How to compile v1.0.1 for Windows?
Shit, I wish I knew how to compile Sia at all, and I'm a developer.
come on wolf ...
now ur playing
...
will try testing sia today with the first machine - 4 x gigabyte 280x oc ...
IF i can get the thing running of course ...
#crysx
Actually, I'm not. I've never dealt with Go before, and after trying to compile Sia, I now hate it.
holy patushka batman! ...
i have been trying to compile the gpu-miner and error after error after error - im close to giving up ...
where can i get some static compiled binaries already compiled for linux ( fedora 19 and 20 ) please? ...
cant believe the difficulty in compiling this miner ...
wolf0 - pallas ... how on earth did you guys get it compiled - let alone running with high hashrates? ...
#crysx
Oh, that shit is simple. If you're on PiMP, as I think you are, simply add -L/opt/AMDAPP/lib/x86 to LFLAGS before -lOpenCL.
If you're on 64-bit PiMP, make that -L/opt/AMDAPP/lib/x86_64
no not on pimp mate ...
just running from fedora 19 x64 ... thats the design i originally needed to make them sellable machines with little downtime ...
still wont compile under fedora 19 x64 ... but managed to download the precompiled gpu-miner ...
now have them running through nested scripts to run automatically from the main script ...
so thanks for the help wolf ...
the issue is at the moment - the four mining threads on this machine are only mining at around 50MH ... these cards are gigabyte 280x oc ...
how on earth can i make them run faster? ...
there doesnt seem to be any parameter switches within gpu-miner to control gpu and mem clocks ( like in sgminer ) ...
anyway of doing that? ...
#crysx
Old school. You need to use aticonfig. But that's not going to help you much.
What would help you, is modifying the global amount of work-items per kernel launch, I think. I've not tried it on that miner, obviously, but I think it'll help.
You really need to modify the code - specifically the call to clEnqueueNDRangeKernel() (or modify a variable passed to it), to do this, however. Now, the command line params give you a sort of indirect, hobbled control with the -c parameter - but instead of raising the number of work-items for the GPU, it instead simply loops kernel executions, which... if I'm reading this correctly, their comment means that they think doing a shitton of calls to clEnqueueNDRangeKernel() with a small amount of work-items will prevent locking up. It might help a tiny bit... but it'll also likely cause a shitton of overhead as instead of using the GPU to its fullest, they feed it tiny bits of work at a time - I would bet some compute units are idle or close to it for a large amount of time, now.
Hmmm... my advice? Just change a few numbers around in the code - raise amount of global work-items until your weakest cards begin to lock up, then raise the amount of time between calls to siad to get work with the -s option. Drop the -c option like a hot brick - raising the amount of global work-items while continuing to loop like that will probably cause massive instability.
wolf ...
can i just buy your miner mate? ...
...
there is NO way that is going to happen with the workload i have at the moment ...
apart from that - i am no developer ... i can do enough to get me INTO a lot of trouble -and that is about it ...
ill leave the coding to you and the devs ...
btw - the blocks that have been mined are huge now - which i think may be a false stat ...
10943 / 11023 / 7953 / 10897 - blocks that have been mined by the measly 50MH per card ... that does not seem right ...
#crysx
I didn't suggest really anything that would even be exceedingly difficult for someone with zero knowledge of coding. I told you what number you needed to change, how to change it, and roughly where it was. The rest is simply command line parameters. Of course, it's not going to be exceedingly simple for someone who doesn't have a clear idea of what to do from having done it before - if it was, there wouldn't be any value in it, don't you think?
The blocks mined stat gets clobbered when siad derps - it sends a REALLY low target, I think, making you think you're mining blocks at an alarming rate. Just restart the miner.
Anyways, I'm open to more questions. You asked me how to make the GPUs run faster; in my previous post, I gave you the tools to do it - but you've have to do it yourself. Besides the fact it's good to learn new things, if I do it for you, there's no reason I shouldn't do it for someone else, or everyone else. Then you'll be right back where you started, except you'll be seeing higher numbers when you look at your miner... and at the network difficulty.
makes a great deal of sense - all that you have suggested ...
but time is of the essence for me - and the farm for that matter ...
which is why the experts in their field do what they do best - and if you want to offload or sell your wisdom via your work - then its a damn good way to do it ...
which ever way - i DO appreciate the guidelines you have supplied ... now for the time to play - which at the moment is non-existent ...
for a man like me - too much fiddling is time wasted - not a learning experience that can advance what i do here ...
this is why i would much rather leave it to you ( the expert ) - and buy your wares if you are willing to sell ( the really lazy but efficient way of doing it for people like me ) ...
btw - i really would like to learn coding ... i was stupid and dropped it all when i was younger and in uni ... there is just no time for me on a personal level to do so ... yet ...
tanx again wolf ...
#crysx
