@rvncn has joined the channel
@Fireduck Why does my miner-pool.bat download the snow file again at every startup even when it is in the folder?
Why exactly is the lenght incorrectly I have downloaded the whole 64GB through the .bat SnowFall process
the miner nerver downloads the snowfields, it either generates them or loads them from the disk to the memory
it's not instant to pull tens of gigabytes from the disk into memory
mining is faster on linux or on windows?
Yes but as you can see on my screenshot I have the files already generated on my diskspace. I already mined for a couple hours.
And now it shows "Unexpected length" on the terminal and tries to generate them new everytime. @Fireduck
Any idea?
disable auto snow to begin with
It looks like your deck files are incomplete. They shouldn't be zero length.
Did snow 7 just got activated?
No where close
Still on 6
my snowfall downloads 7 currently :confused:
It generates 7 to be ready
ah, that makes sense. Thanks!
as said, disable auto snow, it’s mostly just in the way
1-min: 32,588K/s, 5-min: 25,517K/s, hour: 2,126K/s Is that like an okay number? I run this on a i7-6700k CPU and an 1TB Samsung 960 Evo Pro as m2 ssd
960 pro can go to a sustained 100k, but that requires a bit of trickery on the linux kernel side
up to 50k is easy
as a baseline, if it's not saturating your cpu yet, bump the thread count and see what that does for you
I’m not a coder in opencl or cuda but exchange between system ram-cpu and system ram-gpu are about the same , I’m sure that using a gpu to make those read and write with system ram could potentially do more hash than a 8 core cpu.
it's not a checksum throughput problem, it's a seek latency problem
Is the cpu randomly picking hash in the snowfield without keeping in memory those already tested ?
it is picking a random nonce
but the odds of hitting the same one twice is way less than the cost of tracking ones that are tried
It seems like it makes no difference for my hashrate or CPU load if i increase the threads in the .conf file.
Started at 40 -> 60 -> 80 -> 100
hashes are no different
Yeah, IO limited