Topic: Any word on amd vega hash rates? - page 61. (Read 202725 times)

Hi Guys,

I have an issue with 4 x 56 cards flashed with 64 bios.

I have the HBCC settings only on 1 cards, so when i mine with 1 cards it's working, when i try to 4 cards nothing happend.

When i try to disable crossifre the system crash and reboot.

Any help regarding this issue ?

Thx

I am in trouble with the 3 fresh Saphires. 2x 56 (bios modded to 64) and a 64. The rig (these 3 sapth + 2x64 and 1 56 all powercolor) , like 10-15-20 mins, and then suddenly the miner stops.
I've tried several different sets of settings and mods - same result. They are all connected to one 1000W power supply, the 3 power colors that I had for more than a week have zero problems (running at 1407/870 and 1100/850), they are attached to the other power supply.

It seems like the problem is not the memory - with my other rig, the memory crashes were happening way faster. I guess its the voltage, but how exactly should I identify the card that is causing the crash?
After the xmr stak hangs, windows crashes when I close it. And every time, the last card to remain full red before the restart is the one on the PCE-E 16 slot.

Also, at the times where I get real time read on the restart, when pressing H on xmr stak, the first thread of card 0-1 has N/A as value, for a few seconds. But how can that be - it has been working great with even lower voltages and higher memory than the values I am using atm?

As I am already fighting this for straight 16h, I made it stable at 1100 core/875mV and 1025 mem/875mV. Hashing measy 9600 H/s for the 6 cards ... but at least its working since almost 90 mins.



My basic plan for tomorrow:

1. Reinstall drivers, and put the 3 power color card that have been flawless first. Mod em, make sure they work for 2-3 hours .
2. Add a single saphire, mod it , try differnent voltage settings until it works stable.
3. and so on ...

But is it possible, that adding the saphire causes issues with the powercolors? As the whole mess just doesnt make any sense Any ideas how to continue the fight?

Yes, me too.

If I try any of the 3 posted reg files on my XFX Vega56@64 Bios i get a green Screen at Windows 10 Start and cant get into Windows. Any Ideas why?

use the Q key to shutdown miner so it doesnt crash.

Hi guys. On what model the motherboard can work 8 Vega?
Interested in successful work experience 8 video cards.
Google translate.

Can you clarify something:  are you applying  a SoftPowerPlay registry modification, or not?  I ask because like you all my Vega 56s have been flashed with the 64 bios.

If you are using a SoftPowerPlay mod, you can leave it at a stock, safe setting that at least safely reboots.  Then, after rebooting, you can use OverDriveNTool to load whatever custom saved profile you wish.  Only takes a few seconds to hit "load" and then "apply".  No need to retype all the values after each reboot.

4x vega 56 w/ 64 bios and 800mem bios

manual set 1350/1100
i also tried 1350/1050

I have an annoying issue where the cast-xmr will crash. upon reboot the drivers are back to stock voltage etc in overclock.

pretty much have to DDU.

Anyone had this issue before and figure out the cause?

i checked into the logs, it's asif the drivers are getting corrupted..

i ran my miner with hellae's guide on my 56's with no bios flash last night.

i woke up and they were running but when i went to power on the monitor it said no signal? i think it's some weird windows setting i need to manage.

hrm i had that problem and someone on reddit told me to disable and then enable each gpu. magically, i got 1800 h/s.

I have been testing this also out by my self but cant find detailed documentation what exactly was changed in the blockchain driver.

Possibly would want use the bare minimum from it.

I tried to use cast-xmr but it did not work to connect to the nicehash pool.

[/quote]

all my rigs eth with 580's and vega rigs use same mobo. bio tb250btc g4400 16gb ram (580's 8gb) , never mined at ninehash... mining at web.xmrpool with no probs.

The Onda D1800 is six-slot.
Really hoping it'll work. Otherwise it's gonna be back to the drawing board...

I just had an idea...
Remember the trick we used to do to have the RX5xx working with the older drivers? I.e. Physically copying the content of the older driver's folder onto the new one, effectively tricking Windows into thinking it was using the new one?
Maybe something like this could be attempted with the Vega blockchain drivers too... I mean, if you have driver instability (devices disappearing etc) it could be worth a try...

gigabyte z170-ud5 skylake using bios settings here https://www.gigabyte.com/microsite/462/mining.html
64gb ram 3300Mhz
1300w gold psu
monitor plugged into onboard gpu
vega fe in x16 slot using pcie2 setting (ie not in riser)

I had this issue with xmr-stak-amd in the past, screen flickers and one of my cards goes to defaults settings.
But now all is smooth both with cast-xmr or stak.
It can be non optimal clock/voltage settings or some other hardware utilities that cause memory paging / I2C errors with such postponed effect

Make sure you are closing miner with 'q' key, not just closing miner window

Is it just my rig or everyone else has their AMD setting crash every time we close the Cast XMR?

Thanks but I have already tried the devcon tool and it did the same thing.
It claimed to have enabled it but the device manager still shows it as disabled with a yellow exclamation icon.

Only the non-blockchain drivers allow me to disable and re-enable the card.

In my experience yes the reset is required on my MSI X370 Ryzen which does not have an internal gpu.

If you follow the instructions he gave you with devcon.exe and a batfile it might work.