Thanks. That bumped my R9 nanos to 108-110 sols/s. Now better than claymore~!

Is someone getting the 115 sol/s on the R9 nano?
Only getting 96-100sol/s on mine using the Windows build.
Just wondering if that's possible with the linux build?
FYI, running the nanos on stock bios

Hi Claymore,

Is it possible to enable SIA+DCR mining? ETH mining will be dead anyways soon.
I'm okay to pay my devfee with ETH, but SIA+DCR at the same time would be great!

Hi, that's new to me! How can I calculate the mem clock from the core?
I am currently using 850 core clock with the old rom. What's the recommended mem clock? And how does it change with the new rom?

And thanks for sharing the new rom! I hope I'll find some time to test this weekend!
Cheers,
Talan

Hi Eliovp,
How's the status on the new rom? Just wondering...

Hi Marvell,

I had exactly the same problem. Reason is that some of your nanos cannot cope with -175mv while some can.

I had 6 nanos shipped from Amazon and I made the mistake that I applied the same rom to ALL of them. However, each card is different. I even was very unlucky, and had one card that even cannot cope with a -100mv mem voltage! Hence my rig always crashed when increasing the core speed. Unfortunately I also used this gpu when I only tested with one nano...

Therefore my recommendation:
- Test each nano separately and see what mem voltage it does not crash with when you increase the core speed and start the mining

I had the following results:
GPU 1: -100core/-75mem: -100mem lets the system crash. Currently I manually reduce the core clock to -100 (there is only one -75/-75 rom here)
GPU 2: -175 mem
GPU 3: -175 mem
GPU 4: -150 mem
GPU 5: -70core/ -150 mem: Claymore's miner shows a lot of invalid shares for this gpu. I increased the core clock and now it's better though there are still some invalid shares from time to time
GPU 6: Dead - getting replacement

Some other issues: when running headless, max 640x480 resolution when using vnc. Had to purchase a dummy plug. Now it works.

I'm thinking of reducing further the core clocks (except GPU5), but first I want to run this rig for some days to see its stability.

Hope this helps! And Helios, great job! Looking forward to your other roms.

Cheers,
Talan

Hi Eliovp, i got this one working now at 1020 core.
I can reduce the core volt by another 24mv in afterburner

I still run some tests, but this seems okay. Any chance you can do a -100 core/-75mhz for me?

Also in case of not too much work, would be great to have 6mv steps from -100 down to -75. in this case i could try to find the sweet spot.

Thanks so much,
Talan

Thanks for your answer, Eliovp!
I tested with only one Nano connected. Directly plugged into the Motherboard -before that I used powered riser. I have no molex plugged into the motherboard (which should be fine as I have only one card connected and if more, only via powered riser).

As a power supply, I use the ESVGA 1300 Watt. Completely new with one dedicated rail for each gpu. Otherwise there's only the HDD and mainboard that uses power. That definitely should be sufficient.
The Motherboard is H81 btc. I updated to the latest firmware...

What AMD driver version are you running?
I will run some tests tomorrow, but there are not many options left.

Cheers,
Talan

Hi Eliovp,

I tried your ROM but it shows the same effect. I was able to change the core to 900mhz but once I started claymore's miner, the whole system crashed.

I really wonder why no other experiences this issue. What's your program to display your remote desktop? I use ultraVNC. I somehow suspect that the system still uses the GPU to render the remote desktop and causes the issue. Again, once I use internal graphics, no core/mem clock is shown in afterburner and I can't change the core to 1000.

Before I flashed the 75/75 rom, I just let it mine with the -100mv mem@800mhz bios and it worked @ 22mhz. Of course, I want to reach the 30mhz...
Do you think it can be due to asics quality?....

Hi all,

Thanks for the feedback. I ran some further tests today but still no solution found.

That's what I did:
- Tested Gen1 for Pcie: No change. I used to select Gen2 so far.
- Disabled onboard GPU: No change, except that I used the nano for the output

- Flashed -175mv rom. Reinstalled drivers after ddu. Windows reports "Your driver has crashed" in the taskbar. System is very laggy.
- Added a Radeon 390 used for the output which is not underclocked. 2 Nanos plugged in. No monitor attached to them. Windows runs stable now. Ran afterburner, changed to 1000mhz core for the nanos. Ran claymore miner. PC immediately freezed. Had to hard-reset.
- Flashed -100mv rom. Did the same above. Same result.
- Tried a Saphire Nano. Flashed -100mv. Same result.

Here's my system config (all NEW items which I just bought).
- AsRock BTC Pro H81 with 8GB RAM and CPU
- SSD
- eVGA 1300 Watt PSU. At max I ran 3 GPUs. PSU should not be an issue.
- Powered Risers
- XFX R9 Nanos
- Virgin Windows 10 installation.

Here's the register dump from AIDA64: http://speedy.sh/hNsKt/atigpureg.txt

Any further ideas? It's really starting to piss me off as I put so much time into this issue and still no progress.

But thanks so much for your ideas so far!
Talan

Hi lawrencelyl and Eliovp,

I did many tests today- but still the same issue. First of all, you are right, lawrencelyl, this situation is even there with the stock bios. So it's not a mod issue.

I've been running tests today by running it in win7,win8 and win10. all clean installs. driver left and right. I now doubt it's a driver issue.
When core clock increase, the system gets very unstable. As mentioned, nano crashes when I increase the core clock from 800 to 900. I even tried to increase the core clock by 50mV.

@lawrencelyl : Can you run the bios mod with monitor plugged in? What are your settings in afterburner and which mem clock mod are you running? Are you on XFX?

At the beginning I thought it's due to the monitor. Or could it be another issue?
Eliovp, would a snapshot of my stock bios somehow help you to see if we now have a fundamental difference? Let me know.
I'll run some more tests next week, but not many ideas left what to try...

Cheers,
Talan

Hi, Thanks much for your answer.
I completely uninstalled everything. Even with DDU.

However, the following still happens: When no monitor plugged in on the nano, then no data in gpu-z/afterburner is shown. Once plugged in, core/mem clock + temp immediately show up in gpu-z/afterburner.
Of course, with monitor the overclocking capabilities of the core from 800 to above 900 fails and the gpu crashes.
What am I missing here?
I am using the AsRock H81 Pro BTC as mainboard. Any wrong BIOS settings? I also put the gpu directly on the MB without riser. Same result.
Btw someone else had the same issue in this thread. But no result was posted.

Let me know if you have some more ideas.
Thank you!

Hi,

I just tried the -100mv Mem bios on my new XFX Radeon Nano, and strangely I do not get any GPU readings anymore. That includes core clock, mem clock but also temperature.
MSI Afterburner also shows completely nothing (see screenshots).

Did I miss anything? I'm on Win10 with 15.12 driver.
http://gpuz.techpowerup.com/16/05/28/9kt.png
http://gpuz.techpowerup.com/16/05/28/w6n.png

Any ideas?

Additional question: Is it normal that the driver is R9 Fury Series? There's no nano?

Hi Claymore,

Thanks for taking this up. How about you keep it simple by adding one commendline option to send stale shares and keep the status quo as default?
This way, there won't be a need to check all pools as anyways this could change over time. I heard that miningpoolhub and ethpool/ethermine accept stale shares as they could result in uncles.

Thanks

Hi Claymore,
I'm also wondering how your miner handles such cases what genoil has fixed here: After having received a new job, will your miner still submit stale shares for the replaced job?

Hi Claymore,

Setting core or mem clock doesn't work for my Asus 7990. 380+390 both work though. The miner crashes in case i try to set the clocks....

Also can you please add setting the voltage? I want to move completely away from msi afterburner and i had some cases where after i rebooted the voltage was back to standard and the card overheated to 100 degree.

Which leads to another idea for your consideration: reduce intensity or stop mining on gpu when it exceeds a heat threshold. Recently one of my fans broke i only noticed after the whole rig had crashed....

Anyways, keep up the good work!