Some general mining questions
I've been quite enjoyed the past few weeks tinkering with hardware and earning some burst. I think I've got the basics down pretty well and I've been lucky/hustled enough to have picked up some 3tb hard drives second hand from ebay. Which should have arrived within the next 3 weeks (one should arrive soon, the other is just about to ship). Now I know this isn't as ideal as buying a nice stack of 8tb drives in bulk but it's what i've got to work with.
So... knowing these are on their way, I have a couple questions about plotting.
- I've plotted and replotted drives, both due to failure and as experiment using both the cpu wallet plotter and the gpu plot generator in both buffer and direct mode. I've found and got acquainted with the burst nonces calculator.
In spite of this I've noticed that the size of plots suggested as correct by the calculator doesn't seem to be completely accurate depending on the write method. This is madness, sweaty palms, and a very snug jacket territory. XD
I'm plotting in Windows 10 using an NVIDIA gtx 750 ti sc, an intel i5 2500k @3.3 GHZ at base clock, and with 32 GB of ram. Every drive I've plotted into tests as 100% healthy by Hard disk Sentinal.
I first plotted the 3tb drive in buffered mode using the GPU, leaving 2gb empty. It worked fine using the nonces calculator, but then I realized it should be optomized so I replotted it using direct mode. This worked fine. I then tried to plot two 500 gb drives using the GPU plotter again leaving 2gb of free space and all hell broke loose. They simply would not work predictibly, and would not stop plotting in spite of having the correct number of nonces.
Eventually I plotted both of the 500s (starting with the end nonce of the 3tb, and then the next 500 respectively) using the CPU plotter which worked absolutely fine and gets about the same nonces per minute as my GPU (which I was trying with 0 0 256 64 1024)
Obviously it's best to go with whatever works... right? But I really feel like I'm missing something here.
Thanks for any insight.
@Acanterelle I believe you're running into the issue of the number of nonces having to be an exact multiple of the stagger. You can calculate the number of nonces that will fit on your drive, but that number needs to be a multiple of the stagger.
For example, If I tell the plotter to plot 19 blocks with a stagger of 10 blocks, it's going to plot 20 nonces, not 19. That extra nonce will corrupt your file if there is not sufficient space.
To calculate the correct number of nonces:
Floor((<drive size in Bytes>/262144) / <stagger>) * <stagger>
Aaaaah yes. I've set quite a big stagger size. sheepish grin
So to get closer to what I was originally trying to plot I would want to use:
Floor((<drive size in Bytes - 2GB>/262144) / <stagger>) * <stagger>
I've effectively been putting the - 2gb on the outside of pemdas instead of on the inside.