Miner- SILENTARMY v5

alexfox · October 31, 2016, 1:30am

@mrb
mrb !!
New Linux drivers from AMD http://support.amd.com/en-us/download/linux
I hope this helps in your work.

Your2 · October 31, 2016, 1:48am

Today. He’s a recognized dev, so you lost that argument. Miner coming imminently. Walked to water and forced to drink. 120 Sol/s gtx 1070 by coder who produced branches of lbc miner a little while ago:

alexfox · October 31, 2016, 1:51am

shut up please.
bla bla bla (20 char)

Your2 · October 31, 2016, 2:26am

Yeah, no. This is valuable information. The btc talk thread is blowing up. Pool side sol rate validated

mrb · October 31, 2016, 2:40am

I released SILENTARMY v2 which now supports GCN 1st gen GPU. In fact all AMD GPUs should in theory be supported, but keep in mind I still have done zero testing on TeraScale (pre-GCN) AMD GPUs. See:

github.com

mbevand/silentarmy/blob/master/CHANGELOG.md

# Current tip

* Avoid 100% CPU usage with Nvidia's OpenCL, aka busywait fix (Kubuxu)
* Optimization: +10% speedup, increase collision items tracked per thread
  (nerdralph). 'make test' finds 196 sols again
* Implement mining.extranonce.subscribe (kenshirothefist)
* mining.authorize sends an empty string if no password is specified
* Fix memory leaks
* Avoid fatal error when OpenCL platform returns CL_DEVICE_NOT_FOUND

# Version 5 (11 Nov 2016)

* Optimization: major 2x speedup (eXtremal) by storing 8 atomic counters in
  1 uint, and by reducing branch divergence when iterating over and XORing Xi's;
  note that as a result of these optimizations, sa-solver compiled with
  NR_ROWS_LOG=20 now only finds 182 out of 196 existing solutions ("make test"
  verification data was adjusted accordingly)
* Defaulting OPTIM_SIMPLIFY_ROUND to 1; GPU memory usage down to 0.8 GB per
  instance
* Optimization: significantly reduce CPU usage and PCIe bandwidth (before:

This file has been truncated. show original

I would be interested if people with Nvidia GPUs could test SILENTARMY now. There is a chance my fixes help with a separate Nvidia issue: “clEnqueueReadBuffer (-5)”, see No Nvidia support: errors out with clEnqueueReadBuffer (-5) with both CUDA 7.5 libs and system libs · Issue #6 · mbevand/silentarmy · GitHub

Louie2001912 · October 31, 2016, 2:44am

What a god send! Do we have to wait for people to add this into their miners right?

Jiggytom · October 31, 2016, 2:51am

can you compile a windows binary?

Louie2001912 · October 31, 2016, 2:56am

I’ve read this is only a solver. You have to wait for the 3 big miners to implement it. (genoil,etc…)

Exqua · October 31, 2016, 1:06pm

mines twise as fast as nheqminer! i’m very happy

leroy627 · October 31, 2016, 1:51pm

Nope, doesn’t work on my GTX970.

Output:
root@BMPC-Linux:~/silentarmy# ./silentarmy --nonces 100
./silentarmy: /usr/local/cuda-8.0/targets/x86_64-linux/lib/libOpenCL.so.1: no version information available (required by ./silentarmy)
Solving default all-zero 140-byte header
Building program
Hash tables will use 1744.8 MB
Running…
clEnqueueReadBuffer (-5)

I edited libOpencl.so directory in makefile to:
/usr/lib/x86_64-linux-gnu/libOpenCL.so , which is the default location. I’m using Cuda 8 too.

mrb · October 31, 2016, 2:59pm

@leroy627 thanks for testing… I’ll try to get my hand on Nvidia hardware to fix this bug for good (but that’ll be v4).

@Jiggytom Yup I plan to support Windows (this will be maybe in v5).

t2yax · October 31, 2016, 3:02pm

please dont forget cuda 2.1

alexfox · October 31, 2016, 5:27pm

@mrb
There is a suspicion that the kernel works as something incorrect with pools on which there vardiff.
https://forum.zcashcommunity.com/t/genoils-zec-miner/

Sorry. Not a kernel… pools works incorrect with that kernell.
What do you think about?

tylerdurden · October 31, 2016, 5:31pm

where do i get it

20 characters

DanLar75 · October 31, 2016, 6:42pm

What’s the ‘best’ Windows miner that uses Silentarmy at this point? Is it still the fastest solver?

miningpoolhub · November 1, 2016, 9:09am

Is there some limitation that pool must implement to support your kernel?
I saw that last 12 bytes must be zero from github wiki, but seems like that cannot be controlled by pool right?
Our pool uses 4 bytes or minernonce (extranonce1) and remain bytes are all treated from miner itself.

Does it depends on miiner program implementation or pool implementation? I want to be confirmed.

Genoil · November 1, 2016, 10:55am

No the SA kernels are fine in itself. Miner software however needs to respect the requirment of the last 12 bytes of the nonce to be 0. My miner respects that requirement.

However, with the protocol allowing pools to freely specifiy a part of the nonce to be fixed from their end (nonce1), the possibility arises that with a large nonce1, there hardly is any noncespace (nonce2) left for the miner. In practise there’s nothing to be concerend about: For example, suprnova has a 16-byte nonce1. SA take 12 bytes off nonce2. That leaves me 4 bytes (32-16-12) of nonce2 space. This is still good for 4294967296 equihash runs per block. A 30S/s GPU is about 16 runs per second. With a block time of 130 seconds, we have a maximum of about 2100 runs, so make that 10K runs on 5 GPU worker. So this is more than enough.

DanLar75 · November 1, 2016, 1:19pm

Not a all. It’s far more profitable to run a AMD card that cost less for anyone with cheap power.

miningpoolhub · November 1, 2016, 3:34pm

Now I understand better than before.
Thanks.

ajqjjj · November 2, 2016, 5:30am

I am equal to been able to solve the problem?

Topic		Replies	Views
Help on how to use silentarmy on ubuntu 16.04? Mining Support	4	2486	February 3, 2017
Mining on OpenSource AMD (amdgpu/Mesa/llvm) OpenCL Mining	3	1879	April 16, 2017
AMD GPU MINER - Silent Army + Extremal Mining	12	8923	October 28, 2016
AMD Radeon HD 6800? Mining	6	7718	November 4, 2016
CUDA SILENTARMY v4+ Windows Build Mining	74	14617	November 17, 2016

Miner- SILENTARMY v5

Related topics