rebase on top of the ethminer master branch

[hackmod]

Experimental

This is my rebase work on top of the latest ethminer master branch.
The master branch of the ethminer a bit faster and the latest ethash already support ProgPoW v0.9.1

Please see
https://github.com/EthersocialNetwork/ethminer-ProgPOW/commits/progpow-master-rebase

Only CUDA tested.

Screen shot

(rebased ProgPoW ethminer + chfast/ethash master branch with a small progpow CPU verifier wrapper fix)

P106-6G (samsung) ~10.4MH/s (~24.0MH/s for ethash)

[ghost]

@hackmod @ifdefelse @RadixPi

CI test results:

• https://circleci.com/gh/ethereum-mining/ethminer/2535?utm_campaign=vcs-integration-link&utm_medium=referral&utm_source=github-build-link

• https://ci.appveyor.com/project/ethereum-mining/ethminer/builds/20492596/job/1drp6imtv7o4o033

• https://travis-ci.org/ethereum-mining/ethminer/jobs/458471141

log:

CMake Error: The following variables are used in this project, but they are set to NOTFOUND.
Please set them or make sure they are set and tested correctly in the CMake files:
CUDA_nvrtc_LIBRARY
    linked by target "ethash-cuda" in directory C:/projects/ethminer/libethash-cuda
-- Configuring incomplete, errors occurred!

I think CMake can't find the specific library that is defined here

https://github.com/EthersocialNetwork/ethminer-ProgPOW/blame/master/libethash-cuda/CMakeLists.txt#L42

[hackmod]

@naikmyeong // fixed now
anyway, the openCL version have some period_seed issue

[ghost]

@hackmod I can't see any test results, is it possible to rebase your fix again?

[hackmod]

rebased again on top of the latest master branch ethereum-mining/ethminer#1766

[hackmod]

updated. rebased again~
support CUDA-8 (compiled. not tested test passed)/CUDA-9/CUDA-10

[hackmod]

binaries are available at https://github.com/hackmod/ethminer/releases/tag/v0.9.2-progpow

• only CUDA supported
• you can test it locally with --noevel option like as follow
  ethminer --noeval -v 3 --cuda-devices 0 -U -P http://127.0.0.1:8545

See also progpow testnet:

• netstats - http://boot.gangnam.ethdevops.io/
• https://gist.github.com/holiman/c8944c8be26f798e391822b9a70cb91c#gistcomment-2783941

[hackmod]

https://github.com/hackmod/ethminer/releases/tag/v0.9.2-opencl-win-1218
OpenCL supported now

• extract compileKernel() routine and call compileKernel(prog_seed) like as CUDAMiner
• CUDA-8 support added (GTX 680 confirmed)
• temporally disable CUDA to strip nvrtc*.(so|dll) dependency

[hackmod]

fixed epoch change issue
https://github.com/hackmod/ethminer/releases/tag/v0.9.2-opencl-win-1219

• OpenCL: fixed setArg issue after epoch change
• CUDA: fixed manual rebase mistake. (fix incompatible driver context error after epoch change)

[ifdefelse]

Considering https://github.com/AndreaLanfranchi/ethminer is close to ethminer master and runs both ethash and progpow I think we can close this out as fixed. Thanks for all the work.

[solardiz]

@ifdefelse What's your desired/expected future for this repo vs. https://github.com/AndreaLanfranchi/ethminer? ProgPOW is still being revised in this repo, yet in a tree based off older ethminer. Is the intended use of this repo to serve as reference and for testing? Should those ProgPOW spec and implementation changes be getting ported to https://github.com/AndreaLanfranchi/ethminer and at what points? Thanks!

[ifdefelse]

This repro is only intended for algorithm research and development. Andrea's implementation is up-to-date and much more polished for going into production.