Why is sysvinit so despised?

animorphs

‘Omer Simpson

omero

Even imagination only is ugly

A vampire would argue on this

Just wondering about your sick mind

Dr. Se

In my opinion best runlevel implementation

1-6

init default

If you change the runlevel it syncs the processes (stops/starts)

@capezotte brotli on extremely large files ~0.5 MB/s ?

brotli in general is VERY slow when compressing

decompression is about as fast as gzip though

and ratio competes with xz

seems to be a good tradeoff for servers

I’m trying a hilariously dumb action: brotli compressing tar archive of 15 VP9-encoded video lectures with total size of 6GB

my rough estimations tell that it’d take 4 hrs

video doesn’t compress well with general purpose lossless compressors, it might not be worth it

ok

You have to reencode your vid in a lower quality or better codec

They’re already compressed by YT

and unfortunately I don’t have Intel Arc GPU for hardware encoding into AV1

while faster, hw-accel increases losses significally

that’s kind of the point here

any ideas why HEVC 720p 20fps video could have larger size than AVC 720p 20fps ?

length and content entropy are almost identical between the videos

NVENC h.264 and NVENC h.265 encoders were involved

When you test different cpu encoders with same quantization settings, you expect to get comparable quality while speed may vary drastically because of optimizations, code quality. It cannot be applied to gpu vs cpu encoder comparison

Time taken exactly as predicted, compressed size is 96% of initial size

it kinda makes sense if those videos have a lot of repeating parts to compress them with brotli/lzma

but yeah with little benefit as your test demonstrated

back in the day with mpeg2 it used to do wonders

Because they’re already VP9 compressed as I wrote above

no i mean that compression algos work lineary while e.g. lzma analyzes whole file for patterns within it

that’s why lzma is unrecoverable in case of damage