smartbitcoin / arc_plot

CPU & GPU plotter
20 stars 1 forks source link

[Benchmark] rtx 4080 stock, amd 5950x, 3600MH/s ram #16

Open Motophan opened 1 year ago

Motophan commented 1 year ago

[gentoo@arch5950x arc_plot]$ ./arc_plot -t /mnt/nvme01/ -r 16 -f [redacted] -c [redacted] -G


** ARC_PLOT CPU + GPU plotter version: 0.7-d38e551***


  Final Directory: /mnt/nvme01/ Number of Plots: 1 GPU plotting 1 out of 1 (2023/01/13 11:52:06) Found cpu ram size: 131817764 Found gpu: 1 Found gpu:0 ram size: 16493117440 GPU plotting in G24 mode: 0 using gpu 0:[ NVIDIA GeForce RTX 4080 ] for plotting. Process ID: 2409 Number of Threads: 16 Number of Buckets P1: 2^8 (256) Number of Buckets P3+P4: 2^8 (256) Pool Puzzle Hash:
ligma Farmer Public Key: ligma Working Directory: /mnt/nvme01/ Working Directory 2: /mnt/nvme01/ Plot Name: plot-k32-2023-01-13-11-52-119110e2bc5dee543a9e41ee729764ed58e696ef06f6811d29be841cb19bb522  


Phase 1: creating plot


[P1] Table 1 took 7.9771 sec [P1] Table 2 took 15.3115 sec, found 4294918550 matches [P1] Table 3 took 31.612 sec, found 4293109796 matches [P1] Table 4 took 39.4745 sec, found 4290518298 matches [P1] Table 5 took 38.8434 sec, found 4285907734 matches [P1] Table 6 took 29.4285 sec, found 4276865724 matches [P1] Table 7 took 20.3321 sec, found 4258815813 matches Phase 1 took 186.896 sec  


Phase 2: compressing plot


[prescan] took 18.5752 sec G24 MODE: 0 [P2] Table 2.a took 11.9833 sec, wrote 0 right entries [P2] Table 2.b took 25.0771 sec, wrote 3427325037 left entries, 3427325037 final [P2] Table 3.a took 33.7729 sec, wrote 3436107292 right entries [P2] Table 3.b took 24.4721 sec, wrote 3436107292 left entries, 3436107292 final [P2] Table 4.a took 33.2438 sec, wrote 3459210725 right entries [P2] Table 4.b took 24.5901 sec, wrote 3459263491 left entries, 3459263491 final [P2] Table 5.a took 34.1582 sec, wrote 3521034720 right entries [P2] Table 5.b took 25.1385 sec, wrote 3521034720 left entries, 3521034720 final [P2] Table 6.a took 35.4268 sec, wrote 3693158863 right entries [P2] Table 6.b took 26.2747 sec, wrote 3693158863 left entries, 3693158863 final [P2] Table 7.a took 21.0906 sec, wrote 4258815813 right entries [P2] Table 7.b took 22.2449 sec, wrote 4258815813 left entries, 4258815813 final Phase 2 took 341.818 sec, wrote 21795705216 entries to final plot  


Phase 3: writting plot


[P3] Starting to write C1 and C3 tables [P3] Finished writing C1 and C3 tables [P3] Writing C2 table [P3] Finished writing C2 table Phase 3 took 39.0596 sec, final plot size is 108343922948 bytes Total plot creation time was 569.075 sec (9.48459 min)  


Plotting Done!


 

smartbitcoin commented 1 year ago

thanks for submit.

phase 1 rtx 4080 have similiar phase 1 timing vs rtx 3090. but phase 2 take more time as rtx 3090 have 24G vram. and most of compress work was in vram.

Motophan commented 1 year ago

I tried switching -u and -v to improve speed for phase 2 as bladebit on ryzen prefers buckets 64 https://github.com/Chia-Network/bladebit/issues/202 and (through personal experience) madmax likes -u 512 -v 512

Do you have a discord?