quick question, LZ4 vs LZMA2
how much faster is LZ4, how much more space does it take up.
Solution:Jump to solution
DH with Terralith + Tectonic on a 9950X3D:
LZMA2:
Time: 6:23; DB Size: 443,4 MiB
LZ4:
Time: 6:05; DB Size: 971,7 MiB...
31 Replies
cool
basicly: LZMA2 is 50% size of LZ4, but LZ4 take 1000% less time to read and write
so far I've never really had a issue with lzma2
and I prefer my DH DB not taking up a ton of space
is it really noticible when generating lods?
like its a 1ms vs 10 ms diffrence
idk I haven't benchmarked that
its really not noticeable at all if you plan on just playing minecraft
and LZ4 really does take up a LOT of space
on my ryzen 7 5700x i feel that switching to lz4 improved frametimes
in general distant horizons will, so i personally use the highest compression settings
i'm not totally sure if that was the root cause, though
I'm benchmarking it right now, give it like 5 minutes
i cant say ive noticed any frame differences in modes but ive only used it server side so far so that may be why
by the way did they remove the ignore block csz thing
server + client ofc
but using internal mode
cant find it
does it really? i recently compressed my minecraft world in 7-zip (LZMA2, 256 mb dictionary) and the compression ratio for my dh sqlite file was only 70%
wait, i'm an idiot. LZMA2 is a different algorithm from LZMA
Solution
DH with Terralith + Tectonic on a 9950X3D:
LZMA2:
Time: 6:23; DB Size: 443,4 MiB
LZ4:
Time: 6:05; DB Size: 971,7 MiB
holy
hmm
LZ4 is not worth it when your world gen is slow
the speedup barely makes a difference
and in those cases you usually end up with a larger DB than vanilla, so LZMA2 is more worth it there
in conclusion, I recomend to use LZMA2
wait but this was compression right?
not to mention there are mods that can run alongside distant horizons and speed up that generation
yes
perhaps the improvement is in fetching chunks and decompressing them
Idk how I would benchmark that
yeah that's the problem
I'm just benchmarking a bunch of different configurations on my system rn
for that google sheet
could you possibly benchmark how long it takes to load all the chunks from memory?
like leave and rejoin
you would need to do that for your system
I doubt our systems are comparable at all
just want to see if there is any speed up
alr
probably, since LZ4 decompresses faster
but its not gonna make a difference large enough to notice on my system
With high enough DH RD it may be noticeable
i guess