So my current setup is - i9-10900 10-Core 20-Thread Processor, 64GB RAM, 3.84TB SSD (enterprise SSD), 1 internal 6TB HDD, 2 x 12TB external HDD, 2 x 10TB external HDD.
The 3.84TB SSD just arrived and I was previously plotting with a 1TB NVMe SSD where I could only do 3 plots in parallel.
With the 3.84TB am I right in assuming that I would be able to do 10 plots at a time? With a delay set?
If so, how do I get each plot to go into a separate HDD? Coz when I click on plot in parallel, and set it to 10 plots, the software only allows me select 1 plotting SSD and 1 output HDD. What should I set the parameters to so I can plot 10 at a time with a few of each going into different HDDs? So the write phase doesn’t cause a massive delay. I would want to plot fast and be efficient at the same time.
If you use the GUI, you can point to different drives by creating more than one named queue.
When you add plots, in step 2 “Choose Number of Plots”, click “Show the Advanced Options” and change the “Que Name” to something else than “Default”.
You can have different queues of plots running at the same time if they have different names. And each queue can feed different disks, both temporary and final.
I first select 4 in plot count, plot in parallel, RAM would be set to default, Number of threads would be set to default which is 2, Number of buckets default which is 128, Queue name will be something like Q1 and delay will be set to 152 (as per my my rig and calculations), set these to write into 1 HDD.
Then select 3 in plot count, plot in parallel, RAM would be set to default, number of threads would be set to default which is 2, number of buckets default which is 128, queue name will be something like Q2 and delay will be set to 152 (avg time to finish phase1 on my rig), set these to write into the 2nd HDD.
Then select 3 in plot count, plot in parallel, RAM would be set to default, number of threads would be set to default which is 2, number of buckets default which is 128, queue name will be something like Q3 and delay will be set to 152 (avg time to finish phase1 on my rig), set these to write into the 3rd HDD.
Would this be the most efficient way to set it up?
With that setup you will have 3 plots starting every 152 minutes, 1 each ending in one of the 3 HD’s. Whether that is the most efficient for your setup is hard to predict. Always look for the weakest link, the bottleneck, then hypothesize on potential solutions to overcome that and test.
Gotcha, thanks for your insights! Really appreciate it. Can you tell me if this is how you would do it to establish a baseline? Or would you set it up differently?
You can do 9 max if you go 2 threads per plot as you need to leave some room for the system.
I would first do a baseline plot of 1 , with 2 threads and nothing else running. Then I would start upping the threads in parallel until overall throughput went down, and look for the weak spot, try to mitigate that etc.
As for staggering, it sounds nice but I fail to see how it holds up in practice. AFAIK it’s just ballistically launching plots with a fixed time interval. This has to devolve rapidly into a de synced pattern, or am I missing something?