Closed nerdralph closed 6 years ago
I don't understand what you mean. The DAG is generated on GPU.
Anyways, send a pull request if you have done it already.
I know the DAG is generated on the GPU. Showing the percentage progress is of little benefit since it only takes a few seconds. I provided a link to my code, since there are so many changes since I implemented it over a year ago that it's not a simple diff to merge.
Its already done in latest git release.
[2018-01-15 20:59:47][info][miner]: cu 20:59:47|CUDA2 Generating mining buffers
[2018-01-15 20:59:47][info][miner]: cu└╧ 20:547|CUDA6 Allocating light with size: 38141888
[2018-01-15 20:59:47][info][miner]: cu└╧ 20:547|CUDA1 Generating mining buffers
[2018-01-15 20:59:47][info][miner]: cu└╧ 20:547|CUDA5 Generating DAG for GPU # 5 with dagSize: 2441084288 gridSize: 4096 &m_streams[0]: 0x7ff62c000960
[2018-01-15 20:59:48][info][miner]: cu└╧ 20:54:37|CU2 Generating DAG for GPU # 2 with dagSize: 2441084288 gridSize: 4096 &m_streams[0]: 0x7ff63c000960
[2018-01-15 20:59:48][info][miner]: cu└╧ 20:54:37|CU3 Generating mining buffers
[2018-01-15 20:59:48][info][miner]: cu└╧ 20:548|CUDA7 Generating mining buffers
[2018-01-15 20:59:48][info][miner]: cu└╧ 20:548|CUDA4 Generating DAG for GPU # 4 with dagSize: 2441084288 gridSize: 4096 &m_streams[0]: 0x7ff628000960
[2018-01-15 20:59:48][info][miner]: cu└╧ 20:548|CUDA6 Generating mining buffers
[2018-01-15 20:59:48][info][miner]: cu└╧ 20:548|CUDA3 Generating DAG for GPU # 3 with dagSize: 2441084288 gridSize: 4096 &m_streams[0]: 0x7ff630000a70
[2018-01-15 20:59:48][info][miner]: cu└╧ 20:548|CUDA1 Generating DAG for GPU # 1 with dagSize: 2441084288 gridSize: 4096 &m_streams[0]: 0x7ff638000b30
[2018-01-15 20:59:48][info][miner]: cu└╧ 20:54837|CUDA7 Generating DAG for GPU # 7 with dagSize: 2441084288 gridSize: 4096 &m_streams[0]: 0x7ff624000960
[2018-01-15 20:59:48][info][miner]: cu└╧ 20:548|CUDA6 Generating DAG for GPU # 6 with dagSize: 2441084288 gridSize: 4096 &m_streams[0]: 0x7ff620000960
[2018-01-15 20:59:54][info][core]: Sending stats to server...
[2018-01-15 20:59:59][info][miner]: Б└╧ 20:55937|stratum Received new jobb9f5b0243e seed: #ec6e8e0d8056cf9eb61046eaeb754b8e target: #0000000112e0be826d694b2e
[2018-01-15 20:59:59][info][miner]: Б└╧20:59:59|stratum Received new job #7af31371 seed: #ec6e8e0d8056cf9eb61046eaeb754b8e target: #0000000112e0be826d694b2e
[2018-01-15 20:59:59][info][miner]: m└╧ 20:559|ethminer Speed 0.00 Mh/s gpu/0 0.00 54C 40% gpu/1 0.00 46C 40% gpu/2 0.00 53C 40% gpu/3 0.00 54C 40% gpu/4 0.00 54C 40% gpu/5 0.00 54C 40% gpu/6 0.00 52C 40% gpu/7 0$
[2018-01-15 21:00:00][info][miner]: Б└╧ 1:00:0:40|CUDA1 Solution found; Submitted to eu2.ethermine.org
[2018-01-15 21:00:00][info][miner]: Б└╧ 1:00:0:40|CU1 Nonce: 0x01ceba8056f7b947
[2018-01-15 21:00:00][info][miner]: Б└╧ 1:00:00|stratum Accepted.
Showing the percentage progress is of little benefit since it only takes a few seconds.
We don't do that anymore, regardless of how long it takes. Let me qualify that... we don't do it for cuda but I think it's still there in opencl.
@jean-m-cyr Just one line to delete. https://github.com/ethereum-mining/ethminer/blob/master/libethash-cl/CLMiner.cpp#L727
Yeah, I know. But, I try to focus on cuda since cl is not my thing. I don't know how fond the cl crowd is of their startup noise. Seems some are fussy about their console output! I got a few negative revues when I eliminated all the redundant (identical) per GPU 'got work' messages with every new job arrival in cuda.
I don't see any value in it other than maybe when debugging. I'll push a request up the pole and see who salutes.
With on-GPU DAG generation, the DAG load progress should be removed and replaced as I did with ethminer-nr. "GPU1 2.27GB of DAG data generated in 3885 ms." https://github.com/nerdralph/ethminer-nr/blob/master/libethash-cl/ethash_cl_miner.cpp#L483