Improved audio rendering with an optimised version of memcpy (2013)

It&#x27;s important to appreciate that there are several different breeds of audiophile out there...<p>There are those that are obsessed with objectively measurable sound quality. They want to reproduce the recorded waveform as accurately as possible, and they want products that can be measured doing that, but they often pursue accuracy beyond the limits of any human to hear.<p>There are those that like a little distortion or degradation if it sounds nice to them. These are the people who collect tube amps and will openly admit they&#x27;re not as accurate, but accuracy is not what they&#x27;re after.<p>There are those who embrace <i>voodoo</i>. I knew a physics professor who was one of these. He had hideously expensive isolation platforms, resonator weights, unobtanium cables, &quot;ambient field conditioners&quot;... You name it. He knew there was no scientific basis for any them to work, but he had very deliberately deluded himself. If asked, he&#x27;d explain that the ultimate goal of the audiophile&#x27;s pursuit was pleasure. If he could shut down his reasoning and spend hours moving resonator stones around on top of his CD transport until he thought he had improved the sound, and this gave him pleasure, was that pleasure not real even if the sound was exactly the same?<p>Then there are those who combine contradictory aspects of all of the above without any self-awareness or critical thinking skills. That&#x27;s the sort that will torment themselves, and others, arguing on internet forums about why their own snake oil of choice is legitimate, real, and how anyone who isn&#x27;t using it is a cretin.

I&#x27;ve always felt totally alone falling somewhere between hardcore audiophiles and the &#x27;norm,&#x27; which seems to feel a cell phone or cheap laptop speaker is as good as anything else for music. I really notice the difference between cheap crap, and say an older used brand name receiver with freestanding floor speakers, the kind of system you can find on craigslist for about $100. But I cannot notice a major difference between the $100 system, and say a $100k system (or whatever audiophiles spend). I also do notice a big difference between a 96Kbps MP3 and 320Kbps, but not between 320Kbps and lossless.

Actual link to forum: <a href="https:&#x2F;&#x2F;www.audioasylum.com&#x2F;messages&#x2F;pcaudio&#x2F;119979&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.audioasylum.com&#x2F;messages&#x2F;pcaudio&#x2F;119979&#x2F;</a><p>&quot;<i>found that a function called memcpy was the culprit, most memory players use memcpy and this is one of the reasons why memory play sounds worse ie digital sounding. Fortunately there is an optimised version of memcpy from <a href="http:&#x2F;&#x2F;www.agner.org&#x2F;optimize&#x2F;" rel="nofollow">http:&#x2F;&#x2F;www.agner.org&#x2F;optimize&#x2F;</a>, using this version removes the hard edge produced by memcpy.</i>&quot;

Fun fact: vinyl has constant angular velocity, not constant linear velocity. It means that sound fidelity gets progressively worse while going from outer tracks to inner tracks.

How can audiophiles continue to exist with knowledge being so easily available? And how can someone be clever enough to be a software engineer and not clever enough to figure out crystals in bags taped to cables do not affect music quality.<p>Just some nuggets from that thread:<p>&quot;Sounds awesome, the previous version had a slight tendency to defuse the treble, but with VS2012 compile it is a much more complete sound with absolutely no digital harshness, some 16&#x2F;44 albums I could hardly play before without getting a headache are now rendered in their full glory.&quot;<p>&quot;It&#x27;s just lazyness on the part of the player developers that they rely on the old methods, I guess they think bits are bits.&quot;<p>&quot;Goto also sounds better than anything else I have tried.&quot; - the &quot;Goto&quot; in the quote means a goto in C used to replace a loop.<p>&quot;also most players use malloc to get memory while new is the c++ method and sounds better.&quot;

Many moons ago I found myself making parts for the audiophile community. Eventually I got in to making complete headphones. Designing and building audiophile headphones is a multi-modal skill that is very engaging. I was pretty good at it. But I had to exit the audiophile world because it was too crazy for me.<p>I would get messages from customers that were completely insane. And if you say &quot;yo, that&#x27;s insane&quot; word gets out and it damages your brand. So you&#x27;re rewarded for perpetuating these weird myths.<p>If I could build these really cool headphones without having to deal with audiophiles, I probably would. It was challenging and a lot of fun, and I was pretty good at it.

I saw the observation years ago that &quot;Normal people use their stereos to listen to their music. Audiophiles use their music to listen to their stereos.&quot;<p>I guess if that&#x27;s your thing, go for it?

I think I read a thread like this almost 10 years ago when I happened to be doing some audio work, maybe it was the same one.<p>If memory serves, there was somebody saying that exiting the program cleanly sounds worse, you need to crash on a page fault reading the input buffer instead, and that&#x27;s your end of file behavior.<p>Someone else said FLAC makes it sound worse, but I suspect they just couldn&#x27;t figure out how to use libflac.<p>In other words they were bad programmers writing buggy, half baked shit, and they explained away their broken code as sounding better.<p>Even the snippet I see in the linked post, right off the bat they start their loop with WaitForSingleObject, that&#x27;s kind of suspect; the first thing they do is enter the kernel to block at every loop. I suspect they aren&#x27;t doing that in an optimal way.

I once had an &quot;audiophile&quot; sales person in a showroom almost hit me in anger, after confronting him on his loudspeaker cables that had arrows on them pointing out which way the cable should be connected to sound best.<p>As any EE knows from linear elements course A, a linear two-port network is reciprocal. If it is not reciprocal it is not linear.<p>The simple conclusion is that if it matters how you orient the cable, then it has distorsion.

That&#x27;s so dumb. Everyone knows that string.h functions produce the best sound you can get, and you should prefer `calloc&#x27; so you get clean sound.<p>Ideally you want to write your audio rendering code in asm, too, since newer C compilers add a lot of junk to binaries that reduce the quality of the output. If you can&#x27;t do this, simply use older compilers.<p>But if you&#x27;re going to write in C, I suggest Cosmopolitan Libc and running directly from BIOS since operating systems end up doing a lot of work that produces electrical noise on your hardware.<p>You&#x27;re better off running literally only audio rendering code for the best quality rendition.<p>&#x2F;s

The claim made is within the realm of plausibility. A memcpy call in a playback loop was replaced with an optimized version, and this made some harsh audio artifact (&quot;edge&quot;) go away.<p>The poster showed the code and what was done to it.<p>The code does strange things, though, like wait for a Win32 event on every iteration. And not only one time; there are multiple calls like this:<p>WaitForSingleObject(hNeedDataEvent, INFINITE);<p>Is that an auto-reset event that some other thread has to be constantly banging on to keep the loop going? its odd since the code has supposedly buffered the entire file of samples; so you&#x27;d think it would just be iterating over that buffer in some chunks and shoving them into some pipe&#x2F;device. There is also a puzzling ZeroMemory call.

At HiFi Klubbben in Lillehammer, Norway some guy tried to sell me gold cables, because the sound quality becomes so much better with them. I worked as an audio engineer at the time. I just smiled at him politely, then I left the store.<p>Next time I came around, he tried to sell me this CD player with a huge brick in it. &quot;You get much better sound quality when the player has a brick in it,&quot; he told me. So I answered, &quot;For sure! It makes the sound a lot more stable! Badum, tsch!&quot;

To fully grok things, the context of the thread is a person so unhappy with a JPLAY[1] beta that they roll their own command-line audio player with the single-minded goal of eliminating all possible forms of latency. (Upthread they talk about dropping to assembly; further down the thread someone suggests avoiding using the CPU altogether.)<p>JPLAY is the kind of thing that offers, among its features, the ability to use a dedicated PC (ideally running Windows Server) for playback.[2] This includes selling their own SSDs &quot;selected for their sound properties&quot; and &quot;Reference SATA cables to ensure absolute best sound quality&quot;, and playback-optimized USB cards and denoisers, network adapters and power supplies.[3] The best examples are the 375€ &quot;USB Isolator&quot;, which &quot;blocks all noise coming from a PC and re-generates clean USB signal from a <i>(sic)</i> scratch resulting in vast sound quality improvements&quot;[4] and their 12,000€ 4TB server featuring &quot;the only...motherboard designed exclusively for high-end audio playback&quot; that avoids the <i>common</i> pitfalls of worse &quot;office work and gaming&quot; motherboards that &quot;pollute other hi-fi components and contribute to irritating &#x27;digital&#x27; sound&quot;. Its signature review, pulled out in the product page, says it&#x27;s indistinguishable from an analog recording - the highest of high praise.[5]<p>The poster in the linked thread was like, nah, they suck, this sounds awful, but THIS is obviously better. And look how easy it was to do! These C++ developers are all <i>idiots</i>.<p>1: <a href="http:&#x2F;&#x2F;jplay.eu" rel="nofollow">http:&#x2F;&#x2F;jplay.eu</a><p>2: <a href="http:&#x2F;&#x2F;jplay.eu&#x2F;ml_post_slider&#x2F;jplay-dual-pc-mode-how-it-works-why-you-need-it&#x2F;" rel="nofollow">http:&#x2F;&#x2F;jplay.eu&#x2F;ml_post_slider&#x2F;jplay-dual-pc-mode-how-it-wor...</a><p>3: <a href="https:&#x2F;&#x2F;jcat.eu&#x2F;;" rel="nofollow">https:&#x2F;&#x2F;jcat.eu&#x2F;;</a> doesn&#x27;t work in Firefox<p>4: <a href="https:&#x2F;&#x2F;jcat.eu&#x2F;product&#x2F;usb-isolator-usb-audio-galvanic-isolator&#x2F;" rel="nofollow">https:&#x2F;&#x2F;jcat.eu&#x2F;product&#x2F;usb-isolator-usb-audio-galvanic-isol...</a><p>5: <a href="https:&#x2F;&#x2F;jcat.eu&#x2F;featured&#x2F;xact-s1&#x2F;" rel="nofollow">https:&#x2F;&#x2F;jcat.eu&#x2F;featured&#x2F;xact-s1&#x2F;</a>

Speaking of Audiophile superstitions, I do wonder what the point of fully-diminished returns is. I don&#x27;t mean the flat-earth level stuff. I mean standard audiophile discourse.<p>Can humans tell the difference between a laptop DAC &amp; an external DAC. Do expensive amplifiers have noticeably lower white-noise than the standard amp? Do high-impedance headphones actually achieve anything ? Do frequency response graphs mean anything if you can digitally EQ &amp; is digital-EQ fully lossless ? Are planar magnetic headphones a scam ? Is FLAC practically overkill for a human ?<p>I think I can tell the difference between a 100$ and $400 audiophile headphones. Amplifiers are more straight forward, because it is simply a loudness thing. And at high-enough volumes, I can still hear some white-noise on high-gain settings. But I am less sure about the other questions.

Audio and Game devs: Need to port my custom malloc and ring buffer code to get rid of those malloc pauses<p>JavaScript devs: Foolish audiophiles think they can hear when someone calls malloc!<p>Mis-diagnosing a problem doesn&#x27;t mean there isn&#x27;t a problem. If you call memcp in a way that calls malloc, and your app has been running a while or your system doesn&#x27;t have a lot of free memory, you&#x27;re going to get pauses while malloc looks for a free address block. If that pause takes too long, your audio buffer will stall and cause a harsh crackle. People will hear that differently, and explain the problem in different ways.<p>TL;DR: audio and game programming sucks, and if you allocate heap memory people will actually notice.<p>In the extreme:<p><pre><code> Imagine my surprise when the customers chief software engineer said &quot;Of course it leaks&quot;. He went on to point out that they had calculated the amount of memory the application would leak in the total possible flight time for the missile and then doubled that number. They added this much additional memory to the hardware to &quot;support&quot; the leaks. </code></pre> <a href="https:&#x2F;&#x2F;web.archive.org&#x2F;web&#x2F;20170504104857&#x2F;https:&#x2F;&#x2F;groups.google.com&#x2F;forum&#x2F;message&#x2F;raw?msg=comp.lang.ada&#x2F;E9bNCvDQ12k&#x2F;1tezW24ZxdAJ" rel="nofollow">https:&#x2F;&#x2F;web.archive.org&#x2F;web&#x2F;20170504104857&#x2F;https:&#x2F;&#x2F;groups.go...</a>

Okay, I don&#x27;t have enough context to really know, but while it&#x27;s (always) good fun to laugh at audiophiles.. Could it be that the guy is actually experiencing an improvement due to actual real reasons?<p>Imagine: They&#x27;re on a super low-end chip, feeding a bufferless DAC, and they&#x27;re experiencing buffer under-runs, which cause really noticeable degradation ?<p>Imagine: They&#x27;re using some API so wrong that it &quot;works by accident&quot; but only really when timings are &quot;just so&quot; ?

To be fair. If you had poorly shielded audio equipment (maybe because you bought directional copper and forgot to buy the necessary diamond infused audio grade shielding) then its entirely possible that different memcpy implementations might produce different frequencies of noise which would have a noticeable impact on the audio quality. Although, who am I kidding? These people also think Ethernet cables can be directional.

Kind of along the same lines, my local cheapo cable store likes to run a demo by the front door with the same movie playing on identical TVs, with one TV using an expensive cable and the other using a very inexpensive cable. HDMI, of course. The owner likes to ask people to tell him which picture is better. You can guess the results of that.

Forgive me for repeating an old joke, but it does say so much about audiophile culture:<p><i>Q: What&#x27;s the difference between an audiophile and a musicologist?</i><p><i>A: A musicologist uses a stereo to listen to music, but an audiophile uses music to listen to a stereo.</i>

I used to work for Lear corp, they had a fat deal with JLR to develop &#x2F; mass produce an amplifier with 1000+ electric components (5xdual-core Sharc included). MQA software with 192kHz samplerate &amp; audio software processors was included. People of MQA surely have pointy ears like mr Spock. One of our sw architects invited people of MQA in front of a fancy listening setup for speaker quality. Hearing tests with the same speaker set, one with crappy thin copper cables, the other with heavy 6mm2 cables. The people from MQA found the former setup sounding the best. The thing with audio perception is, its like wine tasting, the quality depends on the setting, the people around you and your current mood. I like video algo&#x27;s much better, at least you can see the difference.

I see a business opportunity here in providing very expensive repackagings of compilers and libraries of &quot;audiophile&quot; quality. It just needs to be really, really expensive.

Bet you they are all rocking Shakti stones: <a href="https:&#x2F;&#x2F;www.shakti-innovations.com&#x2F;product&#x2F;shakti-stone&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.shakti-innovations.com&#x2F;product&#x2F;shakti-stone&#x2F;</a><p>&gt; Music reproduction is clearer, with more liquidity, dynamics and focus. The improved inter-transient silence allows the listener to hear ambient cue information essential for accurate perception of stage depth, width and unwavering imaging. High quality video systems will benefit from SHAKTI devices near power supplies, projection guns and laser disc&#x2F;DVD players. Reduced color noise and improved convergence alignment are some of the improvements that can occur.

Those fools... first they praise vinyl, then would ruin everything by playing music through C++? Get over it, folks, for real vintage unadulterated sound nothing beats Turbo Pascal!

Presumably the CPU performing memcpy is sharing power with the DAC and amplifier, assuming a device that does all these things. It&#x27;s not hard to imagine there being an influence, like an analog version of timing-based bit-snooping.<p>At the same time, I venture to say none of these folk could pass a double-blind test with much better than random chance. There&#x27;s not much else to say. Demo that and color me surprised.

I used to own 20$ headphones for few years, I upgraded to 50$ ones for the next couple of years and finally to 200$ ones. I have felt big differences each time. I don&#x27;t see whats funny about that. If I upgrade now (I can&#x27;t quite afford to considering the diminishing returns and my income) but still I expect to get upgrade in sound quality, comfort, build quality, etc.

I&#x27;ve been in premium audio my entire career. There are many in the industry who fight the good fight.<p>The issue is that most of the snake-oil comes from some truth. Here we are at the how much profit is fair debate. Contextualize for your own industry and&#x2F;or product and you&#x27;ll find it tricky to justify a &#x27;fair&#x27; value for anything. Whatever the market will bear, right?<p>Audio hardware is one of the oldest electronic tech segments.<p>&#x27;good enough&#x27; for even the most persnickety audiophiles happened late in the 1960&#x27;s.<p>Since then it&#x27;s been about repackaging and doodads.<p>There are a few technologies that haven&#x27;t found fit yet for market or manufacturing reasons. Some of those will enable big leaps, but require scale and cooperation from content producers.

I mean in theory different versions of memcpy could result in different EMI emissions which could get picked up by various analog paths. It&#x27;s not as crazy as gold plated optical cables. Not really a problem today, but back when everyone had soundcards, EMI could definately screw with your audio if your PC wasn&#x27;t built correctly. Probably a worse problem for those who didn&#x27;t plug in their computer into a grounded outlet as well.<p>I&#x27;ve definately heard hums in computer audio stemming from EMI when the computer is doing something heavy. Granted that was back when having your cellphone near your PC screwed with the sound when someone was calling.

This is totally ridiculous, but a lot of time there is a grain of truth behind some of these myths, that is just diluted by retelling from non technical people. I had a serious musician once tell me that if I rip a recording from a DAT tape to my disk it won&#x27;t sound as good. It sounded about as stupid as this this claim, but the basis of it was conversion between 48khz and 44.1khz sample rates.<p>I also remember a friend of mine learning graphic design many years ago, who told me that their teacher told the class Photoshop on Mac produces higher quality images than on a PC. I don&#x27;t remember what the basis of that was.

I realized I wasn&#x27;t a serious audiophile a long time ago, but still went through a quest to improve sound. My first receiver was a cheap plastic thing that produced an enormous amount of white noise floor- a loud hiss not much quieter than the music itself. I went through several receivers until I found one that didn&#x27;t hiss so badly, but I found that the volume wasn&#x27;t very impressive (turning the knob mainly just increased hiss and distortion). I learned later that most people address this using a preamp, although at the time I was poor and couldn&#x27;t buy one.<p>I mostly listen to soundboard recordings of live music (Grateful Dead and Phish) where some recordings are fantastic and others aren&#x27;t. My personal favorite reference to listen to is Dave Brubeck&#x27;s Take Five- although it has a high noise floor, the recording is still excellent (I like to listen to the creaking of the bass during the drum solo because it&#x27;s quiet and in the background).<p>Over time I moved away from speakers because I&#x27;m rarely in a situation where I can listen at the volume I like without people either asking me to turn it down or leave. I got a pair of Sennhauser headphones- nothing fancy- and am quite happy listening to my desktop computer through that. Switching to another set of headphones- it immediately sounded terrible and I had to play with various settings (turns out the audio driver was trying to send through lots of very low frequency and the headphones made a shambles of that).<p>There was a period where mp3 encoding was terrible and generated obnoxious artifacts. Nowadays many mp3 128k recordings are fine for me.<p>So for me, while not an audiophile, these things matter: reduce hiss to an acceptable level, get headphones or speakers you like, and use EQ to tune it to a sound that works for you.

And then there’s guys like mr. Rainger (<a href="https:&#x2F;&#x2F;www.raingerfx.com&#x2F;shop&#x2F;p&#x2F;minibar" rel="nofollow">https:&#x2F;&#x2F;www.raingerfx.com&#x2F;shop&#x2F;p&#x2F;minibar</a>) just casually inserting an actual pod of liquid into the signal chain to replace a resistor and drive an eq (via its transparency). Coke sounds different from single malt and a protein shake is different again…

I wonder if the said version of memcpy did screw up some bits and indeed affect the music fidelity. I mean audiophiles can detect that change right? It’s like the gm&#x2F; ford vanilla ice cream bug:<p><a href="https:&#x2F;&#x2F;www.marketcalls.in&#x2F;fun&#x2F;vanilla-ice-cream-that-puzzled-general-motors-amazing.html" rel="nofollow">https:&#x2F;&#x2F;www.marketcalls.in&#x2F;fun&#x2F;vanilla-ice-cream-that-puzzle...</a>

Audiophiles are people that need to understand that there&#x27;s such a thing as diminishing returns. Yes, good amplifiers and, more importantly, good loudspeakers and headphones can significantly improve your experience compared to listening to music in some cheap JBL boombox. But once you&#x27;re a step above the cheap stuff, additional expenses don&#x27;t significantly improve your experience.

My favorite is something I read some 15 years ago or so, although I don&#x27;t remember really where, that early CDs had terrible sound because people early on failed to realize the big difference between vinyl records and CDs. Namely that records are read outside in and CDs inside out, and this somehow required adjustments that made everything better, although I again don&#x27;t remember how it was &quot;fixed&quot; :-).<p>Anyway, I don&#x27;t know much about this stuff, but the biggest problem is always lack of education, not that people are stupid. There&#x27;s a lot to know and people usually don&#x27;t and can&#x27;t know everything. I&#x27;d say in certain cases intelligent people are more problematic because they&#x27;re capable of constructing a plausible explanation that makes sense only because they lack the complete picture.

Funny how everyone laughs about this but gets really offended in the subthread about coffee. Instead of claiming it&#x27;s different and that coffee grinding actually matters you should use your interest to try to understand the audiophile&#x27;s perspective. How you feel about coffee they feel about audio.

What is often unknown is that the ear &lt;-&gt; brain interface is not one way but two ways. That means that the brain actively changes physically the mechanics of the ear for better hearing in a feedback loop. Therefore all sorts of &quot;freakiness&quot; can influence our hearing.

Hum, they missed the obvious issue: `+=` requires shielding by `void()` for optimal sound. (It should be obvious how `=` acts as an antenna here, if not properly shielded. In Algol, assignments were still properly grounded by `:=`, which gave you that warm sound.) ;-)

Hearing the difference now isn&#x27;t the reason to choose an optimal memcpy.<p>Some implementations use lossless copying, while other methods are &#x27;lossy&#x27;. The differences between GNU libc and Boost are subtle, but important. What this means is that for each year your audio sits on your hard drive, it will lose roughly 12kbps, assuming you have NVMe - it&#x27;s about 15kbps on M.2, but only 7kbps on eSATA, due to linear velocidensity. You don&#x27;t want to know how much worse it is on CD-ROM or other optical media.<p>I started collecting MP3s in about 2001 (on Windows 98 using an ancient version of malloc and memcpy), and if I try to play any of the tracks I downloaded back then, even the stuff I grabbed at 320kbps, they just sound like crap. The bass is terrible, the midrange...well don&#x27;t get me started. Some of those albums have degraded down to 32 or even 16kbps. If you ever used uClibc, I hope you kept your original vinyls! Proper memcpy rips (or better yet, ripping directly to factory-fresh RAM) from the same period still sound great, even if they weren&#x27;t stored correctly, in a cool, dry place.<p>Seriously, stick to audiophile-grade library functions: you may not be able to hear the difference now, but in a year or two, you&#x27;ll be glad you did.

2013, so computers &gt;&gt; faster than human hearing so the following probably doesn&#x27;t matter but devils advocate : memcpy puts stuff in the heap instead of the stack, which requires a malloc, a system call, which could introduce some latency. Then layer on that our audiophiles are probably listening to ridiculously high bit rate audio which results in large files, such that without swap buffering there may be some periodic buffer-play-buffer artifacts introduced. This is why you buffer but [new] is also a heap allocation, so same syscall, shouldn&#x27;t be any different. Is it calloc? calloc in theory should be worse.

Maybe they were just early to discover a long standing Linux bug in memcpy: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=34095609" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=34095609</a><p>:p

Meanwhile I play all my vinyl through a pair of cranked computer speakers from the early 2000s lol. The pursuit of &quot;highest quality&quot; seems like a good way to take the fun out of music listening.

is this version of memcpy gold plated and developed by Monster?

As a Linux DJ, I just wish the waveform scroll was jerkless.<p>I run MiXXX waveforms at 60hz, it takes about 5% CPU, so it has lots of CPU time to spare, and even then, there are graphic imperfections. Every couple of seconds or so, there&#x27;s a frame skipped, and it is impossible not to perceive it.<p>This has nothing to do with sound, but I think it is relevant to this discussion :)

The original memcpy sounds good after some burn-in

Fundamentally this all boils down to precision measurement&#x2F;observability (for both analog and digital).<p>Occasionally, small changes are observed in high end equipment with high precision instrumentation - but a subjective interpretation to a psychometric observation is applied as cause and effect.<p>Arguably, the cost alone is enough of a placebo to sell anything for a desired effect.

I&#x27;ve been using the same system for decades.<p>Is there a way I can measure how well it does at sound quality other than the &quot;listen to it and see if it seems ok&quot; method? Because as with glasses, you get used to a prescription that is slowly degrading over time, same with hearing.<p>How are people measuring fidelity in a home environment?

There is a saying that hifi using electricity coming from nuclear sounds a big warmer than that from a hydrapower plant...

Hah, good times. In one of my earliest posts on the internet (I was green behind the ears and didn&#x27;t actually know what Usenet groups were, but still hung around those) I mocked some audiophiles because of their debate about whether black disks sounded better than off the shelve CDs (black sounds better, obviously).

Whenever I think of audiophiles, I imagine a couple of wealthy successful older men hanging out on a weekend. While the wives are elsewhere talking about the latest vacation or whatever, they need something to talk about.<p>Because they&#x27;re the kind of men who love maximizing making numbers go up, it needs to have something to do with optimizing performance. Because they&#x27;re wealthy and their success is a big part of their identity, it needs to be something expensive.<p>Then, because they&#x27;re friends who want to feel useful to each other, it needs to be riddled with inscrutable lore. That way there&#x27;s always some handy bit of research, some nuggets of wisdom that they can share with each other. That&#x27;s what this is really about: feeling like they can bond by sharing expertise about some complex hard to optimize problem.<p>The actual problem is self is essentially arbitrary and subjective. In fact, it&#x27;s better if it <i>is</i> arbitrary. Because any problem that can actually <i>be solved</i> means reaching an end to the ability to kibbitz and geek out. That&#x27;s why bringing up objective metrics is such a faux pas in the audiophile world: it kills the game. And the whole point is to play the game together, forever.<p>See also:<p>* Car tuning<p>* JavaScript front-end frameworks<p>* Barbeque<p>* Linux distros<p>* Cycling<p>* Optimizing passenger flying experience

Retro computing just had greater empathy<p>Running programs on modern hardware; the explosion of complexity increases the amount of impure copper; pushing you away from direct and immediate feedback, and towards an abstraction soup that makes my eyes water<p>How can we still trust programs running on todays hardware?<p>That is why I&#x27;ve ported quake to run on an abacus

So this is about sounding &#x27;less digital&#x27; after the fact it was copied? I could understand if a very bad&#x2F;buggy memcpy build was used while streaming audio and causing interference with the DAC. But no, it&#x27;s just idiots discussing something they don&#x27;t understand.

These kinds of things make me wonder what might be the ways to distinguish between the ignorant and the trolls &#x2F; the people who take advantage of the ignorant to make money.<p>Even though that level of (sometimes willful) ignorance is scary, the whole thing is fascinating, in a train wreck way.

part of me wants to give him the benefit of the doubt; he&#x27;s talking about latency and I know from experience there&#x27;s an undesirable &quot;clicking&quot; artifact produced when the sample buffer is empty (ie, the application isn&#x27;t producing samples fast enough); this is caused by the frequency abruptly dropping to 0Hz. It could arguably be described as a &quot;hard edge&quot;. So <i>maybe</i> he&#x27;s in a situation where libc&#x27;s memcpy was a bottleneck and was preventing new samples from being sent to the audio hardware fast enough.<p>But the part where he says that the c++ new[] operator sounds better than malloc is just ridiculous. I&#x27;m not good enough at mental gymnastics to make that one seem remotely plausible.

Archived: <a href="https:&#x2F;&#x2F;web.archive.org&#x2F;web&#x2F;20230312033446&#x2F;https:&#x2F;&#x2F;discuss.systems&#x2F;@dan&#x2F;110008052977994607" rel="nofollow">https:&#x2F;&#x2F;web.archive.org&#x2F;web&#x2F;20230312033446&#x2F;https:&#x2F;&#x2F;discuss.s...</a>

This memcpy debate on one side, and on the other side try getting rid of a ground loop :(

Why not link to the audiophile forum instead of a Twitter reminder?<p><a href="https:&#x2F;&#x2F;www.audioasylum.com&#x2F;messages&#x2F;pcaudio&#x2F;119979&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.audioasylum.com&#x2F;messages&#x2F;pcaudio&#x2F;119979&#x2F;</a>

Well, there is a heap, and to add or remove things from the heap takes time, and that time depends on the memory allocator algorithm used. I can totally see how it could affect audio streaming on really really slow hardware.

What happened to these people, why is this no longer a thing to the same extent it was in the 90&#x27;s and early 00&#x27;s? Was it the internet happening? Did they discover double blind testing?

That is what Ethan Winer called &quot;Audiophoolery&quot; here: <a href="https:&#x2F;&#x2F;ethanwiner.com&#x2F;audiophoolery.html" rel="nofollow">https:&#x2F;&#x2F;ethanwiner.com&#x2F;audiophoolery.html</a>

People just don’t take care of their bodies nor have “good” senses to even tell, I mean half these comments are admitting to no taste, don’t be surprised when they mean that literally

Is this about power consumption at the instruction level, affecting electricity paths on the audio chip? It&#x27;s the only thing that makes sense without reading the original discussion

Weren&#x27;t we amazed a couple of months back when menuetos managed to get audio latency down to 1ms? Yes this is ridiculous but latency does exist and is quite perceivable.

This is why I liked Hydrogen Audio the best. If you ever made a claim about sound quality that didn’t include evidence of double-blind testing, you got banned from the site.

Some people still claim vinyl has the best sound so...

Will using the right version of memcpy help with reducing the effects of rotational velocidensity when ripping to lossy formats?

Did anyone actually read or skim the thread?<p>It seems to be about reducing latency, which definitely varies with memcpy implementations.

Things like this hilight why the scientific method is so important and why double blind trials are so important in medicine!

When you cannot possibly tell the difference between two things, is when the arguments&#x2F;discussion can go on forever.

I usually don&#x27;t mind how memcpy sounds, but the flavor is awful. Especially if there is bit rot.

Was it not just a trolling thread?

Also don&#x27;t forget the gold(en) CDs.<p>And the 32 bit CDs. (They were remastered at 32 bits).<p>Audiophiles will eat anything.<p>It&#x27;s a status game.

Every true audiophile knows it&#x27;s the diet, which can change how the ears perceive sound.

Best comment: &quot;Think of it as a guerrilla progressive tax system.&quot;

Feels like we are laughing at mental illness here.

I had no idea memcpy was lossy. Astonished.

Of course, there is a lot of funny &quot;audiophile&quot; stuff.<p>Including very funny ones who use these unfortunates to show own intelligence or even (save us, Lord) wisdom.<p>For the most part, funny audiophiles are completely harmless old people (see photos from any audience of audiophiles), such children who are constantly waiting for Santa with new beautiful toys.<p>Of course, the music player will sound better if you&#x27;re not a programmer but managed to build some hell of a source code in a dubious version of Visual Studio :) This is an old story about expectations. If something subjective is expected, it can always be achieved.<p>Yes, I know vinyl doesn&#x27;t sound better. I have much better sound sources. But I like my turntable and I like the process of getting ready to listen to vinyl. It gives me pleasure, and I like turntables as very beautiful engineering structures, and no psychopathic fighter with Santa Claus will deprive me of this pleasure :)

artisanal memcpy

It&#x27;s so funny to me how many people post links to Mastodon just to try and get people to use it on HN. There&#x27;s literally no value to Mastodon being the linked to piece of content when one could actually link directly to that content! What&#x27;s worse, it took &gt;10s just to load!