When miner is reporting 0.00 sol/sec after a while nothing happens -> add watchdog please

[frankcoderL]

Bug description When a GPU card stops mining (e.g. when GPU became to hot) the GPU mining doesn't resume

To Reproduce Steps to reproduce the behaviour:

1. Run miner
2. Let the miner run for a while until a mining problem occurs on a GPU card that results in 0.00 sol/sec

Current behaviour GPU stops mining after a GPU problem occurs. Only when manually interrupting the miner and manually restarting the miner the miner resumes

If the OpenCl miner controls more than one GPU card it seems to be more vulnerable to fall back to 0.00 sol/s. A card that stopped regularly after a while in a group of devices, does not has this behaviour when i start a separate opencl miner with the --device switch

Expected behaviour Watchdog in miner: the miner should try to restart automatically when a mining problem occurs.

Platform and build

• Linux
• L(Ubuntu) 10.04 LTS
• AMD-GPUPRO driver
• mainnet-release and current git

Screen output

...
Performance: 8.13 sol/s 3.87 sol/s 14.73 sol/s 12.00 sol/s | Total: 38.73 sol/s 
Performance: 10.33 sol/s 4.47 sol/s 13.67 sol/s 11.47 sol/s | Total: 39.93 sol/s 
...
Performance: 0.00 sol/s 3.93 sol/s 15.13 sol/s 12.93 sol/s | Total: 32.00 sol/s 
Performance: 0.00 sol/s 2.73 sol/s 15.07 sol/s 13.40 sol/s | Total: 31.20 sol/s 
...
-> COMMENT: the first GPU card will never recover and stays on 0.00 sol/s ,
    only when manually stop/start the miner, the first GPU will start mining again