Re: [ANN][DRK|Q2C|QRK|MYR] sph-sgminer - multi-algo GPU miner

I saw some guys started to test it - any results?

Hi! Just registered especially for this theme because of its significance.
So, only one of many kernels are working, the quarkcoin one.
But it's working well!
On my test 7950 it shows 1.500 M avg hashrate with 5.000 M WU.
On the site side I see increasing value of my hashrate aiming to WU. After 25 min its amount reached 3.5 M.
All diffs are acceptable with no rejects and HW errors.
Amount of diffs are 0.5 and 1.0.
Good work, Feeleep!
But this coin is not profitable even with this results for me.

does miner hashrate equals what pool was saying?

Hmm.. So to understand correctly:

The miner which gets rejected on for example suprnova works well on coinmine now and reports the correct hashrate ? no more fake shares ? and it is fixed by a tweak at stratum on coinmine only Huh

i can confirm that exactly yes - i did not download a new compile, i used the same miner, i used it on a windows machine (only one i have) and its the first download from the topic:

works on Feeleeps SRC pool which is the Quark algo as i left it exactly and only changed the pool address .

did not work on any other Quark (or clone) pool, and gave back {share target error} ~ < the error

Thanks guys for testing. There is still something not correct between cpuminer-gpuminer-stratum so my solution if following:
- GPU miners: diff1 = 0x000000ffff000000000000000000000000000000000000000000000000000000
- CPU miners: diff1 = 0x00000000ffff0000000000000000000000000000000000000000000000000000

but GPU miner share is 256 time less valuable than CPU share - dont ask me why but it works. If there is anyone who have an idea let me know please.

I will set up GPU stratum port for all my quark-based coins (Quarks, Securecoin, Particle, Frozen) later today

feeleep

I'd set up a pull request to sph-sgminer on git, which fixes the "share above target" error on quarkcoin kernel and which calculates the Network diff correct.

Here is the link to Uploaded with a precompiled version.
http://ul.to/5xlcv75a

I will have a look after the "real" failure in that code.
but im not so familiar with c.

will you be publishing the source on Git ?

I'd set up a pull request on git.

So the Source is on my git:

https://github.com/DJstone/sph-sgminer