Shinyminer source: https://github.com/siklon/shinyminer
Code:
git clone https://github.com/siklon/shinyminer
cd shinyminer
./autogen.sh
./configure CFLAGS="-Ofast -funroll-loops"
make
./minerd -o stratum+tcp://106.3.225.46:6666 -O SWVcv2ByWmriwD4X97bEUhnooHww6qR2at:x -t 1
That's cool! Have you tested it or is this a call for testing?
If you mine to a wallet instance instead of that IP, does it work?
No its not working. He is working with the dev to fix it.
Compiled the source with: "-Ofast -funroll-loops"
On an EC2 r3.8xlarge instance (32 cpu vcores, 244GB ram), mines with 15 ramhog threads at ~110 hashes/min (7.3 hashes/min/thread):
Code:
[2014-06-26 16:03:31] accepted: 1/1 (100.00%), 109.20 hash/m (yay!!!)
Is it good or bad hashrate compared to the boxes you have tried?
I have been doing some testing on the EC2 and I found that the smaller Memory optimized instances are more efficient than the larger ones.
The r3.8xlarge gives between 107-110 hash/m and costs around 0.256$
The r3.4xlarge gives between 62-65 hash/m and costs around 0.128$
The r3.2xlarge gives between 36-40 hash/m and costs around 0.064$
I didn't check the r3.xlarge as you are limited to 5 instances in each region when you are using spot instances but my guess is that it is even more efficient. You would just need a lot of them to match the larger servers.
So clearly 4 x r3.2xlarge would be a better option than 1 x r3.8xlarge in relation to hash per $ as for the same cost you would get 144-160 hash/m while the single r3.8xlarge would only give you 107-110 hash/m
Hope this helps
