Quote:
Originally Posted by 78372
Thanks. I will be applying on GTA V rpf files, everything is textures except few, should I go for this settings?
lolz -mc8 -tt1 -mt1 -mtb512 -d128 -fba4096 -dto0 -dm00
Edit: Still it's asking for 33hours, I think I should use a more simpler settings
Edit2: Ok, I need an upgrade :P
|
You are using only 1 thread and probably weak cpu so it *will* be slow. I guess that 33h would be ~6h max on my 4690k 4.2ghz with 4t. But I doubt dlz is going to be any faster on same cmp ratio. In my tests it was worse in both ratio and speed although I didn't gave it enough time. Dlz is IMO pretty much old and obsolete version of lolz, they both combine lzma with cm and thats that.
Here is what you need to know:
-lolz seem to be heavily affected by -mc as much as -tt, I think they are chained together so mc in lolz will slow down more significantly than on pure lzma.
-lolz however seems more stable in speed on bigger dictionaries and block size. Lzma is much faster with 64m/64m than say 128m/256m, but lolz seems to be almost same always therefore it does get closer to lzma on bigger dict and blocks.
-mc have much lower effect on textures, 8 should be more than enough
-other than those settings, only -dto0 -dm00 is worth it. -dto0 will gain about 30-40% in speed and -dm00 speedup scanning and I found it sometimes even better than default autodetect. Dm00 is already used on raw images as default. These 2 options affect cmp only minimally, from what I seen.
With above settings, I can get ~4m/s cmp with lolz. With lzma and similarly big dict/block about ~5-6m/s. Cmp ratio suffer minimally, about ~1-3% vs full settings, but oh boy 20h vs 4h! And when you offload non-textures to lzma(which you should because low mc will make lolz compress non-textures worse in such case) then you can further fasten usually to 2-3h, like 20+gb game. Again today I compressed Ni no Kuni 2, got almost same size as FG but it took under 2h to compress 41gb of game data.