Firstly, thanks to Claymore for the improvements here! On most of my rx 470 4GB cards, I am seeing about a 3-4% increase in hash-rate!

However, I was hoping v11 would also alleviate an issue I've been seeing since starting to use v10.2 a while ago.  For some reason, GPU 0 on all 4 of my rigs acts very strangely... most of the time hashing well below the others... fluctuating between 0 and 450, while the other 5 are right at the 700 mark.  Now, if this were only on 1 rig I would think the GPU is going bad, but since it is the exact same GPU slot 0 on all 4 rigs, it is more likely some kind of setting that i need to tweak.

Has anyone else noticed this on their rigs?  If so, any suggestions on how to fix it?

FYI: after just upgrading all 4 rigs to v11.0, GPUs 1-5 are hashing better by about 3-4% but GPU 0 is still bad on all 4. Although, it doesn't seem to be fluctuating as much... pretty much stuck at 232 on 3 rigs and weirdly, 120 on the 4th.

Any help resolving this would be much appreciated!

On an earlier rig setup I found that "set GPU_FORCE_64BIT_PTR 1" worked where "setx GPU_FORCE_64BIT_PTR 1" didn't...odd. (I've read, somewhere, that the 'setx' comands are not really necessary).

Although Cryptopia says you can't mine to there ETN wallet I'm sure I've read of people doing it (might give it a try myself), but I got an address & keys from a paper wallet generator and mined to that wallet, then got the comand line wallet up and running and put the wallet details in to the CLI wallet. The CLI wallet can also generate addresses. It can take some time for the deamon to sync but I do it most days to keep it upto date, then start the wallet and the mined coins will show when it gets to the mined block.

Some help with the wallet:
https://downloads.electroneum.com/offline_paper_electroneum_walletV1.5.html
https://medium.com/@xikimay_33996/importing-your-paper-wallet-to-your-cli-wallet-9e788be08a1a
https://medium.com/@xikimay_33996/how-to-create-an-electroneum-wallet-and-how-to-use-it-b00b787d5bd7

Thanks a ton for that tip! even though my GPU 0 was showing exactly the same MB available as other 5 GPUs I still gave your method a shot.  on startup, all were at -h value auto of 892, so I lowered GPU 0 to 812 as a test and it worked immediately!  Got that card up to 672 (from 120)!  I may play with it and inch up a bit but that's already a huge improvement! Now i will make that same change to my other rigs! Thanks again! Really appreciate it!

Someone already gave his settings in this topic : https://bt.irlbtc.com/view/638915.msg26739857#msg26739857

Well this is depressing... I've seen others report hashrates for the RX 570 in the 800+ range, but I am getting more like 415-425. My RX 570 has Elpida memory and the powersave rom from Anorak with core clock at 1150 and memclock at 1900 (no undervolting). GPU-Z reports 100% GPU load, 24% memory load and GPU-only power draw of 60W (at least that's fair, given the hashrate). My command line currently reads:

NsGpuCNMiner -h 850 -allpools 1 -xpool stratum+tcp://bytecoin.uk:5555 -xwal <BCN address> -xpsw x

Where "BCN address" is my actual address, of course. Claymore is setting -dmem to 1 on its own (I had set it to 1 myself) and was setting -h to 896, but I thought overriding that (with a slightly lower value) might improve things and it did, by a whopping 10 H/s.

I chose BCN to mine because it has a very low difficulty so it would be easy to see how well it is working (I'd rather mine a coin with a less sordid history, all things considered), but if this is any reflection of how ETN or XMR will mine then I don't think I'll bother.

EDIT - looking through the logs I see a lot of AMD adapter not found:

12:15:45:826   44c   ADL: Cannot find any AMD adapters

Which is strange, because it recognizes it as an Ellesmere when it starts up.

Guys, please help me. I've searched high and low  (including this forum) but I cannot find anyone with a similar problem.  Background:
mining cryptonight with two rx cards, 480 (4) and 580 (
both out of the box but the 480 clocked a bit.
the 480 does about 600h/s but 580 does a measly 350 h/s.
I can change the "-h" value on the 580 to get to better numbers like 600 np (will fix bios and drivers next)

Problem:
I cannot for the love of god find where to change the "-h" value when running NiceHash miner legacy v1.8.1.5
doesnt matter If I put it in  config.txt or the .bat-file of claymore or the json files of NiceHash.
please.. help?

settings . device/algo then select card u need to modifyand algo then place -h xxx in the Extra  launch parameter box

it does take some testing and patience to find the right mix for your rigs for sure. I was very lucky to get most of my rigs built back when MSI 470 4GB gaming GPUs were readily available for under $200. Now is a much dicier proposition due to severe shortages of most models/types of GPUs.

One thing i will mention is that there are many seasoned miners that avoid the MSI Armor series at all costs. The consensus is that while almost all other MSI cards are great for mining, the Armor line does not match up, not even for the slightly cheaper price say vs their gaming model.  This is further proven out by the fact that those Armor cards always seem to be among the last in stock when the GPU supply chain has completely gone under.

Anyway, I also thing Armor come in 4GB max, and at this point, i would recommend not buying any more 4GB cards for mining. Much better to go at least 8GB, even if you are sticking with 570/580's. Much better for future proofing.

^^^ You can try daemon+tcp:// but I'm not sure if Clay supports solo.

Apparently some folks don't have the compute option turned on.


edit:

I found this for all you folks who don't use Ubuntu 16.04 64-bit.

-- https://overclock3d.net/news/gpu_displays/amd_adds_compute_mode_and_support_for_up_to_12_different_gpus_with_their_latest_gaming_driver/1

edit2:

I bet MS had to twist their arm to get them to do that...