Topic: Any word on amd vega hash rates? - page 58. (Read 202748 times)

Hey guys, doing my first Vega setup here, I've had a lot of experience with Nvidia rigs but first time doing an AMD one and as expected it's been going rough so far..

So the config is the following:
Gigabyte GA-Z270P-D3
1x EVGA 1000GQ + 1x EVGA 850GQ
Pentium G4400
4 GB DDR4
128 GB SSD
6x Vega 56

I had no issues with the Mobo, recognizing all 6 cards right off the bat. I installed whatever is the current version of Windows 10 Professional (1709) from the official media creation tool, set VM to 80GB and followed this guide:
https://vegamining.blogspot.com/

Now, I've been facing a bunch of different problems - first, when I run my script to enable/disable GPUs - either through the .bat file, through Powershell and I even tried through Python - i get a BSOD most of the times and have to restart. In the few times that this part worked, I then apply the Overdrive settings and start XMR-STAK which initializes properly but then hangs almost immediately.

I did a few more attempts, flashing the Vega 64 BIOS and following this guide but ran into similar issues:
https://www.reddit.com/r/MoneroMining/comments/74hjqn/monero_and_vega_the_definitive_guide/

The only success I had was doing something between the two guides - keeping stock BIOS, applying the Wattman settings from the above guide for Stock BIOS and then applying overdrive settings on top and running XMR-STAK. I know it makes no sense but that was the only thing that worked and was running stable for 8 hours but with terrible power usage - 225+W per card. That's why I decided to start from stcratch and try the Vegamining guide again but ran into the same issues as before. So essentially I've been goin in circles and am highly frustrated..

Any help will be greatly appreciated! As well as any advice on which guide do you think I should follow with my rig configuration.

1. add virtual memory, 16gb per card
2. Make sure you use 23aug blockchain driver
2. re-enable both cards in device manager before mining
3. use cast-xmr miner
4. make sure your temps are below 70 degrees (both gpu and hbcc)

I'm pretty sure that there is no way for anything to monitor the hashrate and force run of this reset script.   You might be able to setup alerting when the api shows rates get under a certain level, but I don't see how you could automated it without initiating the request manually.     as for the script not loading your payment address it's got to do with the config file it's calling and not finding it.

*EDIT*

I was wrong about monitoring the hashdrop..  check this guide..  https://vegamining.blogspot.com/

I have an issue with my setup.  Can someone point out what I'm doing wrong (my first setup).

I'm getting very low speeds on my 2x vega 56s regardless how I set it up.
I've tried the following with the results:

- index 0 only: 1 thread per GPU 2016 results in 600H/s
- index 0 only: 2 threads per GPU 2016+1600 results in 236H/s
- index 0+1: 1 thread per GPU 2016 results in 1175H/s
- index 0+1: 2 threads per GPU 2016+1600 results in weird rates ID 0 = 132, 1 = na, 2 = 221, 3 = 853 and 1 and and I get CL_MEM_OBJECT_ALLOCATION_FAILURE

"platform_index" is set to 1 (doesn't work when set to 0)
Sometimes xmr-stak-amd freezes when setting up 2nd "Device 0 work size 8 / 256" or sometimes on Device 1

My setup:
ASRock H81 BTC R2.0 mobo with G3260 CPU and 4GB of RAM running Win 10 64bit
Both cards are connected via risers (PCI-E #2 (long) and #3 (short) are in use out of 6 total) and powered by 1000w EVGA gold
I am connected to the onboard HDMI for monitoring
There are no HDMI dummy dongles in the Vegas
Running blockchain driver v17.7

Anything obvious pops up?  I'll try disabling onboard video and connecting the monitor to one of the Vegas, but can you see any other issues with this?  Pointers perhaps?  

All of this is running on default 1580MHz/800MHz speeds (haven't tweeked any Wattman settings and HBCC is off)

Thanks a lot in advance.

Hi guys,
Someone tried drivers Crimson ReLive Edition 17.11.2 ?
What about eth hashrate? And what about cryptonight hashrate?
Anyway, is there someone with more than 9x vegas working on the asrock h110 btc+ pro mobo?

Thx

It runs about 100W on idle but I made no effort to reduce running power so I reckon about 250w

I work away from home during the week doing well paid IT work so I only get the opportunity to tweak at the weekend and if it falls over during the week I'm not there to fix it and I can lose up to 5 days mining, hence stability wins, but its just a hobby for me

*winces* .. Dunno what Mobo/CPU you have, but considering that mining is not CPU intensive, I can't imagine that package is any more than 50W. That would mean ~300W for a single card...?

I can get FEs for an okay price (compared to current retail), but it's only 10% more hashrate for almost twice the power draw of a tuned 56. I'd be extremely interested in the results of any tweaking you end up doing.

Yes, that's the code and it works.  However, this does nothing for the miner itself correct?  So, I still have to be physically present to run this batch and restart castxmr when the hashrate drops.  It needs to restart the miner as well for it to work.  

Excuse me for being a newbie, I am pretty tech and hardware literate, but when it comes to creating batch files and software coding I've had practically no exposure beyond some basic HTML for websites.

I found this:   https://pastebin.com/Q0rGa0QS  

I think that's a complete batch file that I'm looking for, but I'm still trying to figure out how to properly edit it to function completely automatically with castxmr.   It's set up to run with xmrstak.  

I got it to work and it starts castxmr, but it doesn't load my payment address when it starts and it also doesn't restart on it's own when the hash falls.

including mobo, wifi etc about 350w, I haven't tried to lower it. I have 3 Vega fe's but I have the same problems as everyone else with getting the drivers and config stable. Wasted lots of time etc so decided to let one card run for a while to earn its keep

outside temp is quite cool so I leave the window open and it runs quiet at about 63/64C

this will disable and re enable all vegas only

Wow. What is the power draw on that; do you know offhand?

does 2100+ out of the box no tweaks, peaked at 2600, had mine running 2100+ non stop for week 3 now

Ok, so I downloaded and made my Devcon folder with the text batch file provided online.  It does indeed disable and re-enable my cards.  Now this batch file will do this automatically every time the hash rate drops correct?

Another issue before even calling that a success though... once I run the devcon file it resets everything, but then in cast-xmr the vega's don't get renabled.  They just show they are getting 0/hs.  Do I need another batch file or something to have cast-xmr restart or re-detect the cards?

You will need 8gb of ram to activate hbcc using Wattman. I've heard it can be done by disabling each card in device manager then activating again but I haven't tried thisywelf as I had bought an extra 4gb before that fix was found. Still not attempted the reg mod for the power consumption yet. Currently running 4x Vega 64s at 900W 1950h/s per card

I dont have a rig yet ( I may be building one soon) but from my research, try these things
make sure in the power plan settings in windows you have the sleep and the monitor set to never turn off
make sure your power plan is high performance

this one is probably the problem though
do the devcon.exe trick to re-enable the two vega cards

I remember reading a post similar to yours and they were just forgetting to re-enable the cards

I was wondering do the vega frontiers perform the same as the vega 56 and 64 (can they get 1800-1900 h/s using the tweaks and the blockchain driver) ?

also If I were to buy the vega 56 and not change the bios but install the blockchain drivers could I get 1800 h/s?

Thank You!

Has anyone found out what exactly is triggering the driver issue that causes the hashrate drop?

I've tried everything I can to get the hashrate to stay up that was suggested.  I am running cast XMR and have 4 cards on my system: GTX 1070, RX570, and 2 Vega 56's flashed to 64. 

The nvidia card is hooked up the monitor.  I get 1800/1950 on both of my cards when I start it up just fine, but like many others lose the speed shortly after. 

I've got my power settings set to never turn off (even the monitor is not turning off at all), I've got everything that resembles crossfire or ULPS in registry disabled, my pagefile is set to 60gb. 

The only thing I don't have is a fake dongle, but the AMD cards aren't even hooked up to the monitor if that matters.

I've got my fans at 3000rpm, but my temps always seem fine.  I am usually under 60c on the core temp.  I haven't checked recently, but I believe that puts the HBM2 ~10C higher. 

I really wish I knew exactly what I was fighting, but I just seem to be trying a bunch of random things to fight this issue without any real idea of what I'm trying to correct.

Is that with the 56's running the 64 bios?   It wasn't clear to me if that was factored in at all. 

using cast xmr sometime now, so far so freakin good ! got free electricity so i follows gandalph guide for 2050h/s (running 56's) at -10% 990 ram and +13% power. anyone managed to find a sweet spot for bigger gains but higher power draw? experimented many many settings but with no luck... best of is 10mins at 2110h/s but it crashed or pixels screen

how low can i go with the voltage in afterburner ?

i m currently at -112mv
did not test lower yet

look like average vddc is 970mv  so i guess can go lower