The contest organizers will be happy to see that I’m about to commit my first borrow-and-improve from xenoncat, hopefully later tonight…
And done!
I can now do fixed heap allocation without increasing peak memory.
With the new layout, xenoncat should be able to save 14/15 on memory;
from 178MB to 166MB.
This is my result, what does it mean? I found 3 solutions in 5.17 seconds? Thank you!
time ./equi1
Looking for wagner-tree on ("",0) with 10 20-bits digits and 1 threads
Digit 0
▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▃▃▄▅▆▆▇███▇▇▆▆▅▄▄▃▂▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ 208 MB
Digit 1 x0 b0 h0
▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▃▃▄▅▆▆▇▇██▇▇▇▆▅▄▄▃▂▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ 208 MB
Digit 2 x0 b0 h0
▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▂▃▄▅▅▆▇███▇▇▇▆▅▄▄▃▃▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ 208 MB
Digit 3 x0 b0 h0
▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▃▃▄▅▆▆▇▇██▇▇▇▆▅▅▄▃▃▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ 208 MB
Digit 4 x0 b0 h10
▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▃▄▅▆▆▇▇▇█▇▇▇▆▅▅▄▃▃▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ 208 MB
Digit 5 x0 b0 h23
▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▃▄▅▅▆▇▇███▇▇▆▅▅▄▃▃▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ 208 MB
Digit 6 x0 b1 h76
▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▃▃▄▅▆▆▇███▇▇▇▆▅▄▄▃▂▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ 208 MB
Digit 7 x0 b6 h166
▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▃▃▄▅▆▆▇█▇█▇▇▆▆▅▄▄▃▂▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ 208 MB
Digit 8 x0 b5 h344
▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▃▄▄▅▆▆▇▇▇▇▇▇▆▆▅▄▄▃▂▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ 208 MB
Digit 9
3 solutions
3 total solutions
5.17user 0.18system 0:05.37elapsed 99%CPU (0avgtext+0avgdata 216200maxresident)k
0inputs+0outputs (0major+53565minor)pagefaults 0swaps
I tried executing eqcuda multiple times, but it could not find solution.
Am I doing correctly?
Running with NVIDIA-SMI 352.63.
make eqcuda
ubuntu@ip-172-31-5-81:~/GitHub/equihash$ time ./eqcuda
Looking for wagner-tree on (“”,0) with 10 20-bits digits and 8192 threads (128 per block)
Digit 0
Digit 1
Digit 2
Digit 3
Digit 4
Digit 5
Digit 6
Digit 7
Digit 8
Digit 9
9 rounds completed in 0.000 seconds.
0 solutions
0 total solutions
real 0m0.329s
user 0m0.007s
sys 0m0.242s
hi
I am unable to get source from git clone git@github.com:tromp/equihash.git
it return this msg
plz help
Solved
after few tries I got this. Thumbs Up Sir
Use git clone https://github.com/tromp/equihash
use make all instead of make eqcuda
Mr. Tromp, as far as I can understand in case of integration yours solver into the official zcashd miner memory consumption for mining will be below 4GB, at least 4GB will be more that enough, am I correct?
Cant Run ./eqcuda and feqcuda
Dear Sir I didnt get eqcuda and feqcuda after compiling so is there any problem with dependencies or something else
Integration with zcashd
Can someone tell that can we integrate it into zcash or is it still in development phase, moreover if it can be integrated into zcashd miner then how can we integrate it into it?
We won’t, developers of zcash will do if consider it necessary, I think they will at least because its the most efficient open source solver.
then what is the use of developing these open source solvers if zcash developers do not integrate it. For that matter they need to keep in view the business case of their Organization. After all its all business for the big guns and no one would ever wanted to compromise on it
These will be use to further optimize the miner. Have patience. We are all waiting for that open source working miner ala str4d standalone.
You people are great by timely responding each other
@str4d has already announced in this thread that he will try to integrate this algorithm into standalone miner over the weekend.
Saying We won’t [integrate into zcash], developers of zcash will do if consider it necessary, is IMHO not the correct approach.
I do find it necessary and would do it in a heartbeat, but I’m quite frustrated that I don’t have the necessary knowledge to contribute to efforts of zcash and other devs.
And it is in alignment with zcash business to provide good algorithms to public at-large. Otherwise the situation will morph into something like Bitcoin, where you have the Chinese de-facto controlling the blockchain. Here it would be the large operators with custom GPU software and cheap electricity.
Instead, the fast, open-source, ready-to-compile CPU algorithms would be in total alignment with the Zcash stated ideal of having as many people mining as possible. They are actually talking about mining on tablets… In fact they have a $30000 Zcash Open Source Miner Contest running with 4 submissions at this time.
Hmm, that always seems to happen if you run it for the first time.
Try simply re-running it. Then you should see
tromp@foobhar:~/equihash$ time ./eqcuda
Looking for wagner-tree on (“”,0) with 10 20-bits digits and 8192 threads (128 per block)
Digit 0
Digit 1
Digit 2
Digit 3
Digit 4
Digit 5
Digit 6
Digit 7
Digit 8
Digit 9
9 rounds completed in 0.133 seconds.
3 solutions
3 total solutions
real 0m0.295s
user 0m0.112s
sys 0m0.176s
You can add option -s to see the solutions in full…
-John
Did you do
make eqcuda feqcuda
?
Make all doesn’t build the CUDA targets, since many people won’t have the CUDA compilers installed.
If the build failed, then please provide the error message…
-John
My equi/equi1 solvers use 208MB, and my faster/faster1 uses 656MB.
But I plan to combine the two versions into one that is both lowmem and fast…
equcuda is showing solutions on my laptop which has Nvidia GTX GPU.
However when I tried to execute equcuda on AWS g2 instance. It is not
showing any solutions.
I tried multiple times today. Still no luck on AWS instance.
No idea what’s wrong with CUDA on AWS.
The most verbose output comes with this define:
nvcc -DHIST -arch sm_35 equi_miner.cu blake/blake2b.cpp -o eqcuda
which prints detailed bucket size histograms.
You can compare those numbers between your working and non-working versions to see where they start to differ. Maybe that can give us a clue…
hi
thanks for your help
i got this msg in return when I used make eqcuda
need your further help