Topic: [ANN] sgminer v5 - optimized X11/X13/NeoScrypt/Lyra2RE/etc. kernel-switch miner - page 248. (Read 877859 times)

Just a heads up the current way the miner is written/coded; it will NOT let you effectively switch between using an xintensity and using a normal intensity.

To mitigate this I did the multiplication of Shaders X [xI setting] and then hardcoded the rawintensity values in my .conf file, I will check back in once there's a few hours under the belt.

This noticeably happens when you switch from X11/X13 on xI:256 to Keccak on I:13 - after the miner switches back to another algorithm, the miner gets stuck on I:13 and obviously you see a very shitty hash rate.

Can we make pool-algorithm SUPERCEDE/cascade over the standard setting from the .conf file?  pool-algorithm works, but it doesn't take precedence or importance when using the new Nicehash BPASMAM released yesterday.

Just a heads up I'm not typically a Nicehash miner but I have built a config with all 4 algorithms for R9 290X that i feel is somewhat optimized.  It is scary watching the miner as it builds the kernels in between the first time it switches algorithms though... That hashrate drop is like whoop.

But it appears the stratum for Nicehash actually works by disabling the other pools that are less profitable?  I'm not quite sure, but you can't force an algo just by picking any of those multiport stratum multiports.

Is it a general problem, when one wants to run this multi-algorithm CGWatcher fucks the shit up?

Figured we would start running into issues with trying to drop in optimized kernels.

Has anybody gotten a working marucoin-mod with the added features from pull request 217 - in Windows binaries?

Could not make it work, all reject share.

Any info on a config file that should work with 280X?

In my rigs,
The efficiency of the 'scrypt' and 'x11' are different depending on worksize.
How can i use 'pool-worksize' configuration?

and, ckolavis kernel is faster than zuikkis kernel. (scrypt, HD5850 378kh vs 320kh)
but ckolavis kernel ignore 'pool-intensity' configuration.
How should i do it? I want 'pool-intensity' configuration using ckolavis kernel.

Sorry for my bad english.

is there a specfic .conf i need to update ? and do i need to change the bamt.conf ?

Error: clSetKernelArg of all params failed.
Can anyone help with this?

Am I the only one that cannot mine on the X11/X13 ports anymore since today?

It look nice but I'm gonna wait until the powerconsumption is also in the calculation..x11 is about 0.5 less power than scrypt. So the profit should be minimal twice as much between scrypt and x11.

Is it possible to only switch between x11 and x13 ? And if the algo choose a scrypt coin u go to your own failover?

I also have to wait untill my hd6970 is working with the miner (x13) I have a rig with 7950 and 6970 in it.

NiceHash brought you another innovation: NiceHash.com Best Profit Auto-Switching Multi-Algorithm GPU Mining. You can get more information here: https://www.nicehash.com/multialgo/

Again, questioning the effectiveness or usefulness in switching between [scrypt, scrypt-N] and [x11,x13].

Still skeptical as to whether you will see peak performance without having to experience a reboot.

Also, are there any pools that have a muliport stratum, for multiports?  i.e. the pool analyzes which algo+coin is the top breadwinner and serves the work through different ports?

We're missing something here.

this now switches and uses the correct algorithm, but does not change any of the gpu settings such as intensity or engine. any ideas?


EDIT: it will change the intensity and gpu engine if you use the "pool-gpu-engine" and "pool-intensity" paramters instead of "gpu-engine" and "intesnity". does not set the intensity correctly when opening (sets it to unless you switch to another pool and then switch back. setmemclock error listed, but i have my memory set to the same speed for all anyway so i shouldn't need that. after a couple of minutes the miner shuts down and displays a summary of works.

[12:38:26]
Summary of runtime statistics:

[12:38:26] Started at [2014-06-03 12:36:36]
[12:38:26] Runtime: 0 hrs : 0 mins : 48 secs
[12:38:26] Average hashrate: 6.4 Megahash/s
[12:38:26] Solved blocks: 0
[12:38:26] Best share difficulty: 0.057
[12:38:26] Share submissions: 28
[12:38:26] Accepted shares: 28
[12:38:26] Rejected shares: 0
[12:38:26] Accepted difficulty shares: 0
[12:38:26] Rejected difficulty shares: 0
[12:38:26] Reject ratio: 0.0%
[12:38:26] Hardware errors: 0
[12:38:26] Utility (accepted shares / min): 36.69/min
[12:38:26] Work Utility (diff1 shares solved / min): 0.11/min

[12:38:26] Stale submissions discarded due to new blocks: 0
[12:38:26] Unable to get work from server occasions: 0
[12:38:26] Work items generated locally: 50
[12:38:26] Submitting work remotely delay occasions: 0
[12:38:26] New blocks detected on network: 5

Phate

Hey all

does anyone have any ideas why i am getting fatal JSON errors here?

{
"pools" : [
        {
      "url" : "stratum+tcp://#server:port#",
      "user" : "#user.worker#",
      "pass" : "#password#",
      "name" : "x13",
      "algorithm" : "marucoin-mod"
                "gpu-engine" : "1145,1114",
      "gpu-memclock" : "1500,1500"
                "thread-concurrency" : "32760",
                "intensity" : "19,20",
                "gpu-threads" : "1"
      "worksize" : "128"
      "no-restart" : "true"
        },
   {
      "url" : "stratum+tcp://#server:port#",
      "user" : "#user.worker#",
      "pass" : "#password#",
      "name" : "x11",
      "algorithm" : "darkcoin-mod"
                "gpu-engine" : "1140,1110",
      "gpu-memclock" : "1500,1500"
                "thread-concurrency" : "32760",
                "intensity" : "20,20",
                "gpu-threads" : "1"
      "worksize" : "128"
      "no-restart" : "true"
        },
   {
      "url" : "stratum+tcp://#server:port#",
      "user" : "#user.worker#",
      "pass" : "#password#",
      "name" : "scrypt",
      "algorithm" : "zuikis"
                "gpu-engine" : "1040,1025",
      "gpu-memclock" : "1500,1500"
                "thread-concurrency" : "32760",
                "intensity" : "21,21",
                "gpu-threads" : "1"
      "worksize" : "256"
      "no-restart" : "true"
        },
]
}

thanks in advance

phate

Yessir, Elun with the fucking save.

brb gonna go be BLEEDING EDGE for a little fucking bit y'all.

Edit: That's a bingo.  x13 is up and going with this miner, will test for 24hours and report back in with any SICK warnings for Hawaii chipsets.

So far so good it appears, not too sure what the difference is between 4.2.1 and 4.1.0 other than the pool settings now.

Looks to be a pretty smooth patching process of migrating the sph-sgminer over to the unified.

Keep up the good work, I'll be waiting in for test results from others!!

Edit 2:  This build seems to be missing the uptime counter [- this is vital for keeping threads alive], and I would assume is missing the ADL GPU idle patch; it does however contain the show-coindiff patch.

DON'T FORGET THAT UPTIME COUNTER Y'ALL.

Windows binaries with marucoin-mod. https://www.dropbox.com/s/z59qwtcc0kb1l2q/sgminer_v5_0.zip
hd5xxx/6xxx still have problems, but lasybear is working on it, will be fixed soon.

Cool, so I'll wait for you to compile it for Windows then, I appreciate it!

Go compiled it yourself, or try being patient for a whole day/two...

The marucoin-mod edition was just released (~6 hours ago) - if you want bleeding edge, learn to compile from source.  Else, wait like everyone else.