Miner download error at initial start

[merlinsbeard69]

Hello, I downloaded v2.5.2 and ran it for the first time. I think it is having trouble downloading the nicehash excavator miner. Probably because they updated their files.

The error reads

"Cannot find .\Bin\Excavator\excavator.exe distributed at https://github.com/nicehash/excavator/releases/."

Also, there is an improved ccminer version that works best with new nvidia cards: Alexis78. I am getting about 30% more in skein algorithm and others compared to other versions. Please consider adding this version. https://github.com/alexis78/ccminer

You can find the compiled version here: https://mega.nz/#F!5vIGUCzC!nLsQtAwDRiV8MYA_P-q9yg

[merlinsbeard69]

Nevermind, I see that the issue is with nicehash being jerks (how ironic). Their miner isn't even that fast.

Still, I think this blocks mpm from initializing and I would like to see ccminer-Alexis78 added (instead?). Thank you for everything

edit: after testing alexis78 version was much higher than Tpruvot but lower than 'SP' mod for skein. I have included benchmarking results with 1080 + 1080 ti:

[Ondalf]

Try using same intensity for "regular" miners as Alexis version uses by default. You might not surprise of the results... Also, the time used for benchmarking, its too low. Reason for this; the time to achieve somewhat okay hashspeed and to force miner API to push out the result for the MPM takes time. For example, ccMinerAlexis78 and skein and 1080. That takes around two minutes for hashrate to stabilize. Its similar with PID controller, like it dampens the jumps and shows value based past 10 submits or so (even though it shows the full speed in grey)...

[Ondalf]

This is basically, what i mean. Sometimes the hop from zero to full hash speed (like at timestamps 5:42:36 and 5:42:40) happens quite slow. Cannot remember exactly, which ccminers suffered from this. Some miners require more time and thus that "one cycle" (60 seconds) might be too low time to benchmark.

[2017-06-04 05:42:17] Adding 536870912 threads to intensity 30, 1610612736 cuda threads
[2017-06-04 05:42:17] Starting on stratum+tcp://mine.zpool.ca:5766
[2017-06-04 05:42:17] NVML GPU monitoring enabled.
[2017-06-04 05:42:17] NVAPI GPU monitoring enabled.
[2017-06-04 05:42:17] 1 miner thread started, using 'blake2s' algorithm.
[2017-06-04 05:42:33] Stratum difficulty set to 2
[2017-06-04 05:42:33] blake2s block 1233685, diff 30866.819
[2017-06-04 05:42:33] GPU #2: Intensity set to 30.5, 1610612736 cuda threads
[2017-06-04 05:42:36] GPU #2: Gigabyte GTX 750 Ti, 675.48 MH/s
[2017-06-04 05:42:40] GPU #2: Gigabyte GTX 750 Ti, 914.02 MH/s
[2017-06-04 05:42:41] accepted: 1/1 (diff 3.191), 854.33 MH/s yes!
[2017-06-04 05:42:45] GPU #2: Gigabyte GTX 750 Ti, 913.45 MH/s
[2017-06-04 05:42:45] accepted: 2/2 (diff 3.616), 879.83 MH/s yes!
[2017-06-04 05:42:49] GPU #2: Gigabyte GTX 750 Ti, 913.99 MH/s
[2017-06-04 05:42:51] blake2s block 9599, diff 111206.384
[2017-06-04 05:42:56] GPU #2: Gigabyte GTX 750 Ti, 913.98 MH/s
[2017-06-04 05:43:01] GPU #2: Gigabyte GTX 750 Ti, 913.78 MH/s
[2017-06-04 05:43:05] GPU #2: Gigabyte GTX 750 Ti, 913.80 MH/s
[2017-06-04 05:43:06] GPU #2: 1332 MHz 22.68 MH/W 39W 67C FAN 89%

[fonyo]

Consider changing the default interval to >60 seconds or adding an exception for the miners in question (if it's worth it)

[aaronsace]

It compensates for fluctuations; it won't take the reading until it's stable. It also wait's 60 seconds before trying to take a reading (there is an option to increase on the command line).

I'm confused as to why alexis78's fork would be faster. It hasn't been updated for a year; has tpruvot not copied the improved code into his version? I'd prefer to have an actively developed miner for most of the algorithms if possible and there is already a lot of ccminers included.

[jakewarren81]

spmod is a better fit than the alexis78 fork, for multi mining. I've done extensive testing and the alexis fork is very inconsistent in starting all gpus and can to 5-10 minutes sometimes. spmod comes up with all immediately and its hashrates are very close to alexis. So factor in the wait time and you probably come out the same.

If you are running just alexis and not switching, then, sure it's better to use that. Just not for multi mining.