Plotting speed issue.. only 10 plots a day on i7 10700?

Hi guys, I have trouble plotting my rig.

As most people on a budget I use a consumer computer to farm and plot.

So my CPU is i7 10700k and atm is hardly creating 15plots a day. I read some people manage to have over 30 plots from the same cpu. So there are only two reasons why I have so low performance.

  1. Wrong plotting technique.

  2. Bad components.

I have 32gb of ram 3200 speeds, good motherboard (ASUS). But I also have only 2 512gb m2 disks running 4 plots in parallel.

When I run only one plot, speeds are about 5 hours per finished plot. When I run 4 plots, speeds are around 10 hours per plot.

I have tried different setups, and even tried to run 9 plots in parallel using HDD to help but the final result was 15 plots per day.

What settings for plotting should I use? Do I need to change my m2 disks? I tried using 2-16 treads of load but no difference. Tried different ram settings, not much of the difference.

What am I doing wrong? I even plot right after phase one 1… and still not much success.

Thank you for any advice.

Andrew

1 Like

Well, I think parallel plotting is the best You can do, but your ssd storage is too small to reach more plots in parallel. Have similar quality of CPU (but AMD Ryzen) and plotting in parallel 7 is the most I can reach. It takes around 10hours, so I am able to create around 17 per day with AMD Ryzen 3900, NVMe m2 2TB SSD and 32 GB RAM. Maybe it is possible to boost it even better, but to be honest - it will change nothing if You will plot 15 or 20 a day.

Hmm. Interesting. I thought everyone with Ryzen 9 plot around 20-30 plots per day… I’m kinda disinformed. My actual plotting time is 9-16 plots per day. In average around 12.

How many plots you do with Ryzen 3900 in average?

Thank you

Around 18 per day. But still would be able to do more I think, when I would buy additional SSD disk, but it is simply not worth.

Confusing low numbers from the perspective of a 5900x-owner.
Ive been doing 24/day while gaming in 4k,
50/day definitely possible if maxing out.

50 per day, let’s assume: 25 in parallel 2 times. 25*256 GiBs > 6 TiBs temporary space needed. I have 2 bro.

Sounds true to me. Especially what I have heard from other guys. Probably 25-35 is your chill average.

I’m doing 20 plots per day with i9 9900k, 1x 2tb nvme, 50gb ram.
The problem is most likely to use a small amount of NVME. If you use 1x2 tb nvme, I believe you will do 20 plots per day.
I wonder how to make 30 plots with the same hardware.

This is exactly my question too. I tried different techniques, but that makes it even worse (8 plots per day)…

I know one guy has i7 10700 and on windows had 24 plots per day with 2x2tb m2 discs…

Soon I would have more information how to plot as much as possible, I would be happy to share.

It seems running plots in paralel at the same time is the way to go. But its good to run 4 plots at the same time and after phase 1, another 4 plots…

Kinda lost in settings, but would cheek logs.

The problem is not the cpu but the ssd storage you are using. I get the fastest per plot time with 4 gb ram and 4 threads per plot which might help but your temp storage is your bottleneck not the cpu.

1 Like

Like nontechguy said 1x 2tb is too little temp space. Also take note of the ssd speed. Not the advertised speed on the box, but the sustained write speed over longer period. Reviewers like tomshardware include those numbers in their tests.
2x 1TB in RAID0 is usually better than 1x2TB for that reason, more write speed.

I am plotting on ryzen 9 3900x, 64GB RAM, 3x 1TB WD black SN750 in RAID0, get 32 plots a day average.
12 plots, 4t, 4000ram, 45 min stagger, max 6 in phase1, windows using Swar plotmanager

Are you on chia 1.1.6, and are your NVME drives absolutely completely empty before you start, and are you staggering ? Maybe try staggering job kickoff by 90 minutes?

e.g.,

  1. start job 1 (uses nvme #1)
  2. wait 90 minutes
  3. start job 2 (uses nvme #2)
  4. wait 90 minutes
  5. start job 3 (uses nvme #1)
  6. wait 90 minutes
  7. start job 4 (uses nvme #2)

(I assume the jobs themselves will immediately repeat when they finish, e.g.,

chia plots create -k 32 -n 99999

Sometimes people forget to clear out their NVME drives from a partial failed run, and that can dramatically slow things down.

Works better, thank you very much.

I have the same processor, 64gb and 2 M2’s, 1tb + 2b, 32 plots / 24 hours, 8 plots in paralel.
I am sure that you only need some vents, most probably your ssd’s are going in thermal threat limits (depends a lot on the motherboard, if the m2 slots are very close to the processor). Also you can change the processor pasta.

1 Like

I am using i9 10900KF, 32 GB RAM 4000MHz, 2 NVMe Samsung 980PRO 1TB and I’m getting bad results. I am mining ETH on the same machine but I guess that uses just the GPU. Could you please share your config? How much delay are you applying while plotting in parallel? Thanks

1 Like

hi

How many plots do you make per day?

Did 7 in parallel, took 14 hours. 4200 RAM for each, and 2 threads. Is that normal?

ATM I’m plotting about 12-13 plots per day. Seems the problem might be in M2 disks… so testing now the new ones and let ya all know

Yep, 7 in paralel should run ok with 32 ram

Yes, I mine on the GPU as well, and it has very little impact on my Chia times (maybe 1% or 2% max)