Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: ALAC: different encode(r)s tested for speed and compression (Read 1362 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

ALAC: different encode(r)s tested for speed and compression

ALAC sucks.  Most of all because being unnecessary.
So I didn't even notice until recently that CUETools' encoder actually implemented some of the measures that over the years have improved FLAC compression.  Could those do anything about one of the reasons that ALAC sucks - namely the performance?

tl;dr: not enough, even if gaining a couple of percent in size over refalac for only ten percent time penalty actually is quite impressive

I ran some encoders and settings (avoiding the dual mono) through the corpus in my signature.  CPU: i5-1135G7, fanless so cooling constrained so everything run multiple times from a warmed-up computer overnight.  Timings aren't completely reliable, but, I'd argue, that isn't necessary. Writing to an internal SSD.
From largest file size, byte count rounded to the millions (not 2^20):
12 569 MB taking 555 seconds for ffmpeg with max prediction order set to 12 (like FLAC -7 and -8; ffmpeg defaults to max order = 6). Mildly surprising, this setting was worse than ffmpeg default:
12 420 MB taking 465 seconds for ffmpeg at default
12 289 MB taking 715 seconds for refalac.  (That is 743 kbit/s.)
12 093 MB taking 802 seconds for CUETools.ALACenc.exe at its default setting -5.  Even a bit slower, but more efficient. Two percent is
12 062 MB taking 40 minutes (three times as long as -5) for CUETools.ALACenc.exe at -8
12 045 MB taking 67 minutes for CUETools.ALACenc.exe at -10

12 032 MB taking 332 seconds for FLAC 1.4.2 at default -5. The comparison FLAC -5 to refalac is in line with ktf's 2022 lossless performance test, although it should be added that FLAC 1.4.x compresses better and slower than 1.3.4 used therein.
But: in that test - done on an AMD CPU - refalac was about the speed of FLAC -8. On this computer, refalac speed is much worse; closer to -8p. Even if timings are not very reliable on this computer, I am putting this on the different CPUs. (FLAC 1.4.x tests have shown some differences over CPUs too: some settings are doing relatively better on AMD, other relatively better on Intel.)