Topic: Bounty for Open-Sourced XMR/Cryptonight GPU Miner Bounties Thread - page 13. (Read 170943 times)

Anyone up for another bounty for a OpenCL-implementation?

tsiv definitely deserves the bounty - no question!

Time for some bounty love, it seems to me.

Could people who compiled in windows please answer my question?  How/what software did you use to compile?


I have done it before, have taken C++ courses, but it was 15+ years ago.  Please enlighten me and others.

Thank you. 

Will report back my experiences with tsiv's code.

Works like a charm on my 750ti's, though it does kill windows a bit.

Pay the man already!

+1 for tsiv to recieve bounty, well done and I'm sure it will only get better from here! I'm runnung 3 seperate rigs 750 ti's and it is functioning exactly as was expressed by @tsiv, thanks a bunch!!

BTW for those of you in this thread who are asking about a windows compilation, here is the one I am using

         https://mega.co.nz/#!YckmgJpQ!etExS5ELpb309WwSgQcA3WaUPYVwnI4ix9UuWNLr4Yc

make sure you do not have any arguments at the end of you .bat, it is set to run @ 8X40 which seems best at this time.

Has tsiv been paid the bounty yet?

If not, what's the hold up?

Just some facts:
No clear criteria was ever set forth, so it's not fair to setup the criteria after a open source miner has been released. Anyone is now able to fine tune this software and it's only fair that tsiv receive his bounty.  Because he did open-source it right away, I'm sure others will now jump in and help solve a couple of the problems noted.  Remember this can't happen if the source isn't available which is the whole point of this thread!

I'm sure he will continue to re-fine it anyway as this is just a first release and tsiv has already proven to the ccminer community he contributes.  He is the reason nVidia miners have X13.  Claymore has already made at least a half dozen releases to fix issues. We are in a sad state of affairs concerning open source GPU miners with Claymore having the only other GPU software with a 5% fee payable to himself (steep) that is not open source and not available on linux and only works on recent AMD cards.

I know myself I get 375 H/s with Claymore (OCed) which is 375*.95 or 356 H/s for me.  With ccminer on 750ti in (not tried to OC yet) windows (runs better under linux I'm told) I get 335 H/s with 2 750ti cards which is right on par if you figure the price of the cards together is less or equal to one 280x.  Power requirement for both 750ti cards is less then one 280x.  I've been told it runs even better on Linux which I haven't tried yet.

tsiv's release compiles and runs under both linux and windows.  It has also been directly integrated into the most popular miner for nVidia which is ccminer.  No additional standalone software needed.

So please do the right thing and pay him the bounty he deserves if this hasn't been done yet.  He HAS done everything that WAS required.

Just looking at this logically and giving my 2 cents,
Carlo

OpenCL/ATI anyone?

tsiv you rock for developing this...

Am I to understand that the miner will bog a 750Ti desktop? 

I would have to disagree with you that pretty much every other algo bogs the machine if under Windows with primary GPU, as I mine nonstop with my rig, only has 750Ti and I can do anything in windows with no lag, no slowing of the miner (cudaminer or ccminer).  Only exceptions are GPU-intensive apps like 3D gaming.

While we are on this topic, sorry for a n00b question, but I think others would also like to know:

How do you compile under Windows?

I downloaded free trial or whatever of Visual Studio Express the other day, trying to compile whatever it was, but the software says "only for Windows Store Apps"

 

Most of the CPU use is likely just polling for completion of the kernel aka http://en.wikipedia.org/wiki/Busy_waiting, which kinda sucks but seems like standard behavior. I'll see about working around it, looks like a "are we ready yet? no? sleep a bit." approach might do the trick instead of just calling a sync which then proceeds to busy wait until the kernel completes, eating whatever cpu cycles are available to it.

There's a windows compile from one of the trusted members at the cudaminer/ccminer thread. It's not all that buggy, but also far from optimized. As it stands right now, this miner is relying heavily on CPU also, so there's certainly room for improvements (50% CPU usage with 4x750 TIs and an i7 4770 on Windows).

The pizza awaits a destination! (though waiting to get acceptance from the bounty author and to see if there is consensus on the miner performance as deserving of the bounty, might require some finishing touches)
~ Myagui

You must compile it yourself but right now it's bugged like hell for Windows.

How do I use it in windows ? It doesn't seem to have an exe ?

I dont have a card to test this but many many thanks!

https://github.com/tsiv/ccminer-cryptonight

Fresh out of the oven, don't have any performance data beyond personal testing during development with my GTX 750 Ti rig. Does around 200-220 H/s per card for a 750 Ti, both stratum and http protocols supported for pool mining. Haven't actually tested solo. Compiles and runs both on Linux and Windows pretty much straight out of the box. Linux highly recommeded over Windows. Has issues with massive lag if run on a Windows primary GPU, practically renders the computer unusable due to the hashing kernel grabbing the GPU for 1-2 seconds at a time. Pretty much what happens with every other algorithm, this is just orders of magnitude worse. Looking into breaking the kernel into smaller pieces to hopefully alleviate the problem.

Anyone find any updates to this? Nvida or AMD... anything?

This thread has nothing to do with BBR. That uses its own miner. MCN should work with the same miner as all the other cryptonote/cryptonight coins (XMR, BCN, etc.)

Haha, I like this post very much. But hmm, I am willing to donate for a BBR gpu miner, or MCN, if that is what this post is about. I can throw 10XMR into the bounty.