I'm using Simplemining OS as well and have issues with the miner constantly freezing and reboot. The other issue I get is the miner will restart but the initializing will fail and the miner will be stuck at mining 0H/s.

I've never played with the -h value. On page 1 it says -h deals with # of hashes that the GPU processes so it probably has to do with how intense you run the cards. I'm going to play with this as well and see how it changes. Sucks that you had to downgrade the cards to 1200/1900, I would have hoped that you can at least run at 2000 mem speed. Have you tried to play with the voltage levels?

What do you have your -allpools set to? Right now my settings are this:

-wd 1 -r 0 -xpool stratum+ssl://cryptonight.usa.nicehash.com:33355 -xwal [yourwallet].$rigName -xpsw x -allpools 0

I'm running RX570 4G at 1150/1750/1000mV and RX580 8G at 1200/2050/1000mV.

I had that behaviour when moved to claymore v10.2 but, to be honnest I had a superior speed with it.  Like you, sometimes the miner restart and stay freezed at 0H/s as well. The only way of getting out of this state was rebooting the whole rig. And by rebooting, i mean shutting it off, wait 2-3s in power-off and then rebooting. Lowering the "-h" parameter bring more stability to the system. So, my current onfiguration is:

Cards: RX580-4G,RX580-4G,RX580-4G,RX570-8G,RX570-8G,RX580-4G,RX580-4G,RX580-4G,RX580-4G,RX480-4G

Claymore command line to XMR mining:
-xpool stratum+ssl://xmr-eu1.nanopool.org:14433 -xwal XMR_ADDRESS/$rigName/your@email.com -xpsw x -mport -3333 -h 850,850,850,1700,1700,850,850,850,850,850

As you can see, you can pass to the miner a comma separated values in that each one is to a specific card from my list above. For reference, the default "-h" value  for the 4Gigs cards is "900" and "1800" for the 8Gigs.

My current frequencies for each card are:
CORE: 1200,1200,1200,1200,1200,1200,1200,1200,1200,1200
MEM: 2000,2000,2000,1900,1900,2000,2000,2000,2000,2000

XMR - Total Speed: 6948 H/s, Total Shares: 1340, Rejected: 0, Time: 06:43                                                       |
XMR: GPU0 715 H/s, GPU1 715 H/s, GPU2 715 H/s, GPU3 686 H/s, GPU4 649 H/s, GPU5 712 H/s, GPU6 715 H/s, GPU7 713 H/s, GPU8 713 H/|
s, GPU9 615 H/s

Hope this help

Can anyone confirm what's the "-h" is doing behind the scenes? and why lowering it a bit bring more stability?

regarding quoted, read post #1, it explains it quite nicely.

Automatically r9 fury 4gb hbm is defined by the miner with the following parameters -h1800 -dmem 1 which is 8gb of gpu memory. You have to start only with your custom settings and this is very annoying especially when using miner nicehash for windows.

just add -h 900 (or whatever) to your bat file string.

40GB is an overkill. I've got it working just fine (on dual ETH) with 30GB (28 works too, but 24 doesn't:) with 12GPUs on 8GB RAM system

^^^ Do you see that message often? Have you tired lowering your -h value?

That I don't see anymore. H/s 450-480 looks low

so do i, my 580 x5 need almost 40GB pagefile (38,xxxxx)

i think i am not alone getting "WATCHDOG: GPU x hangs in OpenCL call, exit" message. Restart in every few hours or sometimes every minutes.. but mostly happened in gpu 0/1 (with long type pci-ex slot/pci-ex x16), maybe i'll try to only use x1 type pci-ex..