Updates on my Half a peta
-
Upgraded from kaby i5 to i7 on both of my units. Also upgraded from 16gb to 32gb ram
Old read speeds up to 600mb
Curent read speeds 900mb.
Helped quite a bit. So i'm reading 250tb in 90 seconds or less now with un optimized plots. (optimizing now)
Question and problem
My Jminer eventually errors out because not enough resources. And kills the miner. Also get the windows notification.
Jminer uses all 32gb of my ram now. How can i solve this memory problem?
-
@ChuckNorris I believe you will have to look for an optimal number on the 'ChunkPartNonces' parameter, so that you gain between reading speed and memory saving.
-
@vExact Which direction do i go? i've tried 320000 160000 960000 The higher the number it saves memory? and it goes to go in additions of 160000 at a time?
-
@ChuckNorris said in Updates on my Half a peta:
@vExact Which direction do i go? i've tried 320000 160000 960000 The higher the number it saves memory? and it goes to go in additions of 160000 at a time?
I use a power of 2 that is closest (greater than or equal) to 4096 x (# of TB in the farm), it yields, for me, the fastest time with the lowest amount of memory usage.
If you have 250TB on a miner, try 2^20 or 1048576 chunkpartsnonce which is slightly greater than 250 x 4096 = 1024000.
-
@rds Thank you so much. That was the formula i was looking for. I just tried your number. And my usuage is now 24gb of ram of my 32gb. That just might do the trick. i think i've also heard once i optimize these drives the memory usage will also be reduced.
-
@ChuckNorris what @rds said. I've heard different kind of suggestions, but I like this one too :)
-
@vExact said in Updates on my Half a peta:
@ChuckNorris what @rds said. I've heard different kind of suggestions, but I like this one too :)
I ran a test awhile back (I talked about it in another thread) ran through 512 MB to 2GB with different size plots. This formula worked best for me.
