Help Chuck Norris. Plots fail at the end of plotting.
-
I don't know what's causing this. I'm plotting two drives at once. Both windows look just like this
Device settings for gtx 1070
0 0 8192 256 128 for both cardsAm i getting a memory leak.
I have 16gb ram and i'm using:
gpuPlotGenerator generate buffer C:\MountPoints\1-6\burst\plots\14653230141344995543_600000000_30400000_20000
pauseThis is happening on several different machines of mine.
Please help Chuck.
-
@ChuckNorris, hello! I'm a "Norris" too, in real life. I've got a GTX 1060 and gpuPlotGenerator as well. My 1060 suggested devices.text settings were 0 0 4096 something something. I couldn't use the 4096 value; I think because my monitor is also tied to the graphics card and using some of its resources. I now use 2048 instead (I have not really tried to optimize though).
I've had random lockups while plotting (direct and buffered) with larger plot files (one file for a 6 TB drive, for example). I now use multiple files of about 2TB each and I've not had the problem occur since. I know that's not an explanation of why it happens, but it did "fix" the problem for me. No real speed penalty when mining with several files on one drive.
Did you check the drives before plotting for bad sectors? CHKDSK /F? I've had some mucked up drives, even when new.
How much wiggle room did you leave in terms of free space? I've bought identical drives, yet some listed less free bytes than others and a plot that fit on one drive did not fit on one of these just a bit smaller drives.
Just some random thoughts. I'm tired; it's been a LONG day. This Norris is signing out...good luck!
-
@ChuckNorris Smaller plot files are a good idea to help mitigate situations like this. I would recommend running a single instance of GPU Plot Generator on each machine rather than two. The two instances may be fighting each other for resources. You can also reduce the amount of RAM it uses with a smaller stagger size. However, it seems to use up most of the available RAM at some point during the process anyway, which can cause problems if any other app is trying to use more RAM than it already has allocated. My settings for 8TB drives with my 1070:
devices.txt:
0 0 4096 512 128How I would plot your drives:
gpuPlotGenerator generate buffer E:\14653230141344995543_600000000_1525760_10240 F:\14653230141344995543_630515200_1525760_10240
gpuPlotGenerator generate buffer E:\14653230141344995543_601525760_1525760_10240 F:\14653230141344995543_632040960_1525760_10240
gpuPlotGenerator generate buffer E:\14653230141344995543_603051520_1525760_10240 F:\14653230141344995543_633566720_1525760_10240You can repeat this pattern for 20 lines to fill up both drives with 20 x 400GB plot files. Just copy, paste and add 1525760 to the starting nonce each time. Fills up the drive 99.99% with just a few hundred MB left over. Also, it's good to do a full format on your drives if you haven't already. If there's some flaw with the drive itself, a full format should catch it before you start plotting.
-
I leave at least 200gb free usually on a 8tb drive. Plots end up being 7.24 out of 7.45tb
It's worked on quiet a few drives. I think maybe what's happening is. I'm testing the speeds and nounces. and i stop the miner sometimes mid write. It's almost like there is shadow space it's taking. Maybe i should format after testing.
I've even had the same problem with plotting 1 drive with 1 instance. I'm going to reduce my scoup down to like 4gb per drive for plotting. All good ideas guys and helps me tip off an idea of what's doing it. Also keep in mind i'm plotting a petabyte. I don't want to be messing around with a bunch of files. I want to figure out what works really darn good on a drive. And replicate it for 100 plus drives.
-
@ChuckNorris said in Help Chuck Norris. Plots fail at the end of plotting.:
i stop the miner sometimes mid write. It's almost like there is shadow space it's taking
Run Plotschecker program if you have done this, and it should "cut" where you finished and repair the file. I run Plotschecker anyway on all files after I am done.
@ChuckNorris said in Help Chuck Norris. Plots fail at the end of plotting.:
This is happening on several different machines of mine.
Regarding your error at over 98 percent, I am pretty sure that this is because you have already filled your drive to the max and ran out of space. Run Plotschecker to snip it.
When you calculated your amount of plots, did you use the byte value found in drive properties in explorer? Byte value is more accurate. When you have found the amount of nonces you wish to plot you must also make sure they are a multiple of your stagger, so you can divide nonces by stagger and round off to nearest integer to find the most precise number of nonces you can plot. If you did it right you will avoid that error. But, Plotschecker fixes that anyway, so it's no biggie.
help Chuck Norris
LOL, since when did Chuck Norris ask for help btw?
When Chuck Norris gets bitten by a zombie he does not turn into a zombie, the zombie turns into Chuck Norris.
-
Sorry everyone. Have been filling up 80tb's a day. Finally got a breather. I figured out what was going on with the failing plots.
I would run the bat i would test the speeds. Make the tweaks and what not. Get her dialed in. Then delete the plot files out of the folders. Well... This is my old theory and now same theory. It's happened in the past. Whenever it's writing a plot file. If you close it as it's writing or just running i don't know which. It leaves something in like a cache form on the drive. So when it's filling it up. It ends up being short because the invisible cache is there. I diskpart clean the drives when i first install them. And if i fail a plot or am testing. MAKE SURE i FORMAT before i write the official plot.
Answer me this. I read somewhere a long time ago something about setting the default allocation size i think when formatting has some effect on something?! Puzzle Face.
-
@ChuckNorris said in Help Chuck Norris. Plots fail at the end of plotting.:
Answer me this. I read somewhere a long time ago something about setting the default allocation size i think when formatting has some effect on something?! Puzzle Face.
I seem to remember that a higher allocation size can be beneficial when plotting optimized, whereas in theory SMR drives can get the adverse effect, even when plotting unoptimized. You will get a few extra MB space with higher allocation size, but it's not much. All in all I don't reckon it's worth it, so I just keep it at default.
-
@Propagandalf Thanks
