Plotting, mining and stagger with multiple drives on one PC
I read that a larger stagger size will reduce the stress on your HDD, so I purchased another 16gb to bring my PC up to 24gb. I also have three 8tb HDDs to mine on.
I would like to mine with 18gb (75%) of my system ram. When I plot my drives I am plotting it planning to use the whole 18gb stagger.
How does this work when mining on multiple hard drives?
1 - Does the miner only access one drive at a time but in sequence?
1a - If so, will that slow down my mining speed because it only accesses one drive at a time?
2 - Would it be faster to mine 3 drives simultaneously with 6gb of stagger?
2a - Would I then plot my drives with a 6gb stagger instead of 18gb?
Sorry for the basic questions and thank you for any information!
- no drives are read in parallel
- always use optimized (stagger = numberOfNonces)
- No - they're done in parallel.
1a) - N/A
- Use xplotter or gpuplotgenerator in direct mode to create an optimized plot.
2a) See 2)
- No - they're done in parallel.
Thanks for the information!
So is my thinking right to want a bigger stagger size? I have gpus I could use to plot faster, but I get out of resources errors (I am assuming because I dont have 18gb of ram on them). It seems like my choice to have a bigger stagger size is significantly slowing down my plot time.
When I mine, how do I set stagger size? Or is it read from my plots?
@mhminer With Xplotter or Gpuplotgenerator in direct mode, the stagger is basically irrelevant - the plotter is going to create a plot with stagger = number of nonces. For gpuplotgenerator set the stagger to 0.33 * the number of megabytes you have of RAM. ie on an 8GB system: 8 * 1024 * 0.33 = 2703.
mhminer last edited by mhminer
@haitch Thank you so much! I will test this out, I am sure I can get it working great. I am also a "but why" kind of guy.
If it is irrelevant, am I wrong in thinking "more stagger is better"? In the doc linked here it says:
"<stagger size> - You can think of stagger size as a ‘buffer’. This is how much will be read into memory before writing to the hard drive. The place that stagger size really comes into play, is during mining. During mining, the larger the stagger size the better on your drives, the less seek time, and faster your computer will read through the plots, thus faster deadline submission and more of a chance of hitting a block. Setting your stagger size equal to 75% of your total memory (or up to the amount of memory you have,) is a decent default. I.E. if my computer has 10GB of RAM, I would set my stagger size to 8192. Or something close to that, my math is not exactly perfect, but a stagger size of 8192 or larger is great, if you have the memory for it!"
edit: I also want to gpu assist plot but only cpu mine.
@mhminer If you plot the file sequentially, then yes, you'll have "stagger size" of scoop X from nonces all laid out, then further up in the file there will be another stager worth of scoops, then further up ..... With xplotter or gpuplotgenerator direct, the stagger = the number of nonces, so every single scoop X is sequentially laid out, you slurp them all up instead of jumping all around your disk. The plotters I referenced can do this without regard to the stagger you specify - the stagger is just telling the plotter how much memory it use, it does not impact the final file - that will be optimal.
@haitch Ok perfect. So how much RAM will the miner use if I set the plotter to 150% of my system ram and plot in direct mode?
@mhminermy statement about 150% of gigabytes was wrong - totally wrong. Don't know where that brain fart came from - do not set memory to use > 33% of you physical RAM.