I've upped the diff a bit, it was still quite low due to my testing earlier, should be more reasonable now, let me know if that works.
Good Evening, First, overall looks great, looking forward to launch. Just tested 7 Gig against it. Was still getting a lot of reject due to duplicate shares and diff still seems low getting 1-2 share a second. Side note: Blake256 is brutal on GPUs. I took them down to -20 power tune which loses some hash but otherwise topping 90 even with a fan on them.
I'll bump the diff a bit more when necessary. cgminer is really a wreck at its current state, I hope it will be fixed until launch, those duplicates all come from cgminer and (imho) the limitations of the getwork protocol. Stratum or at least GBT is the only real solution to get these issues completely fixed. We're currently mining like in the early days of bitcoin from 2011
Incorrect incrementing of nTime and duplicate hashes have been fixed with this pull request. There are a number of other fixes in dcrd master that also prevent stales from mining. We should have new binaries tomorrow. https://github.com/decred/cgminer/pull/10 For the highest efficiency, run one instance of cgminer per GPU.
Looks good here, it fixes the duplicate issue and also the roll_work ntime issue is fixed. I've pushed the changes to my github Repo too, now we need the pull merged into the official branch and then new windows compiled version.
Sorry I had to fix ceejep's fix, the roll_work bug was still not fixed, its now fixed in this commit: https://github.com/ocminer/cgminer-decred/commit/ea8684794ed5af40c1531a965b4fd1a28c3de5d1 ceejep please compare my roll_work function with yours and merge into your branch
I think neither solution was fully correct. The ntime is already in big endian, so there seems to be no need to swap it at all. Code: static void roll_work(struct work *work) { uint32_t *work_ntime; uint32_t ntime; work_ntime = (uint32_t *)(work->data + 136); applog(LOG_DEBUG, "work_ntime %x", *work_ntime); ntime = *work_ntime; ntime++; applog(LOG_DEBUG, "work_ntime incremented as ntime %x", ntime); *work_ntime = ntime; applog(LOG_DEBUG, "work_ntime after \"swap\" %x", *work_ntime); local_work++; work->rolls++; work->blk.nonce = 0; applog(LOG_DEBUG, "Successfully rolled work"); /* This is now a different work item so it needs a different ID for the * hashtable */ work->id = total_work++; } Code: [2016-02-03 23:56:11] Pushing work from pool 0 to hash queue [2016-02-03 23:56:11] work_ntime 56b2d9eb [2016-02-03 23:56:11] work_ntime incremented as ntime 56b2d9ec [2016-02-03 23:56:11] work_ntime after "swap" 56b2d9ec [2016-02-03 23:56:11] Successfully rolled work I'll force push the update to my branch.
Just FYI, please use the latest version of cgminer as several bugs have been fixed there. You'll get rejected shares (duplicate/time-too-new) if you're still using the old version. The new version is rel 0.0.2 as binarys or just pull from github: http://github.com/decred/cgminer
can't compile. on ubuntu made next: 1) git clone https://github.com/decred/cgminer 2) ./autogen.sh output ------------------------------------------------------------------------ cgminer 3.7.2 ------------------------------------------------------------------------ Configuration Options Summary: libcurl(GBT+getwork).: Enabled: -lcurl curses.TUI...........: FOUND: -lncurses OpenCL...............: Detection overrided. GPU mining support DISABLED configure: error: No mining configured in in PC two 750Ti installed. if i will try to you old binary that was provided at this forum it will work. what can be an issue?
Thanks. actually i just try updated binaries for linux and it work just fine. your pool now has +1Gh's
Anyone has any idea why i am not getting any shares? I tried with all the possible settings but still without success. No rejected no accepted
Really weird, did not see that before .. Does it work when you mine locally directly to your wallet ?
Can't really decide that, miner is mining but since my speed is pretty low i don't know if i can get a block or not
What makes me wonder is that string about dynamic mode.. usually you DONT want dynamic mode.. You want a fixed Intensity..