Does it really make difference 5900x 2x 2tb 980 pro 40 gb Ram

[nem0n]

Hi Guys,

So I was looking into this project and trying to understand if it is really benefical to my case. Currently on windows I am plotting.
Here is how its going approx I am getting 27-30 plot per day.

So my question is with my setup switching to madmax would make difference ? or should I just stick with it.

Thank you

[Dartellum]

Try it out. 😀

[nem0n]

I want to :D but if I stop the current ones just for testing will be taste of 15-20 plot already so wondered if someone did it

[thiagottjv]

@nem0n nem0n, with a similar setup (5900x, 32gb RAM, 2x 970 2Tb, 2 external USB3 HDs and clearlinux) I was getting around 40 to 42 plots.
I just finished testing this and each plot is taking around 1750s, witch mean that Ill do around 49 plots a day using only one HD. The best thing about this plotter is the same thing that you dont want to stop yours: you`ll lose the current plots. With this you just lose one and you dont need 3 end HDs to lower the copy times. No need to figure out staggering times and so on.

So answering your question. Yes, I think it would make a difference.

P.S.: I dont know witch system youre using, but if youre using windows you should take something around 10 to 15% off what I just said. If you`re using a gui distro you should take around 5% or so.

[nem0n]

@thiagottjv first of all thank you for the feedback. The thing is currently I have one 256 gb 980 pro and 2 tb due to my system is installed I was not able to connect the other 2 tb that I have so waiting for converter. But after posting this I decided to give it a try to mad max and it took 55 minute to get the plot done which approxtmly will be 25-26 plot maybe as a result of end of the day. But with swar I am able to get around 27-30 plot and hoping that with another 2 tb it will increase. But so far I didnt see the advantage of the mad max.

And yes I am using windows because of this is my main machine and have many stuff installed on it and gpu mining at the same time.

[thiagottjv]

You could do what I did. I partitioned my nvme with 100Gb for the OS and the rest for plotting. With this I have 1.7Tb with one 970 pro and another with the OS instalation and more 1.6 for plotting.
The next test Ill do is plot for windows. Since I'll have 64Gb and I'm using only around 20Gb with this plotter Ill try to use 23 threads (instead of 24) alongside windows and compare the times.

[Dschijn]

6 in parallel on a 5900x isn't that great, look at your CPU and RAM usage. You could do at least 10, maybe 12 in parallel staggered with SWAR.

[nem0n]

Generally I am doing parallel for 7-8 plot with current delay. But because I have kind of 1 x 2 tb nvme and 1x 256 gb I think that becoming my limitation and thats why if I try to go over that I am ending up with 20 plot or maybe even less. With my current approach I am getting around 27-30 plot. But not sure if this is good enough for my setup currently

[Dschijn]

True. I am pretty sure, with two propper 2TB SSDs you could do 12 staggered. 40GB of RAM might be an issue in Windows though.
I got a 14core Intel CPU and 2x2TB plotting in Ubuntu with SWAR, 14 in parallel and up to 16 when starting 1-2 early. Manage 44-46 plots per day that way.
Not sure if madmax could help me, since I don't habe enough memory in that system.

[nem0n]

I tried madmax but apperaently it taking 55 min for one plot which is compare swar not better

[thiagottjv]

I tried madmax but apperaently it taking 55 min for one plot which is compare swar not better

Are you using all of the 24 threads?

[nem0n]

Yes I tried use all of my threads but my second temp was not ram it was another ssd which is 256 and which result of 55 min

[casa10]

Can you share your configuration of swar ? (if you want of course)

[Dschijn]

consider using "max_concurrent_with_start_early" with one more plot than your 9 max_concurrent. Maybe with a short delay within phase 4 of 5-10min.
By that a new plot will already start shortly before phase 4 is finished and only phase 5 is copying the plot (which is not CPU intensiv).

[casa10]

Thank you.
i am surprised : you make 27-30 plots by day with 3 threads using only one NVME ? and one NVME?
@thiagottjv : you do 40-42 plots by days with your configuration withour using this new way to plot? can you share your configuration of swar please (if you accept) ?

@Dschijn
: same question

[nem0n]

@Dschijn I didnt play with that feature much but for sure will give it a try

[nem0n]

@casa10 yeah I do around 27-30 after it settled what I mean by settled is that like when you stop it until its fully back up and plotting in pararel. When it reaching to that stage then yeah I am getting around that but will try even more increase it

[Dschijn]

My config for an Intel i9 9940x with 14 cores, pltting on two Intel SSD DC P4510:
global:
max_concurrent: 16
max_for_phase_1: 8
minimum_minutes_between_jobs: 10

jobs:

• name: first
  max_plots: 999
  farmer_public_key: xxx
  pool_public_key: xxx
  temporary_directory: /media/ssd1/chiatemp
  temporary2_directory:
  destination_directory: /media/tower/chiatemp1
  size: 32
  bitfield: true
  threads: 4
  buckets: 128
  memory_buffer: 5000
  max_concurrent: 7
  max_concurrent_with_start_early: 8
  initial_delay_minutes: 0
  stagger_minutes: 30
  max_for_phase_1: 3
  concurrency_start_early_phase: 4
  concurrency_start_early_phase_delay: 5
  temporary2_destination_sync: false
  exclude_final_directory: false
  skip_full_destinations: true
  unix_process_priority: 10
  windows_process_priority: 32
  enable_cpu_affinity: true
  cpu_affinity: [ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13 ]

• name: second
  max_plots: 999
  farmer_public_key: xxx
  pool_public_key: xxx
  temporary_directory: /media/ssd2/chiatemp
  temporary2_directory:
  destination_directory: /media/tower/chiatemp2
  size: 32
  bitfield: true
  threads: 4
  buckets: 128
  memory_buffer: 5000
  max_concurrent: 7
  max_concurrent_with_start_early: 8
  initial_delay_minutes: 20
  stagger_minutes: 30
  max_for_phase_1: 3
  concurrency_start_early_phase: 4
  concurrency_start_early_phase_delay: 5
  temporary2_destination_sync: false
  exclude_final_directory: false
  skip_full_destinations: true
  unix_process_priority: 10
  windows_process_priority: 32
  enable_cpu_affinity: true
  cpu_affinity: [ 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27 ]