Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[zoom+slomo]

This discussion may only serve to reveal my less than complete knowledge of A/V digital signal processing rather than raise questions which audibly or not impact signal quality. However, the concerns raised stem from here. https://www.google.com/amp/s/arstechnica.com/gadgets/2021/01/linus-torvalds-blames-intel-for-lack-of-ecc-ram-in-consumer-pcs/%3famp=1

The report initially describes and quotes Mr. Torvalds’ contempt for Intel’s refusal to support error correcting EEC memory which apparently impacts his interest with server hardware, computational computing and other non-home theater related tasks. But later discussed are claims, though unspecified, where consumers-presumably home users-can also suffer from non-EEC memory generated errors.

Thus, if the lack of EEC memory would leave the computer audiophile vulnerable to errors, however infrequent, in what form would they be? Would they manifest as permanent audible “clicks” during the re-digitizing of music in 24 bit audio and/or when downloaded from stores like https://www.hdtracks.com/ ?

Has anyone experienced this? If not perhaps because such errors may be even more noticeable and more frequently occurring from much more common be still respectably sounding 16 bit audio?

OTOH, would such errors be more noticeable if downloading or playing 1080p or especially 4K video, though both requiring far more digital bandwidth (e.g. bit depth?) than even 24 bit audio? However, the ears are likely more sensitive than the eyes to digital audio errors.

In any case, do AMD brand Ryzen CPUs and motherboards support ECC memory?

And which if any other motherboard brands do also?

But if so would such consumer desktop systems tend to create more heat induced fan noise than desktops with comparable Intel CPUs and motherboards performing the same kinds of work loads, such as 1080p video editing-and full length movie playbacks via https://jriver.com/overview.html ?
 

[Awesome Donkey]

I can't contribute anything else to this topic, but I can answer this...

In any case, do AMD brand Ryzen CPUs and motherboards support ECC memory?

Yes they do.

And to top it off all of the Ryzen 5000 series of CPUs are better than Intel's entire current 10th generation lineup. If you can find a Ryzen 5000 series CPU and motherboard, there's no reason to consider Intel at the moment. AMD has the performance crown right now (first time, in what, 15 years?).

[wer]

Hi Zoom.

Your question is a reasonable one, and I don't think you're trying to start a debate, but questions like this can easily devolve into one, and I have zero interest in debating issues of audibility with any audiophiles who might show up. I'm just going to briefly address some of the technical aspects of your question.

First, if you have been using a PC for any period of time, you HAVE experienced a memory error. It is a certainty. Life continued, the Earth still rotates on its axis.  It's part of life, like rain.

If you want to know what Intel CPUs support ECC memory (it's actually the memory controller) then look here:
https://ark.intel.com/content/www/us/en/ark/search/featurefilter.html?productType=873&0_StatusCodeText1=3,4&0_ECCMemory=True

I'll also say I agree with AD: AMD is putting out better products at the moment.

Don't worry about downloaded files. Data transfer protocols have data integrity checks. If you want to worry, download the track twice and run an md5 checksum on it.  Or do research on data integrity. Educating someone on the lengths computers and computer systems go to to ensure data integrity could take months. University courses are designed around it.

Finally, the effect of a single-bit memory error can range from catastrophic to totally insignificant. It is impossible to predict, because it is context dependent. If the memory error occurs in critical kernel or driver code, your computer will likely crash. If it occurs in memory used as frame buffer, one pixel on your screen will momentarily be a slightly different color, and you will never know.

For video playback, the bit error might be totally unnoticeable (that pixel is 1/128th more red than it should have been) or it might cause a bit of pixelization.

For audio, the same is true. It depends on how your audio is encoded. If a single-bit error occurs in a DSD stream, the effect will not only be inaudible, it will be unmeasurable.  If it occurs in a PCM stream, there's a small chance it might be audible depending on context and which exact bit. If it occurs in some other encoded formats, the track might not decode at all and won't play or will cause your player to crash.

But it is only hubris that makes audiophiles worry about this. Memory errors are random, unless you have defective hardware. If only 1% of your system's memory is used to store audio data, then you have only a 1% chance of the memory error actually occurring in the audio data. 99% chance it will be elsewhere. Audiophiles seem to think that because the memory holding audio is the "most important" that it will be especially susceptible to errors. Not so.  So then if the error DOES occur in audio memory, what are the chances it will be audible? Not predictable, but it's not 100% either. So multiply that by the 1%. And multiply that by the chance that the memory error will occur while you're listening in the first place.

The odds of you having an audible effect due to lack of ECC memory might be 1 in 10,000 or 1 in 1,000,000.  It is much more likely it will just crash your computer.   Seriously.

ECC memory is good. Intel preferred cost cutting. But worrying about single bit errors during audio playback is not productive or beneficial to your enjoyment of music. "Computer audiophiles" are already vulnerable to a plethora of things, both real and imagined, and single bit memory errors are not the greatest of these. The click an audiophile hears from a speck of dust that just landed on their vinyl record might be greater than the effect of a single bit error.

So "don't worry, be happy" is my advice.

[zoom+slomo]

I can't contribute anything else to this topic, but I can answer this...

Yes they do.

And to top it off all of the Ryzen 5000 series of CPUs are better than Intel's entire current 10th generation lineup. If you can find a Ryzen 5000 series CPU and motherboard, there's no reason to consider Intel at the moment. AMD has the performance crown right now (first time, in what, 15 years?).

First time in 15 years? Was it that AMD was not able to compete technologically with massive Intel or that they didn't need to because they held comfortable profit margins serving some niche market (s)?

In any case, does the lead that most of the four core Ryzen CPUs/motherboards currently have over Intel, when doing 1080p editing with software like this https://www.cyberlink.com/products/powerdirector-video-editing-software/features_en_US.html?r=1 also mean that such midsize desktops (dual 120mm fans, radiator) will likely run cooler AND quieter?

Any speculation on how Intel may turn things around in this regard whenever the Alder Lake processors and new motherboard architecture arrives? 

[zoom+slomo]

First, if you have been using a PC for any period of time, you HAVE experienced a memory error. It is a certainty. Life continued, the Earth still rotates on its axis.  It's part of life, like rain.

If you want to know what Intel CPUs support ECC memory (it's actually the memory controller) then look here:
https://ark.intel.com/content/www/us/en/ark/search/featurefilter.html?productType=873&0_StatusCodeText1=3,4&0_ECCMemory=True

I'll also say I agree with AD: AMD is putting out better products at the moment.

Don't worry about downloaded files. Data transfer protocols have data integrity checks. If you want to worry, download the track twice and run an md5 checksum on it.  Or do research on data integrity. Educating someone on the lengths computers and computer systems go to to ensure data integrity could take months. University courses are designed around it.

Finally, the effect of a single-bit memory error can range from catastrophic to totally insignificant. It is impossible to predict, because it is context dependent. If the memory error occurs in critical kernel or driver code, your computer will likely crash. If it occurs in memory used as frame buffer, one pixel on your screen will momentarily be a slightly different color, and you will never know.

For video playback, the bit error might be totally unnoticeable (that pixel is 1/128th more red than it should have been) or it might cause a bit of pixelization.

For audio, the same is true. It depends on how your audio is encoded. If a single-bit error occurs in a DSD stream, the effect will not only be inaudible, it will be unmeasurable.  If it occurs in a PCM stream, there's a small chance it might be audible depending on context and which exact bit. If it occurs in some other encoded formats, the track might not decode at all and won't play or will cause your player to crash.

But it is only hubris that makes audiophiles worry about this. Memory errors are random, unless you have defective hardware. If only 1% of your system's memory is used to store audio data, then you have only a 1% chance of the memory error actually occurring in the audio data. 99% chance it will be elsewhere. Audiophiles seem to think that because the memory holding audio is the "most important" that it will be especially susceptible to errors. Not so.  So then if the error DOES occur in audio memory, what are the chances it will be audible? Not predictable, but it's not 100% either. So multiply that by the 1%. And multiply that by the chance that the memory error will occur while you're listening in the first place.

The odds of you having an audible effect due to lack of ECC memory might be 1 in 10,000 or 1 in 1,000,000.  It is much more likely it will just crash your computer.   Seriously.

ECC memory is good. Intel preferred cost cutting. But worrying about single bit errors during audio playback is not productive or beneficial to your enjoyment of music.

Though given my limited my knowledge, I couldn't imagine a more complete and useful reply. Thank you.

But at the risk of raising questions possibly inappropriate to this forum, many enthusiasts likely lament that had digital audio arrived earlier then maybe record labels would have skipped over 16 bit CD standard in favor of 24 bit DVDs. And on this point, would you know if most of the better audiophile DACs (~ $5000 range) employ some kind of interpolation or even extrapolation schemes to predict the values of missing data in a 16 bit recording that might have been extant if the recording were a 24 bit recording?

If yes, are you familiar with any make and/or model DACs which are especially good at doing this, along with having well designed output stages and power supplies?

[wer]

Thanks for your comment.  Regarding your follow-up:

The bit depth of recordings constitutes the (possible) dynamic range: how loud vs how quiet. It also impacts signal to noise ratio.

Such interpolation as you describe would not be useful.

Sampling rate is more important for audio quality. Google sampling rate for many educational articles.

A lot of DACs can't accurately provide even 16bit resolution even today.

Regardless, you can't really enjoy music with a full 24bit dynamic range.  The sound pressure level provided by 16bits can damage your hearing.  The human ear can only hear about 20bits of dynamic range without pain. 24bits, 144db, causes instant hearing damage.

So bit depth isn't what you think it is.  Better do a bit of research: https://www.soundguys.com/audio-bit-depth-explained-23706/

If you want to know about the technical capabilities of DACs you can actually buy, go over to audiosciencereview.com.  They publish detailed measurements.

Incidentally, you don't need to spend $5000 to get a DAC with stellar quality, probably better than you can perceive. I suggest channeling the extra money into your speakers and amplifier.

I hope this helps...  Take care.

[zybex]

Regardless, you can't really enjoy music with a full 24bit dynamic range.  The sound pressure level provided by 16bits can damage your hearing.  The human ear can only hear about 20bits of dynamic range without pain. 24bits, 144db, causes instant hearing damage.

Eheh, this makes it sound as if just listening to 24-bit audio is dangerous
Just to clarify what Wer means (excellent link, btw): 144db is the SNR of a 24bit signal. Meaning, you would have to listen to it at at least that level to hear any quantization noise (not advisable). 16bit SNR is 96db which some people claim to hear, so 24bit is definitely a get-off-my-lawn improvement in that area. That said, I'm known to appreciate even 128Kbps MP3 audio, so I'm not the best person to talk about these things

And on this point, would you know if most of the better audiophile DACs (~ $5000 range) employ some kind of interpolation or even extrapolation schemes to predict the values of missing data in a 16 bit recording that might have been extant if the recording were a 24 bit recording?

There are no "missing data" values. It's just different levels of sampling precision. i.e., where a 16bit recording may say 0.1234, the 24bit recording will say 0.123456. This makes the recording more precise, closer to the actual original sound, so the implied added quantization noise is smaller => higher fidelity.

Going back to ECC memory, what may happen when a memory bit gets randomly flipped. If that bit belongs to the already decoded audio buffer that is just about to be sent to the DAC (the probability of this happening is so vanishingly small that you shouldn't worry about it, like Wer said) - anyway, what happens is that one sample is corrupted by the bitflip. Depending on which bit is flipped, the change in the actual sample value may be insignificant or may be very large (ie, turning a 0x8000 into 0x0000 or vice versa in the worst case). From the DAC point of view, there's no indication that this sample is corrupt - it's just another value. However, the samples right before and after this one are normal, so this corrupt out-of-line value will basically get smoothed over by the DACs filters and other stages, leading to a very small analog change. At 44.1KHz, 1 sample represents just about 23 microseconds, and the output wave is disturbed for no more than 4x this interval (Nyquist–Shannon). So in practice, you'd be hard pressed to hear a single bitflip effect on your speakers even with a cheap DAC/amp.

What you CAN hear is when there's an error or bitflip on the encoded audio - let's say a sector cannot be read from the CD/DVD/BR disc, or fails to be decoded by the codec due to some data corruption - in this case, it's a contiguous block of data representing a few milliseconds of audio that is corrupt/missing, so all the software can do is repeat the previous block or interpolate to smooth it over; in any case, we can definitely hear this in the output.