Mad Max Plotter, some test results, some general info

Chia Plotting
653

So, i can’t create hardware raid 0 with 2 pci-e drives, software raid 0 works like single disk speed. Decided to use ramdrive on Windows.

Process ID: 9680
Number of Threads: 20
Number of Buckets P1: 2^9 (512)
Number of Buckets P3+P4: 2^7 (128)
Working Directory: q:
Working Directory 2: u:
Plot Name: plot-k32-2021-07-08-05-48-25eef0f19d558549d27201c97051698bdbb0886c0770e1fb29fb992e50d92fcc
[P1] Table 1 took 20.281 sec
[P1] Table 2 took 131.664 sec, found 4294815212 matches
[P1] Table 3 took 208.756 sec, found 4294539690 matches
Copy to o:\Plots\plot-k32-2021-07-08-05-09-f862a0b982348895084716ac5e0f043223e5deef40e7135fcf4feb2c7d4aae35.plot finished, took 518.071 sec, 200.332 MB/s avg.
[P1] Table 4 took 202.557 sec, found 4294080119 matches
[P1] Table 5 took 195.256 sec, found 4293228433 matches
[P1] Table 6 took 170.869 sec, found 4291529357 matches
[P1] Table 7 took 146.111 sec, found 4288068102 matches
Phase 1 took 1075.83 sec
[P2] max_table_size = 4294967296
[P2] Table 7 scan took 19.5669 sec
[P2] Table 7 rewrite took 36.7059 sec, dropped 0 entries (0 %)
[P2] Table 6 scan took 32.4485 sec
[P2] Table 6 rewrite took 54.9772 sec, dropped 581725204 entries (13.5552 %)
[P2] Table 5 scan took 31.0244 sec
[P2] Table 5 rewrite took 52.5355 sec, dropped 762494288 entries (17.7604 %)
[P2] Table 4 scan took 30.6112 sec
[P2] Table 4 rewrite took 56.2178 sec, dropped 829231191 entries (19.311 %)
[P2] Table 3 scan took 30.4667 sec
[P2] Table 3 rewrite took 58.1732 sec, dropped 855385155 entries (19.918 %)
[P2] Table 2 scan took 30.4158 sec
[P2] Table 2 rewrite took 52.738 sec, dropped 865760352 entries (20.1583 %)
Phase 2 took 488.265 sec
Wrote plot header with 268 bytes
[P3-1] Table 2 took 51.3687 sec, wrote 3429054860 right entries
[P3-2] Table 2 took 42.4191 sec, wrote 3429054860 left entries, 3429054860 final
[P3-1] Table 3 took 63.2727 sec, wrote 3439154535 right entries
[P3-2] Table 3 took 45.1291 sec, wrote 3439154535 left entries, 3439154535 final
[P3-1] Table 4 took 63.986 sec, wrote 3464848928 right entries
[P3-2] Table 4 took 45.6382 sec, wrote 3464848928 left entries, 3464848928 final
[P3-1] Table 5 took 64.3813 sec, wrote 3530734145 right entries
[P3-2] Table 5 took 46.8304 sec, wrote 3530734145 left entries, 3530734145 final
[P3-1] Table 6 took 67.6044 sec, wrote 3709804153 right entries
[P3-2] Table 6 took 48.1065 sec, wrote 3709804153 left entries, 3709804153 final
[P3-1] Table 7 took 58.1739 sec, wrote 4288068102 right entries
[P3-2] Table 7 took 55.7968 sec, wrote 4288068102 left entries, 4288068102 final
Phase 3 took 653.406 sec, wrote 21861664723 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 90.8212 sec, final plot size is 108741957936 bytes
Total plot creation time was 2308.42 sec (38.4737 min)

Cpu slowdowns process for now. My previous result was 68 min without ramdrive. I wanna try linux, maybe can beat record)

P.S. maybe i should overclock cpu and mem?

654

To use MM with to plot portable plots, can you just install the normal way, or do you have to use the pool-puzzles branch?
Not too familiar with git but it seems that branch has not been committed to main?

655

Oh yes, that was a long time ago. In my experience, if it crashes after starting the answer is to reformat the drives you are using as temp drives and ensure there is enough space on your final drive.

656

Right now you should use pool-puzzles brunch. It’s not migrated yet.

1 Like
657

It gives 2 options (you can use 1 or the other, but not both)

Option 1, use pool contract address + farmer public key:
-c *** -f ***

Option 2, use pool public key + farmer public key (no idea yet if this becomes portable):
-p *** -f ***

658

pool-puzzles branch has been committed, so you can install normal way now

1 Like
659

Hi everyone, I have been reading through this thread for half an hour now and couldnt find a solution to my problem.The best plot time I can get with mad max is 103 minutes. It seems like others here are getting closer to and less than 60 minutes/plot. I use taskset to limit the process to only one of my CPUs since letting mad max run wild on both CPUs gives me plot times in the 140 - 150 minute range.
My current system setup:

Ubuntu 20.04 LTS
Cisco UCSC-C240-M4L
CPU: Dual E5-2697 V3
Ram: 64 Gigs of DDR 4 2333 Mhz
Temp folders: 2 x 4 SAS 10k drives running in raid 0

Any help or suggestion would be appreciated.

Thanks

660

I decided to buy an NVME to test plotting speed with it as temp 2 and got down to 88 minutes/plot. It is connected through a PCIE adapter to a 4x PCIE slot. Details of the ssd below.

661

maybe add 64G ram to form a ramdisk as temp2 will help

662

the best choice for you is add 64GB ram ro total 128GB you will get very fast plotting speed.

663

The issue I have with buying more ram is that with a dual cpu system I have to distribute the ram evenly among both CPUs, so if I buy 64 gigs, half that will go in the slots for CPU 1 and the other half to CPU 2. In that case I am not sure how effective a ramdisk would be since I set the affinity of mad max to only one CPU.

664

best buy two 32GB ram to get better performance, the plotting time should less than 45 minutes. can reduce 75% writing for SSD.

665

My server uses 2333 mhz DDR4 ECC registered ram. cheapest I have found for 16 gig sticks is 100 bucks Canadian. I will probably go that route which is gonna cost me $400 cad, but for now my budget is maxed out.

666

how many plots you want to plotting? you can buy two 500NVME disk, you can run two processe, then you can get two plots in 100 minutes, but the SSD Endurance you need consider it.

667

Right now I would be happy with 1 hour plot times, so about 24 plots a day. That would be about one 3 TB sas hard drive a day. Thanks for the advice by the way.

668

Ok so I started some parallel plotting last night and today I see that it takes one CPU 1 hour 53 minutes to complete a plot and the second CPU takes 2 hours and 3 minutes. I think this is best I will be able to do speed wise with my setup. I will stick with this configuration until I have the budget for some more ddr4 ram.

669

Hi Harris

Your records are very impressive

I would like to know how to had distributer your thread as what I could see u have 4 nvme
Is that 20 threads is for each NVME u allocated with a total of 80 threads
What’s your strategy u adopt

I am having the following configuration please guide me the best strategy to boost the speed of plotting

Ryzen 3950x 16c/32 threads
64 GB RAM
1 980 PRO 2TB
1 ENOMUTUS FUZEDRIVE 1.6 TB
1 970 PRO 1 TB

I running with the following strategy

  1. with 1× 980 pro 2 TB: 12 threads, both temp drive same nvme,
  1. with 1× fuzedrive pro 1.6 TB: 10threads, both temp drive same nvme
  2. with 1× 970 pro 1 TB: 10 threads, both temp drive same nvme

With that I am spreading all my 32 threads and the performance with that is as follows

  1. For 980 pro: plot duration around 75-78 mins
  2. For fuzedrive: plot duration around 80 mins
  3. For 970 pro: plot duration around 90-100 mins

So in summary: I am getting an average of 85 mins for 3 plots in parallel so average of around 28 mins per plot

What I have seen since yesterday, that my 970 nvme time is been increased from 85 mins to 103 mins I don’t know why any troubleshooting for the same I tried trimming and formating as well and the temp is also within the limit it was floating in the range of 42deg c even though the time is not reducing

Could you please share your view point on above how to reduce the plotting time for 970 pro and in general which strategy I should adopt with above configuration so as to reduce the plotting time overall in general, I could see the time your stated for your machine is very impressive, appreciate if you can share your strategy of plotting with the configuration you have

Awaiting for your feedback
Thanks

671

with similar setup of 10850K ,64GB DDR4, 2T Samsung 970 evo *2, I mange to have around 46 minutes , BTW, all default with no overclocking at all.

672
  • Cheap machinist motherboard for dual xeon v3 processors.
  • Dual xeon 2640v3 (hack unlook turbo boost) all cores at 3.4ghz when needed.
  • 64RAM 2133mhz (1866 mhz limited due processors)
  • Ubuntu 21 system.
  • Temp disk: Crucial SSD 2TB
  • 2 Temp disck: Crucial p5 ITB nvme

Process ID: 46788
Number of Threads: 32
Number of Buckets P1: 2^8 (256)
Number of Buckets P3+P4: 2^8 (256)
Pool Puzzle Hash: 0361c3158a748********************
Farmer Public Key: a86143e7d0a1422d53
Working Directory: /media/nouser/crucial2000/TmpDirParcela/
Working Directory 2: /media/nouser/crucialp5/TmpDirParcela/
Plot Name: plot-k32-2021-07-14-12-20-2a63ca94bb8624d70560a7*************
[P1] Table 1 took 37.2993 sec
[P1] Table 2 took 149.594 sec, found 4295019206 matches
[P1] Table 3 took 243.061 sec, found 4294973608 matches
[P1] Table 4 took 310.649 sec, found 4295043683 matches
[P1] Table 5 took 313.305 sec, found 4295076218 matches
[P1] Table 6 took 253.861 sec, found 4295028827 matches
[P1] Table 7 took 140.585 sec, found 4295004139 matches
Phase 1 took 1448.94 sec
[P2] max_table_size = 4295076218
[P2] Table 7 scan took 17.9 sec
[P2] Table 7 rewrite took 81.5235 sec, dropped 0 entries (0 %)
[P2] Table 6 scan took 60.0028 sec
[P2] Table 6 rewrite took 96.9637 sec, dropped 581331640 entries (13.535 %)
[P2] Table 5 scan took 76.8112 sec
[P2] Table 5 rewrite took 84.5557 sec, dropped 762034414 entries (17.742 %)
[P2] Table 4 scan took 80.3267 sec
[P2] Table 4 rewrite took 65.4072 sec, dropped 828898486 entries (19.299 %)
[P2] Table 3 scan took 79.859 sec
[P2] Table 3 rewrite took 65.1829 sec, dropped 855056077 entries (19.9083 %)
[P2] Table 2 scan took 80.1993 sec
[P2] Table 2 rewrite took 57.6427 sec, dropped 865610694 entries (20.1538 %)
Phase 2 took 863.606 sec
Wrote plot header with 252 bytes
[P3-1] Table 2 took 61.1798 sec, wrote 3429408512 right entries
[P3-2] Table 2 took 45.8331 sec, wrote 3429408512 left entries, 3429408512 final
[P3-1] Table 3 took 79.3036 sec, wrote 3439917531 right entries
[P3-2] Table 3 took 38.5586 sec, wrote 3439917531 left entries, 3439917531 final
[P3-1] Table 4 took 82.7937 sec, wrote 3466145197 right entries
[P3-2] Table 4 took 38.9191 sec, wrote 3466145197 left entries, 3466145197 final
[P3-1] Table 5 took 85.5981 sec, wrote 3533041804 right entries
[P3-2] Table 5 took 37.3983 sec, wrote 3533041804 left entries, 3533041804 final
[P3-1] Table 6 took 87.0852 sec, wrote 3713697187 right entries
[P3-2] Table 6 took 42.0147 sec, wrote 3713697187 left entries, 3713697187 final
[P3-1] Table 7 took 127.21 sec, wrote 4295004139 right entries
[P3-2] Table 7 took 71.4367 sec, wrote 4294967296 left entries, 4294967296 final
Phase 3 took 801.622 sec, wrote 21877177527 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 111.787 sec, final plot size is 108835629458 bytes
Total plot creation time was 3226.09 sec (53.7682 min)

1 Like
673

Im having same issue of degrading performance. Have you solved it?