You need to work on your marketing.    Sell it as, "saving power when the network has no work available," or something.

The biggest mistake a FLOSS developer can make is letting users drive development.  99% of users don't have the skills, experience or insight to make development decisions.  This is a prime example of why.  Of course it is ideal to keep the GPUs pegged at 100% usage. But if there is nothing productive to do, then there is zero sense in burning the watts.

Do what's right and ignore the uninformed, baseless complaints.

Heh, those bounties are a joke, right?  You guys probably ROI the lot of them in <24hrs.

Insider?  https://devtalk.nvidia.com/default/topic/878455/cuda-programming-and-performance/gtx750ti-and-buffers-gt-1gb-on-win7/

Dynamically loading non-GPL code into GPL code is legal grey area.  Not that there's any precedent for any of this stuff.

It also does nothing to address the issue of the fee being removed.

OK you toy lawyers... I've been in FLOSS dev for over ten years and have plenty of experience with GPL compliance.

1) ccminer is hopelessly non-compliant.  The core code is GPLv2.  It links to some dependencies which are mostly LGPLv2 system libraries.  A few of the algo kernels are licensed MIT, BSD or Apache.  So far, so good.  Now there are a couple dozen kernels with no stated license whatsoever.  The sources being public does NOT make them open.  They default copyright to the original author.  Using them in any way, without written permission from the original author, leaves you open to legal action.  ccminer linking to them breaks GPL-compliance and distributing its binaries is illegal.

2) Just because someone uses GPL'd code in some private project does not give you any legal recourse towards them.  You have to legally obtain a binary release first.  It is perfectly fine to use GPL code in something you never release to the public.  It is perfectly legal to sell a binary release privately, so long as you give the customer some means to obtain the code for that release.  The sources need not be posted publicly by the original author.  They can be sent to the customer via email, on a USB stick, CD or even printed on 100,000 pages and mailed, etc.  Additionally, purchasing a binary only gives your rights the the sources used to build THAT release binary.  The author does not have to give you SCM history and you are not entitled to future release sources without legally obtaining said future release in binary form.

3) As a continuation to #2; Anyone who legally obtains the source code to a GPL compliant project has the right to do what they want with it within the GPL licensing framework.  This includes modification and redistribution, including to the public.

So as djm34 has been saying, adding a mining fee to ccminer is a lost cause.  Even if distributed privately, it only takes one person to exercise their GPL rights to obtain the code.  That person doesn't need to even know anything about code to publish it.  Next some will remove the mining fee and use it privately.  Eventually someone will remove the fee and release fee-less versions of the code.

Taking the opposite side of this argument is untenable.  You think people will morally sit back and pay that fee if they don't have to?  Bet your ass they won't.  I'd be the first one in line to obtain the sources and strip the fee for my own use.

The only solution is a completely new, closed source, fee-based miner with completely new kernels written in isolation from the specs.  The resulting binaries would further need to be copy protected to dissuade reverse-engineering and binary patching.  How many people do you think would trust that?

I achieve this on Linux with the following xorg.conf

In the "ServerLayout" section, I have the nvidia cards assigned to screens to allow NVML to work.  It costs 18MB of VRAM to X, but I noticed no performance decrease.  It is simple enough to disable this by commenting screens 1 and 2 and uncommenting the "Inactive" lines for them.  You will lose temperature, fan speed, etc though.

EDIT:  On Ubuntu, you'll also need to disable "gpu-manager" or the dumb ass thing will overwrite your xorg.conf everytime you log in.  I just commented the entirety of /etc/init/gpu-manager.conf.  Allegedly, this also can be done by passing a kernel param in newer (than 14.04) versions.  Though I don't recall the name of said param.

Could be launch params changes in a recent commit.

C&P FTW, I suspect.

Unused variable warnings are innocuous either way.

I've updated my previous post with what I've found regarding neoscrypt wallet status.  If you're having issues with one that's already fixed, then you need a newer version of the wallet.  That may require waiting for their devs to cut a release.  Of the two that are reported, I'd suggest posting in the GH issue thread links I've posted.

As for output, did you see zTheWolfz's post a while back?  Apparently that's how it's always worked.  It could be something in the kernel.

Much appreciated, sir!

Which coins are you having wallet trouble with specifically? PhoenixCoin is fixed in their master branch, so should be OK whenever they cut a new release.  I opened an issue for FeatherCoin at here.  You can use the --broken-neo-wallet flag until they sort things out.


I also wanted to clarify that by "closed," I meant "closed source."  That is the only way to enforce something like a small fee.  Generally in open source I don't care about such things because, for someone else to monetize my work, they have to put in some effort on their own.  However, with a miner its just; start program, profit!

EDIT: I checked sources for Feather, Pheonix, Orbit, UFO and Halcyon.  They're all correct except for Feather, which I already reported and UFO, which I've reported here.

Not at all.  You actually sped things up by posting output.  It saved me waiting around on my piddly setup.  We're past even by my book.

ccminer has NO control API of any kind.  It's just a config/text log app.

More like a spare bedroom, one GTX960, one GTX750Ti and 0.10USD/kWh power.  Amateur at best.

I'm not sure it matters how many ways to donate we have. Miners seem to be way better at complaining than donating.  I had my all of my GPUs pointed at broken soloing for nearly a week and have only received two donations (thanks again myagui and zTheWolfz!) since the fix went in.  That's despite there being at least a half dozen people on here complaining about the issue.  It really makes me wonder whether my time would be better spent fixing ccminer or teaming up with some of the kernel guys and doing something closed from scratch...

sp_ just merged my PR adding a new flag "--broken-neo-wallet" to work around the known issue of the malformed data parameter.  So that will be available in 70, whenever it's cut.  There may be some other problems though, as I never saw any GPU status updates after soloing PhoenixCoin for ~1hr.

That'd likely have to be on someone else's plate.  My CUDA chops are noob at best and will remain that way until I'm happy with core code.

EDIT: This dumbass forum needs markup syntax checking...

IIRC the blue text is only for stratum info.  You're soloing, so no stratum.  You'll only see GPU report at the end of a round and accept/reject.  Just let it run

Yeah we worked it out on IRC.  I'm 100% sure this needs fixed in the wallets, not the miner.  I'm willing to add it based on a non-default flag for the time being though.

Do what you want, but so far as historical record goes, it's completely useless.  I could squash the entire git history to 1 commit and that counter would go back to 1.  Or rebase away some commits in the middle and reduce it by 500.  Anyone who reads that number in the future may or may not wind up looking at the same commit (nevermind the fact that you'll have to manually count commits to find it).  IMO, making up your own nomenclature only serves to confuse anyone trying to help themselves and thus is a detriment to the community.  The SHA1 is right there to c/p.