Comparing plot speeds across CPUs

I have been running my surface pro 7 with an external Orico 20gbps USB enclosure and a 2tb 980 pro for non-stop parallel (4 hour staggered) 4t/8gb plots for quite a while now… it does really well!

I especially like that since the plot disk space requirements are pretty low, it can pile up 9-12 plots on the output folder of that external USB drive before I copy them off, and since the machine has two USB ports (one USB-C, one USB-A) it copies the plots off at 500mb/sec which is altogether rather nice…

5950x, 32gb 3600CL16, MP510 4tb (temp1), 500gb SN850 (OS)

HDD for storage

Win GUI, 1hr stagger, 4000MiB, 18t

21:53
Time for phase 1 = 5003.076 seconds. CPU (496.760%) Tue May  4 23:17:03 2021
Total time = 16425.433 seconds. CPU (207.770%) Wed May  5 02:27:26 2021

22:53
Time for phase 1 = 5492.738 seconds. CPU (467.260%) Wed May  5 00:25:13 2021
Total time = 17872.484 seconds. CPU (197.260%) Wed May  5 03:51:33 2021

23:53
Time for phase 1 = 5982.631 seconds. CPU (412.770%) Wed May  5 01:33:24 2021
Total time = 18926.703 seconds. CPU (182.630%) Wed May  5 05:09:08 2021

00:53
Time for phase 1 = 6458.575 seconds. CPU (362.660%) Wed May  5 02:41:20 2021
Total time = 19573.451 seconds. CPU (170.380%) Wed May  5 06:19:55 2021

01:53
Time for phase 1 = 6648.331 seconds. CPU (352.690%) Wed May  5 03:44:30 2021
Total time = 19151.137 seconds. CPU (172.710%) Wed May  5 07:12:53 2021

02:53
Time for phase 1 = 6583.250 seconds. CPU (353.730%) Wed May  5 04:43:25 2021
Total time = 18055.258 seconds. CPU (179.800%) Wed May  5 07:54:37 2021

03:53
Time for phase 1 = 7073.209 seconds. CPU (336.030%) Wed May  5 05:51:36 2021
Total time = 17050.584 seconds. CPU (189.290%) Wed May  5 08:37:53 2021

04:53
Time for phase 1 = 6570.544 seconds. CPU (371.170%) Wed May  5 06:43:13 2021
Total time = 15458.703 seconds. CPU (209.300%) Wed May  5 09:11:21 2021

Really want to get plotman running but for some reason if i do exactly the same run using Jim’s begginers build batch, i can’t even get 1 plot to finish in 10hrs. CPU perf for me in ubuntu is dire, even with kernal 5.12. Fresh install today and swapping out to 64gb 3600CL16, see what happens.

I’ll most likely stay on win if there’s and auto run script floating about. I work away every 2 weeks so really want to get something automated up and running asap. If i can do that the i can use clock tuner. (Insane L3 cache read/write/copy).

From data above it looks like i could do with an extra 25min stagger (1500s) but need to run again with 64gb ram and see what difference if makes upping mem to 6750MiB. I’ll probably just que 29 plots, run, see what happens in 24hrs. Still not figured out how to run more than 1 staggered que at a time…(probably need to look into CLI for win).

Off topic slightly, really curious to hear from Storage JM about his brother in law 10850k build. 4TB/day in win gui is awesome, would love to hear what the config is for doing that. Guessing 2nd NVME for temp2 also helped… @storage_jm

1 Like

Is that 18 threads per plot? That’s too high for a 5950x as it’s crossing the CCX boundary. I believe 8 threads max has been recommended here.

Your 32GB RAM is a little stretched at the 4000 setting for 8 parallel plots, and your SSD’s sustained write performance may also be a bottleneck.

1 Like

Yep, not something i’ll run again, more for curiosity on core behaviour.

Indeed, i think 6-7 was the sweet spot but wanted to run 4gb/plot, 8 plots to compare running win gui vs ubuntu with Jims batch just to make sure everything was still alright in win. Looking at the times, maximum plots on the go at once was 6 so less than 24gb.

I’ve got a few 2tb sn750’s arriving in a few weeks which were for something else so will grab a gen 3 hyper, populate it and use temp2 onto those drives, should also help cut time down.

Still lots to play with / figure out for me. Getting auto plotting sorted is a must whether that be win or ubuntu. Finding it all really interesting regardless of xch rewards although slow plot time in ubuntu is frustrating me now.

Are there any specific reason for choosing MP510 instead of MP600?

MP510 is cheaper. I think that is the only reason.

Price. Wasn’t chosen for chia, bought it before i knew chia existed and got it for a ridiculously low price!

Found out about SWAR yesterday so have moved to powershell. Just doing a 2x3 to see what’s going on with it and output is 2 plots/2hrs (21k/s each plot).

And it might actually perform much better for this work load then the MP600, at least in my testing (of the new “B” revision) they have consistently done well!

3 posts were merged into an existing topic: Optimizing plots on a 5950x (or any other 16c/32t CPU)

@codinghorror
Hi CH, do you have an update regarding your 12 plots in parallel?
What optimal settings did you discover for the 5950x? Can you post some new numbers?
Thank you.

1 Like

Click on the link directly above your post :point_up_2:

Hi, I recently bought 64 GB of 2666 MHZ RAM to plot with (TEAMGROUP Elite DDR4 64GB Kit (2 x 32GB) 2666MHz), but my plotting speeds seem to be rather slow. I’m running on an MSI MPG x570 Gaming Edge Wifi, Ryzen 3900x, 1 TB Samsung 970 EVO, and a Crucial - P2 2TB 3D NAND. My CPU usage seems rather low rarely spiking above 50%, I’m only using around 32 GB of RAM, and my SSD write speeds fluctuate anywhere from 5MB/s to 500MB/s. I’ve been plotting around 7 in parallel with a 90 minute delay using 4 threads each and 4000 MB of RAM and each plot takes upwards of 14-15 hours. My SSD temps sit around 50 C so I don’t believe they’re throttling. Any insights on how to improve these times? Could it be a RAM speed bottleneck?

Highly unlikely to be RAM speeds! So you are doing a lot of plots to two destinations. My money is on the Crucial P2 being weird I/O wise; you might want to look up stats on it.

SSD temp is okay, but what is CPU temp?
Your ram is more than enough.

Try 1 plot on 970EVO with the final destination at itself and see how long it take.

Hi, I have Ryzen 9 12/24, 32 GB RAM and 2x samsung 980 pro 1 TB - the best speed plot?

You should be able to get around 5 to 6 hours running ~8 in parallel. But this assumes a well tuned Linux install running plotman.

i7-3770K @ 3.5 GHz, 4core 8thread, 32GB 1333 RAM, 1TB WD 3D SSD SATA 6G, 300 min delay, 6000 ram, 3 threads, Farm is 18TB WD Elements USB 3.0 (6G)

Total time = 32798.003 seconds. CPU (121.170%) Sat May 15 08:09:23 2021
Total time = 34124.709 seconds. CPU (118.860%) Sat May 15 13:31:28 2021
Total time = 33591.081 seconds. CPU (123.240%) Sat May 15 18:22:35 2021

Using GUI 1.1.5

My times have varied from 8 to 14 hours with different ram, threads and delays. I have a list of 50 if you are interested.

1 Like

Hi there, my modest contribution :slight_smile:

Intel I9 9900K (All 5ghz WC) 8c/16t, 6 parallels Plots, 32go Ram, 3300 per plot

  • 3 plotting drive
  • 1 Intel P3700 Nvme 800go
  • 2 intel S4610 Sata 920go

planning to fill my 89tb farm in 40 days, waiting for the pool update to start, only testing and tuning for now.

I use only 12threads for now, my computer is doing other things than plotting.
I’m going more on sustainability and durability than speed and throwing nvme away (Massey Ferguson > Ferrari :rofl:)

I’m getting 18 to 20 plots a day, the variation is due to the intel nvme beeing way faster than the two sata ssd, and sometimes “overlaping” their phase 1

As far as durability goes, my goal is to fill my 89tb of drive with plot and then planing to move to sas datacenter grade mechanical hdd for plotting, but massively parallel (with hpe server, rack format) Seagate XE14 seemed to be the ideal way to go.

Maybe selling the nvme / ssd’s when done, the price already go way above the regular market price and they may farm a lot more plot than I need… we’ll see what the future may be !

A post was split to a new topic: What does the percent of CPU time mean in total time?

Dell Precision Workstation
2 x Intel Xeon Silver 4215R CPU
128 GB RAM
8 x 1 TB Samsung 980 Pro NVMe

30 x K32 Plots Parallel = 24 hours.