50+ plots/day on a single 1TB NVMe for temp 1&2 with a 5950x

Hi folks!
This is my first post as a long time lurker so I wanted to start with a giant Thank You!!! to everyone in this forum and in the Chia community in general. I’ve learned many things from you all (including how little I actually understood about SSD technologies :smiley: )

Anyways… my 50+ plots/day setup

OS: Ubuntu 20.04 Desktop
CPU: AMD 5950x @ 4ghz
Mem: 64GB (4x16GB) G-Skill F4-3200C16 @ 3200mhz
MB: ASUS TUF GAMING X570 Plus
Temp 1&2: Gigabyte 1TB M.2 NVMe SSD GP-AG70S1TB - ext4 (discard,noatime,commit=600)
Final: 8TB Avolution usb3.0 HDD - ext4 (no sleep)
Commit: 16f9793b78b60eb9a4e84ff8048998c3ac62a0e5

chia_plot -n 75 -r 32 -u 512 -t /foo/bar/nvme1/ -2 /foo/bar/nvme1/ -d /foo/bar/hdd8tb1/ -p ABC012 -f XYZ345 >> madplotlog_test4.log

Multi-threaded pipelined Chia k32 plotter - 16f9793
Final Directory: /foo/bar/hdd8tb1
Number of Plots: 35
Crafting plot 1 out of 35
Process ID: 24702
Number of Threads: 32
Number of Buckets: 2^9 (512)
Pool Public Key: ABC012
Farmer Public Key: XYZ345
Working Directory: /foo/bar/nvme0
Working Directory 2: /foo/bar/nvme0
Plot Name: plot-k32-2021-06-14-13-52-abcxyz123456
[P1] Table 1 took 14.0462 sec
[P1] Table 2 took 79.4959 sec, found 4294922660 matches
[P1] Table 3 took 121.275 sec, found 4295043348 matches
[P1] Table 4 took 139.896 sec, found 4295105423 matches
[P1] Table 5 took 127.754 sec, found 4295051975 matches
[P1] Table 6 took 112.256 sec, found 4295115411 matches
[P1] Table 7 took 79.6454 sec, found 4295199278 matches
Phase 1 took 674.41 sec
[P2] max_table_size = 4295199278
[P2] Table 7 scan took 9.19296 sec
[P2] Table 7 rewrite took 26.7727 sec, dropped 0 entries (0 %)
[P2] Table 6 scan took 27.2039 sec
[P2] Table 6 rewrite took 43.4694 sec, dropped 581330986 entries (13.5347 %)
[P2] Table 5 scan took 26.3494 sec
[P2] Table 5 rewrite took 42.0555 sec, dropped 762004269 entries (17.7414 %)
[P2] Table 4 scan took 25.9751 sec
[P2] Table 4 rewrite took 41.1127 sec, dropped 828947124 entries (19.2998 %)
[P2] Table 3 scan took 26.0536 sec
[P2] Table 3 rewrite took 41.0379 sec, dropped 855102775 entries (19.9091 %)
[P2] Table 2 scan took 25.7834 sec
[P2] Table 2 rewrite took 40.8842 sec, dropped 865500546 entries (20.1517 %)
Phase 2 took 384.835 sec
Wrote plot header with 268 bytes
[P3-1] Table 2 took 38.0272 sec, wrote 3429422114 right entries
[P3-2] Table 2 took 50.0202 sec, wrote 3429422114 left entries, 3429422114 final
[P3-1] Table 3 took 57.1191 sec, wrote 3439940573 right entries
[P3-2] Table 3 took 57.6703 sec, wrote 3439940573 left entries, 3439940573 final
[P3-1] Table 4 took 41.786 sec, wrote 3466158299 right entries
[P3-2] Table 4 took 38.9953 sec, wrote 3466158299 left entries, 3466158299 final
[P3-1] Table 5 took 42.8739 sec, wrote 3533047706 right entries
[P3-2] Table 5 took 38.4781 sec, wrote 3533047706 left entries, 3533047706 final
[P3-1] Table 6 took 44.9517 sec, wrote 3713784425 right entries
[P3-2] Table 6 took 41.8087 sec, wrote 3713784425 left entries, 3713784425 final
[P3-1] Table 7 took 58.9838 sec, wrote 4295199278 right entries
[P3-2] Table 7 took 49.1742 sec, wrote 4294967296 left entries, 4294967296 final
Phase 3 took 562.658 sec, wrote 21877320413 entries to final plot
[P4] Starting to write C1 and C3 tables
[P4] Finished writing C1 and C3 tables
[P4] Writing C2 table
[P4] Finished writing C2 table
Phase 4 took 37.5768 sec, final plot size is 108836215983 bytes
Total plot creation time was 1659.53 sec
Started copy to /foo/bar/hdd8tb1/plot-k32-2021-06-14-13-52-abcxyz123456.plot
Copy to /foo/bar/hdd8tb1/plot-k32-2021-06-14-13-52-abcxyz123456.plot finished, took 951.733 sec, 109.058 MB/s avg.

3 Likes

If you are only using the one drive you might want to try not setting it and only use the 1 for -t. It may slow the drive down with 2 write/read instructions is you don’t use 2 drives.

I have a similar setup and @ 51/day (>1600s):

5950X
32GB (4x8) DDR4-3600
Aorus X570 elite wifi
1TB NVMe Gen 4
2 TB NVMe Gen 4

1 Like

I will give that try shortly, thanks!

Why are the buckets set so high?
Doesn’t that mean it uses less ram… Since reducing buckets requires more ram.

I assume that’s helping to improve the time somehow?

lol this aint a good result. with double 4x300gb 15k sas raid0 arrays as tmp1 and 256gb nvme as tmp2 I get 47 ppd on 10yrs old dual xeons. note that this whole rig I use is cheaper than 5950x alone!

Holy crap! I just tried this plotter and got a 48 min single plot!! My fastest time before on the GUI was ~8 hours😳
I’m STOKED

1 Like

hi, i have an old dual xeon server too running windows though. may i know the type of xeons you have and the madmax settings you are using?