Finally, and please don't take this the wrong way, but providing those binary files in the built is not a good idea for several reasons. One of them being that each GPU even same model GPUs behave differently due to the fact that each memory chip is different thus making your binary only truely efficient for your cards.
I don't think so. Memory is not currently used in the miner at all.
But it'll be interesting to read about this. Please send a link if you have it. Thanks
I didn't say memory was used nor that it was used in the miner. I said the memory chips (withing the GPUs) are all different thus each configuration is different. Part of those binaries data include how the GPU's memory chip perform thus if you make them with your own GPUs they will be 100% customized for your GPUs. 2 GPUs of the same brand and same model might very well not perform the same way because of the memory chips inside them. Hope it's clearer like this.
This is interesting information and that's why I'm asking if you can point me to some article on this. I really would like to know more about .bin files that AMD uses. NVidia .bin are really easy, they are clear text.
An interesting thing is that Piticarn and Tahiti binaries are interchangeable.
And you probably know how the .bin naming strategy is used in sgminer. So if you had two R9 290 from different brands in your mining rig, they'll still be using the same .bin. Not different.
I'm by no means an expert. I simply used my logic since when configuring a GPU in the .conf file it nearly always lead to having slightly different configuration values for 2 different cards (in order to maximise their potential) although they are identical brands/model. You can actually see it in the hashrate (above) of the 4 cards. You will notice that the 1st and last one are slightly hashing more than the 2nd and 3rd one. The 4 cards are 280x Windforce Gigabyte cards but 2 of them have Elpida chips while the 2 others have the Hynix chip.
Anyways, I also installed grs-sgminer. I keep getting all rejects. I've disabled tc, lookup-gap, changed to groestlcoin kernel, disabled difficulty multiplier and added the normal hashing parameter. This is my config:
"api-listen" : true,
"api-allow" : "W:127.0.0.1",
"api-mcast-port" : "4028",
"api-port" : "4028",
"_device" : "0,1,2,3",
"intensity" : "22",
"_difficulty-multiplier" : "0.0039062500",
"_rawintensity" : "8191",
"gpu-threads" : "2",
"worksize" : "256",
"_thread-concurrency" : "0",
"shaders" : "2048",
"gpu-engine" : "1200,1200,1200,1200",
"gpu-memclock" : "1500,1500,1500,1500",
"gpu-powertune" : "0",
"_lookup-gap" : "2",
"auto-gpu" : true,
"auto-fan" : true,
"gpu-fan" : "0-100",
"temp-cutoff" : "89",
"temp-overheat" : "85",
"temp-target" : "65",
"temp-hysteresis" : "3",
"failover-only" : true,
"_scrypt" : true,
"_vectors" : "1",
"kernel" : "groestlcoin",
"normal hashing" : true,
"gpu-memdiff" : "0",
"gpu-vddc" : "1.200,1.200,1.200,1.200",
"gpu-dyninterval" : "7",
"gpu-platform" : "0",
"hotplug" : "5",
"log" : "5",
"no-pool-disable" : true,
"queue" : "1",
"scan-time" : "1",
"expiry" : "120",
"shares" : "0",
"kernel-path" : "/usr/local/bin"
}
EDIT: I just saw the subtlety. The normal-hashing parameter goes just under the pools config info (below the user).
EDIT 2: Sweet jesus! just getting 2.8Mh more per card than with sph-sgminer. It's like having ~2 more cards
(I was mining ATH on suprnova pool) When I changed kernel back to groestlcoin, it started to show my hashrate on the pool again.
