Quote from: chrysophylax on March 30, 2018, 04:06:42 PM

Quote from: Newton90 on March 30, 2018, 09:37:31 AM

Quote from: chrysophylax on March 30, 2018, 04:31:35 AM

Quote from: hashappliance on March 23, 2018, 07:15:25 PM

Quote from: Zebra# on March 22, 2018, 08:51:38 PM

Quote from: MMichaelix on March 22, 2018, 07:14:26 PM

Quote from: Zebra# on March 22, 2018, 12:48:43 PM

How do I obtain my private key from my Boolberry windows wallet? I like to document my keys on paper and keep them in a safe just in case.

Also, roughly how many coins can I expect per day can I expect with a gtx 1070? I've never mined on wild keccak before so I'm not sure which hashrate to enter into the calculator.

Im not too sure about your first question (sorry about that) but as for the second one, Ive been using a 1060 for the past couple months and its comfortably getting from 1.1 to 1.2 coins a day which is about 1.3 to 1.45 mH/s. But i used to use a 780 and i was getting just as much so that kinda surprised me. Also a friend is using a 1080 and he was getting maybe 100 more kH/s than with my 1060. Maybe I'm just doing something wrong with configs or something but Ive never been able to get past 1.5 mH/s with just one card.

Thanks!

1080's have GDDR5X instead of GDDR5 on 1060/1070, and this is why the 1070 is about the same performance on ETH and more efficient than the 1080.

The most efficient card I have seen is the RX 470, which is about 16 kH/watt, the Titan V is 40+kh/watt, but at $3,000, I hate to call it the most efficient. Wink

Here are some #'s from my experience.

GTX1060 - 1.2-1.3 MH/s (mem +120 core +800)
GTX1070 - 1.7-1.8 MH/s (mem +0 core +1400)
RX470 - 1.6 MH/s (mem 1990 core 1190)
RX570 - 1.6 MH/s (mem 1980 core 1160)
RX580 - 1.9-2 MH/s (mem 2040 core 1200)
RX Vega 64 / FE - 2.4 MH/s (I believe this could be optimized to be 4 MH/s+ based on its performance on CryptoNight)
Titan V - 4.2 MH/s

Optimization plays a massive part here.

If the Algo is optimized for the SDK/Driver, then you will get better performance. In our case, nVidia (CUDA).

#crysx

as i know any improvements probably aren't possible:

https://bitcointalk.org/index.php?topic=693118.msg17229968#msg17229968

There is ALWAYS room.

It depends on whether a small or larger investment of time and money is worth the small or large improvement in code.

We had (and still to this day) one of the most optimized miners in Crypto. I subsequently withdrew the miner from the market due to unforeseen circumstances, but it nonetheless is still one of the most highly optimized miners to date, and there is still room to move with the progression of SDK and code optimizations, for the Algos that we support in CWIgm.

Depending on what 'can; be done will determine what 'should' be done. Either way, it is worth a shot if it will give an edge in mining.

We just cannot justify the cost involved at the moment for it.

I will fire up the miner again soon when I have the time, and see if WildKeccak was one fo the Algos we worked on. If so, there will be some testing to be done.

#crysx

Agreed.

We need a new miner built on GCN assembly. Nvidia has an advantage on raw compute, but AMD has VOP_DPP modifier at no cost compared to __shfl, which is far from free! Even though Nvidia is faster per thread, it only has 32 threads/warp while AMD has 64/wavefront, but you can parallelize to even that out. Meanwhile it would be nice to get the Vega 10 RRG.