Optimizing Plot Speeds

Chia Plotting
1

Hi all,

Looking for a little guidance here… I’m trying to optimize my rig and I’ve gone through much trial and error. I’ve got 200 TB plotted and about 200 to go for now. I’m getting 35-40 minute plot times with the following build and specs. I feel like I can be doing better, based on what I’m seeing from other people’s similar and sometimes lesser rigs…

Specs:

SAMSUNG 970 EVO Plus SSD 2TB
128 GB DDR4 ram
Asus Tuf Gaming Motherboard Z590
Intel I9 10900k (10 core)

Plotting with MAdMax k32 plots on Ubuntu. My script for running is the following…

screen -d -m -S chia1 bash -c ‘cd ~/chia-plotter/build && ./chia_plot -f -c -n 1 -r 16 -u 256 -t /media/hespori/allotment/ -2 /media/ramdisk/ -d /media/hespori/magic/ |tee ~/mmlogs/madmax1_1.log’

I’ve taken away addresses and for the sake of this post. This is averaging around 36 minutes per plot. Running 16 threads because that seems to be what the available ram can handle after the 110g ramdisk.

Anyone offer me any advice whether or not I can achieve faster speeds?

-Patrick

2

Take a look at this YT channel below. It has (IMO) the best advice on how to fine tune your plotter.

3

I have a i9-10850K/64GB RAM/980 PRO 1TB and use -r 20 (20 threads) and get about 29 to 30 minute plots without a RAM disk. I think you can do better. Is your -d (destination) drive slow? If so, consider using another SSD as the intermediary, even if it’s a SATA SSD.

4

Brilliant, I hadn’t thought of that. All my -d are 7200 rpm spinning drives. I am using 20 threads now. My buddy with a similar build is getting 49 plots a day w/o ram disk. ram disk sped me up to 43 plots a day… but still feels slow.

Do you automate the copy to sata ssd then write to final destination? If so, how do you do it on your setup?