Macbook Pro M1 Max plots in 47 min

I was curious what a $3500 Apple laptop would do for plotting. Maybe this has been covered before, so apologies if this is old information, but I thought it was interesting to try this.

I ran a Madmax plot just from my home directory on a 10-core, 1TB M1 Max and was able to complete a k32 plot in 47 minutes. Watching this run and comparing it to my other computer, it seems clear that the built-in storage on this laptop is insanely fast.

Creating plotters folder within CHIA_ROOT: /Users/[redacted]
Multi-threaded pipelined Chia k32 plotter - 2092041
Network Port: 8444
Final Directory: /Users/[redacted]
Number of Plots: 1
Crafting plot 1 out of 1 (2022/02/16 22:01:01)
Process ID: -1
Number of Threads: 10
Number of Buckets P1:    2^8 (256)
Number of Buckets P3+P4: 2^8 (256)
Pool Public Key:   [redacted]
Farmer Public Key: [redacted]
Working Directory:   /Users/[redacted]
Working Directory 2: /Users/[redacted]
Plot Name: plot-k32-2022-02-16-22-01-[redacted]
[P1] Table 1 took 34.6552 sec
Progress update: 0.01
[P1] Table 2 took 175.944 sec, found 4294852221 matches
Progress update: 0.06
[P1] Table 3 took 179.407 sec, found 4294665200 matches
Progress update: 0.12
[P1] Table 4 took 206.56 sec, found 4294388131 matches
Progress update: 0.2
[P1] Table 5 took 183.054 sec, found 4293802327 matches
Progress update: 0.28
[P1] Table 6 took 216.79 sec, found 4292616835 matches
Progress update: 0.36
[P1] Table 7 took 186.341 sec, found 4290340875 matches
Progress update: 0.42
Phase 1 took 1182.85 sec
[P2] max_table_size = 4294967296
[P2] Table 7 scan took 17.7309 sec
[P2] Table 7 rewrite took 50.7884 sec, dropped 0 entries (0 %)
Progress update: 0.43
[P2] Table 6 scan took 30.689 sec
[P2] Table 6 rewrite took 95.3102 sec, dropped 581520508 entries (13.547 %)
Progress update: 0.48
[P2] Table 5 scan took 26.9554 sec
[P2] Table 5 rewrite took 94.1669 sec, dropped 762313449 entries (17.7538 %)
Progress update: 0.51
[P2] Table 4 scan took 28.925 sec
[P2] Table 4 rewrite took 90.9306 sec, dropped 829126360 entries (19.3072 %)
Progress update: 0.55
[P2] Table 3 scan took 28.2425 sec
[P2] Table 3 rewrite took 91.8993 sec, dropped 855237238 entries (19.9139 %)
Progress update: 0.58
[P2] Table 2 scan took 27.5779 sec
[P2] Table 2 rewrite took 89.4755 sec, dropped 865704678 entries (20.1568 %)
Progress update: 0.61
Phase 2 took 677.863 sec
Wrote plot header with 268 bytes
[P3-1] Table 2 took 62.4104 sec, wrote 3429147543 right entries
[P3-2] Table 2 took 59.4272 sec, wrote 3429147543 left entries, 3429147543 final
Progress update: 0.66
[P3-1] Table 3 took 86.3764 sec, wrote 3439427962 right entries
[P3-2] Table 3 took 60.9231 sec, wrote 3439427962 left entries, 3439427962 final
Progress update: 0.73
[P3-1] Table 4 took 85.4422 sec, wrote 3465261771 right entries
[P3-2] Table 4 took 58.4836 sec, wrote 3465261771 left entries, 3465261771 final
Progress update: 0.79
[P3-1] Table 5 took 86.9766 sec, wrote 3531488878 right entries
[P3-2] Table 5 took 62.1041 sec, wrote 3531488878 left entries, 3531488878 final
Progress update: 0.85
[P3-1] Table 6 took 92.4453 sec, wrote 3711096327 right entries
[P3-2] Table 6 took 63.8995 sec, wrote 3711096327 left entries, 3711096327 final
Progress update: 0.92
[P3-1] Table 7 took 73.6675 sec, wrote 4290340875 right entries
[P3-2] Table 7 took 72.2165 sec, wrote 4290340875 left entries, 4290340875 final
Progress update: 0.98
Phase 3 took 866.246 sec, wrote 21866763356 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 136.477 sec, final plot size is 108772773543 bytes
**Total plot creation time was 2863.5 sec (47.7251 min)**
Started copy to [redacted].plot
Renamed final plot to [redacted].plot

I definitely don’t want to burn up the internal storage, so I won’t make this a habit!

3 Likes

I’d like to see how fast this would be using an external SSD via Thunderbolt.

And sooner will appears PCIE 5…

My 400$ server (2x xeon 2680v2 + 384GB DDR Ram) is making a plot in 26 minutes.
No need to burn the SSD of your Mac :slight_smile:

1 Like

Pretty cool. Certainly a good plotting speed!

How did you do that? I am around 29 minutes with 256MB ram and same CPUs. Practically even a bit more as without waiting for copying I had stability issues…

In term of power efficiency, you probably just broke the world record for the creation of one plot. Congratulations :slight_smile:

I run debian with 384GB ram (110GB temp1 & 220GB temp2), no SSD is involved

Wonder if using two distinct ram drives makes a difference? I am using one tmpfs with 256 MB, which works OK.

What are the setting you are using for the plots?

I’m running a single ram drive in Linux Mint, and can complete a K32 in 22 minutes all in RAM , 18 core Xeon E-2699v3 256GB DDR4

PS must get around to trying Debian, heard it’s potentially faster.

@Impuls, nothing special:

nohup chia_plot -n 165 -r 20 -u 512 -t /mnt/ram2 -d /mnt/mergerfs/plots/ -2 /mnt/ram/ -f xxx -c yyyy > /tmp/plot.txt &

For me 512 buckets were slower than 256. Guess I will give 512 another chance. Are you using a MadMax build that does not have the -K option or are you not using it intentionally?

My bad,

I am using 256 buckets atm, the command is from a shell file when I had only 128GB in the beginning and 512 buckets made more sense.
If you omit -k standard size is k32.

Was referring to uppercase K:

-K, --rmulti2 arg    Thread multiplier for P2 (default = 1)

Without that parameter you most likely will observe quite low CPU load in phase 2 because not all cores are used. If your MadMax build does support the parameter you might be able to squeeze out few more seconds.

At least that is the experience I made… considering how “slow” my plots are I could be wrong :smiley:

Ah i see. Never needed to play with that. Build is fairly old though (and was built on a ubuntu system)

I’d like to see how fast this would be using an external SSD via Thunderbolt.

Yeah, me too. So I got my hands on a Plugable TBT3-NVME512 Thunderbolt 3 drive and re-ran the same type of plot. It took exactly twice as much time. I could see while watching it that the CPUs weren’t pegged like before. The internal storage on these Macs is incredible.

Multi-threaded pipelined Chia k32 plotter - 2092041
Network Port: 8444
Final Directory: /Volumes/TBT3_SSD/chia/plots/
Number of Plots: 1
Crafting plot 1 out of 1 (2022/02/19 11:34:20)
Process ID: -1
Number of Threads: 10
Number of Buckets P1:    2^8 (256)
Number of Buckets P3+P4: 2^8 (256)
Pool Public Key:   [redacted]
Farmer Public Key: [redacted]
Working Directory:   /Volumes/TBT3_SSD/chia/
Working Directory 2: /Volumes/TBT3_SSD/chia/
Plot Name: plot-k32-2022-02-19-11-34-870[redacted]
[P1] Table 1 took 47.7344 sec
Progress update: 0.01
[P1] Table 2 took 240.242 sec, found 4295050962 matches
Progress update: 0.06
[P1] Table 3 took 405.926 sec, found 4295017288 matches
Progress update: 0.12
[P1] Table 4 took 436.289 sec, found 4295032340 matches
Progress update: 0.2
[P1] Table 5 took 335.651 sec, found 4295062819 matches
Progress update: 0.28
[P1] Table 6 took 349.061 sec, found 4295060949 matches
Progress update: 0.36
[P1] Table 7 took 246.102 sec, found 4295050833 matches
Progress update: 0.42
Phase 1 took 2061.15 sec
[P2] max_table_size = 4295062819
[P2] Table 7 scan took 22.7586 sec
[P2] Table 7 rewrite took 126.193 sec, dropped 0 entries (0 %)
Progress update: 0.43
[P2] Table 6 scan took 27.5799 sec
[P2] Table 6 rewrite took 154.245 sec, dropped 581346211 entries (13.5352 %)
Progress update: 0.48
[P2] Table 5 scan took 91.032 sec
[P2] Table 5 rewrite took 218.007 sec, dropped 762040365 entries (17.7422 %)
Progress update: 0.51
[P2] Table 4 scan took 29.9228 sec
[P2] Table 4 rewrite took 229.76 sec, dropped 828838099 entries (19.2976 %)
Progress update: 0.55
[P2] Table 3 scan took 37.4438 sec
[P2] Table 3 rewrite took 317.748 sec, dropped 855094575 entries (19.909 %)
Progress update: 0.58
[P2] Table 2 scan took 59.7561 sec
[P2] Table 2 rewrite took 267.979 sec, dropped 865629090 entries (20.1541 %)
Progress update: 0.61
Phase 2 took 1584.41 sec
Wrote plot header with 268 bytes
[P3-1] Table 2 took 192.76 sec, wrote 3429421872 right entries
[P3-2] Table 2 took 127.564 sec, wrote 3429421872 left entries, 3429421872 final
Progress update: 0.66
[P3-1] Table 3 took 162.595 sec, wrote 3439922713 right entries
[P3-2] Table 3 took 149.017 sec, wrote 3439922713 left entries, 3439922713 final
Progress update: 0.73
[P3-1] Table 4 took 143.102 sec, wrote 3466194241 right entries
[P3-2] Table 4 took 133.083 sec, wrote 3466194241 left entries, 3466194241 final
Progress update: 0.79
[P3-1] Table 5 took 135.165 sec, wrote 3533022454 right entries
[P3-2] Table 5 took 126.39 sec, wrote 3533022454 left entries, 3533022454 final
Progress update: 0.85
[P3-1] Table 6 took 124.674 sec, wrote 3713714738 right entries
[P3-2] Table 6 took 112.459 sec, wrote 3713714738 left entries, 3713714738 final
Progress update: 0.92
[P3-1] Table 7 took 197.983 sec, wrote 4295050833 right entries
[P3-2] Table 7 took 175.181 sec, wrote 4294967296 left entries, 4294967296 final
Progress update: 0.98
Phase 3 took 1781.75 sec, wrote 21877243314 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 241.151 sec, final plot size is 108835899633 bytes
Total plot creation time was 5668.53 sec (94.4755 min)
Started copy to /Volumes/TBT3_SSD/chia/plots/plot-k32-2022-02-19-11-34-870[redacted].plot
Renamed final plot to /Volumes/TBT3_SSD/chia/plots/plot-k32-2022-02-19-11-34-870[redacted].plot

…aaaaand I just realized this drive is formatted ExFAT.

With a proper macOS file system, it finished in 62 minutes.

1 Like

I was going to say this speed seems slower than I would have expected because I have the old M1 MBP and was able to finish a K=32 plot on an external SSD in 46.8 minutes. I will be getting my M1 Max with 64GB in a few weeks and will re-post my results here. See also: https://www.reddit.com/r/chia/comments/t8azie/imac_with_intel_cpu_vs_macbook_pro_with_m1_cpu/

Did you try with an active cooling below the CPU? As far as i know, Mac slows the processing power to reduce the heat and avoid any issue. This is why M1 Mac book air doesnt have a fan.

I did nothing to cool off my MacBook Pro M1. However, I did put the external SSD on a marble countertop as a massive heatsink and I also put an active cooling fan on top of the SSD enclosure.