idv.tw/?p=71
This person added a metal casing on his Audigy and tested its line out with a CardDeluxe and found that it has a 2dB reduction in noise (upper results=before, lower results=after)


IOW, not worth the trouble.

QUOTE
If the noise are transmitted from motherboard then USB/firewire/PCI interface with breakout box could be affected as well because they are all physically connected with the motherboard.


The mythology says that PC's are unusually horrible places to run audio gear. The fact is that any piece of audio gear that has digital and analog signals in it is a mixed-signal application. Mixed signals is like a room that the designer enters and as long as he follows the rules he can do just about anything he needs to do that is reasonable. It is the first TTL line running millimeters from your analog signal that sets the tone for the rest of your day. Adding more doesn't make that much difference.

Nobody has figured out how to build a DAC that doesn't put digital signals in intimate contact with analog signals, and I don't think anybody ever will.

QUOTE
It is often said external interfaces are usually better but in the same price range I often see otherwise. For example, I never see actual reports that external interfaces under $250 could have noise below -115dB in a loopback test.


I can fearlessly predict that we will see such things in a few years at the latest, and they will come in both internal and external models. It's all about chip prices. 10 years ago 115-120 dB chips cost maybe $20 or more. This year some good ones cost less than $5. If a chip producer wants to get a premium price for his chip this year, he needs to have numbers >120-125 dB. However, the time/price curve of that chip seems predictable. 5-10 years from now it will probably be under $5.

The end point of the dynamic range game is probably set by utility. Once gear gets beyond 100 dB dynamic range, it is hard to come up with practical (audible) reasons to go further. Chips in the 105+ dB range make designing 100 dB gear easy enough. We are there.

QUOTE
15 RMAA reports including M-Audio, Echo, RME, Lynx etc:
http://arkiv.idg.se/fileArchive/studio/24/...51020154428.zip


Not the best choice of interfaces, but some kind of an indication.

This post has been edited by Arnold B. Krueger: Apr 23 2013, 12:30
Go to the top of the page
+Quote Post
Wombat
post Jan 12 2014, 21:27
Post #25





Group: Members
Posts: 1121
Joined: 7-October 01
Member No.: 235



For completeness i recorded the noise with an EVGA GTX-670 in the exact same system as the GTX-660Ti sample above. This one sounds much quieter to the ear because it doesn't create these obvious spikes all over the place as the GTX-660Ti does. Must be higher quality parts or QC.



This post has been edited by Wombat: Jan 12 2014, 21:36
Attached thumbnail(s)
Attached Image
 

Attached File(s)
Attached File  GTX670_noise.flac ( 1.17MB ) Number of downloads: 49
 
Go to the top of the page
+Quote Post

2 Pages V   1 2 >
Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 27th December 2014 - 15:35