CPU + GPU plotter.
update:
v0.7 have performance increase for 24G gpu like rtx 3090. perf increase 40%. best result : 400 sec /plot.
new Announcements about "compressed plot"
Few facts about "compressed" plot
a.) standalone 32G ram + 256 SSD
b.) standalone 128G ram + 128G ssd
c.) local 32G ram + NRD disk 128G x2 pc
d.) 128G cpu + 4G+ vram GPU.
How to run:
get help:
./arc_plot --help
cpu: 32G + SSD
./arc_plot -r cpu_core_num
cpu:110G
./arc_plot -R ...
GPU: default setting
./arc_plot -G ...
GPU: using multiple gpu like gpu 0,1,2,3
./arc_plot -G --gpuid "0123" ...
GPU: using 5 gpu context threads for 12G gpu
./arc_plot -G --gthreads 5 ...
above example with short parameters require apend -r cpu_core_num -n num_of_plots -u 256 -v 256 -t your_ssd_path -d your_distination_path -c your_contract -f your_farm_key at the end.
please check:
https://github.com/smartbitcoin/arc_plot/blob/main/release_notes
GPU plotting requirements:
hardware:
cpu: 64bit x86 cpu with pci-e 3.0 x16 bus
ram: 128G ram
gpu: nvidia cuda compatibility > 7.5 GPU ( 20 or 30 series) with 4G+ vram, support up to 8 GPUs.
ssd: *MLC or TLC with good continuous sequence R/S speed. QLC should use raid0. ( found some SSD have fwrite() error if speed lag GPU too much.)
Software:
ubuntu 22.04. ( or any linux distribution with kernel compiled by gcc11 )
please submit your benchmark by a github ticket
arc_plot 0.7 for ubuntu 22.04 download link:
https://github.com/smartbitcoin/arc_plot/raw/main/arc_plot.0.7.ubuntu22.04.tar.bz2
windows user please check: ( not sure how wsl support gpu though.)
https://github.com/smartbitcoin/arc_plot/wiki/arc_plot-windows-wsl-guide.
a benchmark can be also found in wiki:
https://github.com/smartbitcoin/arc_plot/wiki
briefly compare speed of gpu vs cpu. rtx 3090 = 2 x i9 13900kf for phase1.
https://www.youtube.com/watch?v=CfmZbIM17ZQ&t=87s
Special thanks to Bladebit as arc_plot learned lots of good design from them and here show highly respect for these complicated project and pioneer's hardwork.
ARC_PLOT is freeware and if you like this idea and love the arc_plot freeware, please don't forget donate some XCH to support this project.
xch12qge50ttf5c6lcx7at3295q92wndfapyuvr920j88zdnfqzc8ggqaja0dn