Pool-specific intensity setting vanishes sometimes

[Zuikkis]

This is with the latest develop version.

I'm using nicehash. Sometimes when the algorithm changes, intensity values are not changed. This results in slow hashrate or even crashes, depending on how far of the previous intensity was..

What's more, even if I manually switch pools back and forth, the "ill behaving" pool always uses the previous pools intensity! It's like the pool-specific settings are forgotten completely?

I'm using the profile-approach in my config, the relevant parts are below.. The last it happened, it switched from Neoscrypt to X11_Bak, which worked nicely, then returned to Neoscrypt but intensities were 18,18,18,18,17,18 like in X11..

{
"pools" : [
     {
          "name" : "NiceHash_Neoscrypt",
          "url" : "stratum+tcp://stratum.nicehash.com:4341",
          "user" : "xx",
          "pass" : "f0=0;f2=0;f3=4.4;f4=3.1;f5=345;f6=2.7;f7=15.7;f8=0.3",
          "profile" : "neoscrypt"
     },
     {
          "name" : "NiceHash_X11_Bak",
          "url" : "stratum+tcp://stratum.nicehash.com:3336",
          "user" : "xx",
          "pass" : "f0=0;f2=0;f3=4.4;f4=3.1;f5=345;f6=2.7;f7=15.7;f8=0.3",
          "profile" : "x11"
     }
],
"profiles" : [
        {
                "name" : "x11",
                "algorithm" : "darkcoin-mod",
                "intensity" : "18,18,18,18,17,18",
                "gpu-threads" : "2",
                "worksize": "128"
        },
        {
                "name" : "neoscrypt",
                "algorithm" : "neoscrypt",
                "intensity" : "13,13,13,13,13,12",
                "thread-concurrency" : "16384,16384,16384,16384,16384,16384",
                "gpu-threads" : "2",
                "worksize": "64"
        }
]
}

[ystarnaud]

Actually there are other relevant parts of the config. Could you post the full thing somewhere?

[ystarnaud]

Also any chance of possibly getting a debug log of the pool switching? I will run my own tests later when I have a chance but in the meantime it would be helpful. Thanks.

[Zuikkis]

Here's the full config: http://zuik.org/sgminer.conf

I try to make a log. I have never been able to trigger it when switching pools manually, but it has happened a few times when pool becomes dead and sgminer falls over to the next one.. But not every time there either.

[ystarnaud]

Can you check with 4ba8a68043a14f9f3d12c6ae3838cc802ddadd65 see if this problem is resolved? I was unable to reproduce it on my test build. Let us know, thanks.

[Zuikkis]

This might be a HW issue on my system. I have two machines, and the bug only seems to happen on one of them. And it doesn't happen very often, especially now when NiceHash seems to be 99% neoscrypt so pools don't change very often. :)

[ystarnaud]

I also have my rigs on nicehash and my settings are neoscrypt: xI 3 and Xn: intensity 19. It doesn't happen often like you said but I witnessed my miner on both algos at the proper intensities without a restart so it's promising hehe.

[platinum4]

This appears to have been fixed I think, your thoughts on the matter?

[Zuikkis]

Yeah, hasn't happened in a long time, so probably fixed.