Optimizing Disk Space. Finding the Balance Between Capacity and Performance?

Chia Plotting
1

I just finished plotting a 16TB disk (actual size 14.5TB) and filled up around 14TB, leaving approximately 500GB of free space. I’m wondering if it affects the disk’s performance when it’s so heavily filled, or if I can confidently add four more plots. The differential drive size (16TB disk / actual 14.5TB) serves as a safety margin to ensure proper disk functioning.

How do you handle this? Do you fill it up completely or leave around 1TB of free space?

2

Forget plotting old 100GB plots, use madmax or another compressed plots. And yes, fill to full

1 Like
3

Im plotting on MadMax but K32 so 108GB. Are those compressed plots so much better ? I have only 80TB at the moment so i don’t think ist relevant for me at so low space ?

4

I don’t know. Maybe it isn’t worth the bother for ~$4 per month.

1 Like
5

Since you are not writing and changing data on the disk regularly you don’t gain anything keeping open space. My self im useing ext4 and pulled out all the reserve space ext4 normally keeps. In Ubuntu tune2fs -m

3 Likes
6

If you use compressed you can get a basically free extra 15-25 TB effective space

16 Terabyte (TB) = 14.55 Tebibyte (TiB) thank Microsoft for mixing up the notations.

Filling a hdd to the max will impact write speed, the last few plots will take longer to copy. But after that you only read from it, and this is no problem. So you can just fill it to the max.

7

Can I do this with my Ryzen 9 3900X, or do I need a different polling method?

1 Like
8

Depends how much ram you have, not sure on the requirements for Madmax Gigahorse compressed plots.

Chia Bladebit cuda currently requires 256GB ram, so no good for you, 128GB ram requirements is in the works and may be available soon, no idea if they’ll release a version with lower ram requirements though.

9

It’s actually easy to calculate. Windows calculates GiB,TiB but shows as GB,TB.

1 Like
10

You can CPU plot without any special RAM requirements or GPU and still enjoy the compression advantages. It’s just not much faster than the regular madmax plotter if you go that route (and you have an SSD to hammer).

11

With gigahorse (by madmax), yes you can make compressed plots the old fashioned wat, with cpu and minimal ram.

Bladebit (by chia) not yet, but it is in the works, coming out with chia version 2.0

1 Like