Regarding this question, I use OverclockNTool and the CLI to automate settings update prior to running the miner.

IIRC: OverdriveNTool.exe -p1"Setting1"

You're the man. Updated and rock solid. I noticed that on a few of my bios versions (LC for example), the location of the change is different. An easy search and replace (on the ones closest to the location you describe) was all it took. 2050 h/s

Yes it seems like it's flying under AMD's radar... along with many other issues LOL

Has anybody found success on Linux with the "pro" AMD drivers, specifically for Cryptonight?

Could you share what hex values you changed to bump the SoC limit? And any other changes you made ontop of the orig ref file would be awesome! Prefer to manually update my currently working regs

Gave it a go tonight, got everything up and running (1709 + 17.11.2) correctly. Found the following issues:

1. Apparently 17.11.2 down to 17.10.2 does NOT support Vega Compute Workload... no toggle and confirmed that it's not just "automatic" b/c my CryptoNight hashrates are down to ~1300 from high 1900s
2. Also 17.11.2 doesn't support the SoftPowerPlayTable mod anymore

Can anybody confirm these findings?

Hmm, tips  I'm a treasure trove man if you want to PM me I'd be happy to give you some guidance on any particular problems you're having. That said, I've also read many people stating that the Vega series of AMD cards aren't for the "first timers" and I have to believe them. Looking back it's amazing the amount of problems I had getting everything situated... most of it directly attributable to the AMD drivers and working to make everything run as efficiently as possible.

I think the #1 recommendation would be: Keep it simple.

#1a Try to NEVER use WattMan or Afterburner. Stick to OverdriveNTool and registry mods. You'll never have to touch Wattman, and you can automate your undervolt/overclock via the CLI (OverdriveNTool). Actually, I feel like everybody needs to spend at least a week fuxing w/Wattman. Then you'll appreciate ONT and automation a whole hell lot more.

#2 Always make sure your system boots with NO DRIVERS installed. Eg, install all your hardware, make sure they initialize (for the Vegas a single red LED will glow on successful bios init, if not then something is wrong with hardware or bios config), and then when Windows kicks on, make sure Windows recognizes them all as the generic display adapters (Device Manager). Once you get this, THEN install your drivers. And install your drivers with all your hardware connected and enabled.

#3 Everything is a variable. Limit the variables and change things (overclocking/voltage/intensities/etc) incrementally so when shit hits the fan, you know what happened and you can intelligently back pedal. And take notes. I have 24 Vegas. 4 of them have been RMA'd. 1 DOA. 3 just were crap. Tracking those down and making special 'exceptions' for them is a PITA but possible. Just keep that in mind when things *should* be working but aren't

Hope it helps!

Thank you. I dread updating to 1709 (again) after I blindly installed and it fubar'd everything LOL. But that was the one variable I didn't try given that the new 17.11.2s are for 1709 technically. I'll give it a go and report back. Are you having success with this setup in particular?

Long time lurker, first time posting... Wealth of knowledge, THANK YOU to everybody here 

I'm curious if anybody has had any success running:

Windows 10 (v1703)
More than 8 AMD Vegas

I have 3 rigs of 8 Vegas running on the Aug23 Blockchains no problem, rock solid on ETH-only (~44Mh/s), ETH-dual (depends on secondary) and XMR (1960 Mh/s) to the maximum extent. With all units running below 160W for all 3 ops.

Until recently I was limited by my PSU to only 8 Vegas. Now looking to add a couple more GPUs to each rig now that the "masochism" is over.

So, my woe's:

a. With Aug23 Blockchain drivers installed, simply tried adding the 9th GPU. Upon restart, immediately ran into issues. Full DDU. Computer saw all 9 GPUs as generic (minor success!). Cleared out AMD folder, full reinstall of Aug23 Blockchain drivers. No dice. Same issues on restart.

b. Then tried the latest 17.11.2 drivers (with all 9 GPUs hooked up). Full DDU, install 17.11.2. Failed installation 2-3 times, no relevant error messages. Gave up on this round

c. Then tried 17.11.1 drivers (with all 9 GPUs hooked up). Full DDU, install 17.11.1, installation success but on multiple restarts, AMD Settings crashes immediately. Thought, that's fine, all I do in AMD Settings would be to change the driver to "Compute" anyhow. I can make a Regedit hack for that. So..

Thus I had 17.11.1 installed on Windows 10 v 1703, with all 9 GPUs recognized as Vega correctly --->> drivers successfully installed (but as described, I couldn't get AMD Settings to launch). So then I went to install the PowerPlayTable mod, and immediately upon restart, all the GPUs were no longer recognized (more accurately described: Driver Error).


Anybody finding success running 9+ GPUs on Windows10 (either 1703 or 1709)? If so, what drivers are you running and any special considerations?


Thanks guys, looking forward to being a contributing member here