Expected 1080 Ti performance


#41

Should not be too hard to run multiple instances. I guess you need a separate directory for each. Then make sure none of the ports overlap (i.e. defaults can’t be used) so you need to set

external:
  port: xxxx
internal:
  port: yyyy
websocket:
  channel:
    port: zzzz
sync:
  port: aaaa

And then you need to change [email protected] in releases/1.0.1/vm.args into say [email protected]. Beware that is from memory and not something I’ve tried. Not sure it will be any more efficient though but maybe worth trying.


#42

Hahaha I was just saying on the windows one this does not work. Having different ports, different folders and different erlang contexts is not enough in windows.

However in linux I have done this with docker and it works quite nicely.


#43

Did you do a tut on that ? Or just even a bash history ?


#44

For which one? the linux or the windows one?


#45

guys, tell me please how many AE can you mine per 24 h with 1080 Ti or 1060 at current difficulty?


#46

The current target (Height = 6620) is 506663214 this corresponds to a difficulty of ~328000 - so in order to solve one block you need to solve 50 * 328000 = 16400000 cuckoo problems (1 in 50 attempts gives you a solution and 1 in 328000 solutions meets the target).

If a 1080Ti is able to do 3 problems/s, I think, so on average you will mine 0.02 blocks in 24h.


#47

have updated cuda binaries:
https://github.com/Zwilla/BitCaine5_aeternity_miner/tree/master/binaries

cuda29_GTX_1030	special builds	2 minutes ago
cuda29_GTX_1050	special builds	2 minutes ago
cuda29_GTX_1060	special builds	2 minutes ago
cuda29_GTX_1070	special builds	2 minutes ago
cuda29_GTX_1080	special builds	2 minutes ago
cuda29_GTX_2080	special builds	2 minutes ago
cuda29_GTX_970	special builds	2 minutes ago
cuda29_GTX_980	special builds	2 minutes ago
cuda29_GTX_titan	special builds	2 minutes ago
   cuda29-generic          for all others at moment

source code included

example: 10xx includes always 10xx I

let me know how the drivers perform: here: https://github.com/Zwilla/BitCaine5_aeternity_miner/issues/new

12/12/2018 binaries updated!!!

how to test the performance?
this are a test result!!!
0 total solutions with 5000 nonces

real 0m12.810s
user 0m9.163s
sys 0m3.567s


#48

Why there 0 solution with 5000 nonces?


#49

dry run, normal run will find 100 or more solutions and takes as long


#50

Hi,Zwilla.I replaced it with your cuda29, but found that only one GPU worked, the other five GPUs occupied 0. With hans CUDA29, there are 6 GPUs working. The configuration file is the same. Where is the problem?


#51

which version?
./cuda29_GTX_1070 -s

but, epoch starts the mining app!!!
So if epoch starts -d 0 and -d 1 to -d 5 then it starts as executed.

If only one card is working then there is anywhere other a problem!

Why?

Because if epoch is able to start once the mining engine, then epoch is able to repeat it million times.

Use this:
/usr/bin/nvidia-smi
to check if cuda it self find your card

or this after epoch start
/usr/bin/nvidia-smi pmon
ctrl c c will exit the Nvidia monitor


#52

I have run it 3 hours,but there still 0 solution in epoch_pow…log


#53

which card do you have exactly and which cuda-xxx are you using?

  • simple test

  • stop epoch

  • $./cuda29_GTX_1080 -g
    If you don’t see similar lines, then you have the wrong version or maybe need an update.

    GeForce GTX 1070 Ti with 8119MB @ 256 bits x 4004MHz
    Looking for 42-cycle on cuckoo30("",0) with 50% edges, 64*64 buckets, 176 trims, and 64 thread blocks.
    Using 6976MB of global memory.
    nonce 0 k0 k1 k2 k3 a34c6a2bdaa03a14 d736650ae53eee9e 9a22f05e3bffed5e b8d55478fa3a606d
    2-cycle found
    30-cycle found
    754-cycle found
    findcycles edges 63447 time 24 ms total 367 ms
    0 total solutions with 1 nonces

DOCS UPDATED: https://github.com/Zwilla/BitCaine5_aeternity_miner/blob/master/binaries/README.md


#56

It’s ok now,but slowly than 8 hours ago’s cuda