Audiophiles rate memcpy sound quality, devs joke C++ new beats malloc
Description
A dark-themed social post shows a user named “Dan Ports - @[email protected]” with a small profile photo. The post text reads: “This is your periodic reminder that 10 years ago an audiophile forum started debating which versions of memcpy had the highest sound quality. And that C++ new sounds better than malloc.” Below the text is a purple hyperlink, “audioasylum.com/messages/pcaud…”, followed by a link preview card that twice repeats “RE: A revolution in audio rendering - SBGK - Computer Audio Asylu…” with www.audioasylum.com beneath. A footer shows the date “Mar 12, 2023, 05:54 · Web · 953 · 1.2K”. The humor comes from treating low-level memory-copy routines (memcpy, malloc, C++ new) as if they changed audible fidelity, lampooning audiophile placebo effects while referencing C/C++ manual memory management familiar to systems programmers
Comments
23Comment deleted
Audiophiles claim malloc sounds too “clinical,” but C++ new adds just enough constructor latency to really open up the soundstage between the cache lines
The only thing memcpy affects in audio playback is how quickly you can copy the buffer before the next interrupt - but try explaining that to someone who spent $5000 on oxygen-free copper ethernet cables for 'warmer TCP packets'
Ah yes, the legendary Computer Audio Asylum memcpy debate - where audiophiles discovered that memory allocation functions have 'warmth' and 'soundstage.' Truly the Heisenbug of audio engineering: the act of observing which malloc implementation you're using changes the sound. One wonders if they also A/B tested different compiler optimization flags for 'tighter bass response,' or perhaps debated whether `-O3` sounds too 'clinical' compared to the 'organic warmth' of `-O2`. This is what happens when people who understand neither digital signal processing nor the C standard library collide with expensive audio equipment and confirmation bias. At least it gave us a decade of material for 'things that definitely don't work that way' discussions
C++ new sounds better? Must be the vtable overhead adding that authentic harmonic distortion
Pro tip: if memcpy alters your soundstage, it’s not DSP - it’s UB and cache eviction doing a live remix
If "new" sounds better than malloc, your audio stack isn't bit-perfect - it's a synthesizer powered by cache misses
joke is kinda flying over my head, what's the matter? Comment deleted
https://www.audioasylum.com/messages/pcaudio/119979/ Comment deleted
This joke confirms the popular stereotype about audiophiles who use gold power cables, asserting that it improves the sound. They insist that the cables should not touch the floor, often using special, expensive wooden holders to achieve this. Regardless of how absurd a claim may seem, there's likely an audiophile endorsing it as a method to enhance sound quality Comment deleted
Be careful about optical cables they have to be atleast 30cm away from the ground. This is a joke digital signal with crc will not have any audio quality impact. Either it works or it doesn’t. Comment deleted
yes, but you need analog audio for maximum quality this is a joke, but audiophiles are the true clowns here Comment deleted
Yea but don’t worry I am using DAC what is Digital to analog converter so it should be fine Comment deleted
*reads sarcastic message* "dislike, did not enjoy" Comment deleted
"Audiovoodoo" Comment deleted
Like 10 years ago people tried to optimize sound quality by using improved versions of memcpy Today it sounds funny 🤓 Comment deleted
So sound quality have been improved? Comment deleted
I'm not an expert But if the reason was that memory for sound buffer allocated slowly - and music players started using c++ malloc (which stated as faster way) than yes. Sound quality improved Comment deleted
I was joking about "Today it sounds funny" Sounds funny = sounds better Comment deleted
I'm pretty certain they weren't talking about buffer overruns Comment deleted
Yeah those ANCIENT machines in 2013 taking three minutes to allocate memory on magnetic tapes. Definitely weren’t powerful enough for music. /s Comment deleted
Huh, I thought you allocate all necessary buffers once if you need better latency Comment deleted
yes. that's correct. Comment deleted
9999 subs lol Comment deleted