have you tried other miner software? Claymore (with built-in overclocking)?
...disabled crossfire?
Claymore XMR-Cast XMR-Stak Crossfire disabled via registry keys and double checked in Crimson. |
|
|
|
Just to be sure... you have set your bios to UEFI, right?...
Yes, have tried both 4G enabled and disabled |
|
|
|
Fresh install of Fall creators update.
Download/unpack AMD drivers.
Shut machine down.
Install 3x Vega cards
Remove network cable
Boot machine and install drivers with no network connection.
Reboot machine
Go into crimson, enable HBC max.
Reboot
Connect ethernet
Begin mining XMR, 3,500h/s for 3x Vega 56x
Grrrrrrr
you don't need HBC turned on and set your pagefile as high as possible. On/off it does not make a difference. System memory is at 16GB and pagefile is at 70GB. What is your hashrate with 2 vegas? Maybe the third vega is causing the issue. Disabled HBC, restarted, running two miners, 2,000h/s "gpu_threads_conf" : [
{ "index" : 1, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 2, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 1, "intensity" : 1600, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 2, "intensity" : 1600, "worksize" : 8, "affine_to_cpu" : false },
], |
|
|
|
Fresh install of Fall creators update.
Download/unpack AMD drivers.
Shut machine down.
Install 3x Vega cards
Remove network cable
Boot machine and install drivers with no network connection.
Reboot machine
Go into crimson, enable HBC max.
Reboot
Connect ethernet
Begin mining XMR, 3,500h/s for 3x Vega 56x
Grrrrrrr
you don't need HBC turned on and set your pagefile as high as possible. On/off it does not make a difference. System memory is at 16GB and pagefile is at 70GB. |
|
|
|
|
Is support coming for the rapidly growing list of CryptoNight miners out there?
xmr-stak
xmrig
cast-xmr
These are quickly growing as the CryptoNight scene is exploding and Claymore's XMR miner is lagging hard in the competition.
I need support for these and will be upgrading my license once one/some of them are supported..
|
|
|
|
So I have my KORE Masternode up and running! ...I think. STATUS: Enabled ACTIVE: 00:00:00 LAST SEEN: 20+ minutes ago Should this be updating? Do I need to wait some time for that data to get updated? Or have I messed something up? Edit: nevermind! Check the next step and we're good! Thanks guys!  |
|
|
|
|
Fresh install of Fall creators update.
Download/unpack AMD drivers.
Shut machine down.
Install 3x Vega cards
Remove network cable
Boot machine and install drivers with no network connection.
Reboot machine
Go into crimson, enable HBC max.
Reboot
Connect ethernet
Begin mining XMR, 3,500h/s for 3x Vega 56x
Grrrrrrr
|
|
|
|
After running for 8+ hours at 920 my HBM readings are 72/63/65.
Before I went to bed I rebuilt the rig, removed all risers and plugged 3 GPUs directly into the x16 slots. DDU'd, fresh install of Blockchain drivers. Made sure HBC was on. Overclocked the memory to 920. I got 3,500h/s stable all night long from 3 Vegas with the above config.
That's not very good is it...  Have you try the disabling/re-enabling the GPUs as mentioned further above? Did it make any difference? Yes, used devcon to reset them then am forced to reboot. My machine will always claim that a card is not present until I have rebooted. |
|
|
|
I'm starting to this something is bottle necking... so mining with this: "gpu_threads_conf" : [
{ "index" : 0, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 1, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 2, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 3, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 0, "intensity" : 1000, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 1, "intensity" : 1000, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 2, "intensity" : 1000, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 3, "intensity" : 1000, "worksize" : 8, "affine_to_cpu" : false } My machine mines at 5,000 h/s, with the 1920 threads running faster than the 1000s. Running with this:"gpu_threads_conf" : [
{ "index" : 0, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 1, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 2, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 3, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 0, "intensity" : 1500, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 1, "intensity" : 1500, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 2, "intensity" : 1500, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 3, "intensity" : 1500, "worksize" : 8, "affine_to_cpu" : false } ...produces 5,000 h/s, with the 1920 threads running slower and the 1500 threads making up the difference.. What are your HBM temperature readings in hwinfo64? After running for 8+ hours at 920 my HBM readings are 72/63/65. Before I went to bed I rebuilt the rig, removed all risers and plugged 3 GPUs directly into the x16 slots. DDU'd, fresh install of Blockchain drivers. Made sure HBC was on. Overclocked the memory to 920. I got 3,500h/s stable all night long from 3 Vegas with the above config. |
|
|
|
I'm starting to this something is bottle necking... so mining with this: "gpu_threads_conf" : [
{ "index" : 0, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 1, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 2, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 3, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 0, "intensity" : 1000, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 1, "intensity" : 1000, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 2, "intensity" : 1000, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 3, "intensity" : 1000, "worksize" : 8, "affine_to_cpu" : false } My machine mines at 5,000 h/s, with the 1920 threads running faster than the 1000s. Running with this:"gpu_threads_conf" : [
{ "index" : 0, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 1, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 2, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 3, "intensity" : 1920, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 0, "intensity" : 1500, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 1, "intensity" : 1500, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 2, "intensity" : 1500, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 3, "intensity" : 1500, "worksize" : 8, "affine_to_cpu" : false } ...produces 5,000 h/s, with the 1920 threads running slower and the 1500 threads making up the difference.. |
|
|
|
Using those settings above I was able to max out (2) Vegas for a couple hours. Added the third in with the same settings and now I'm getting about the same hashrate as before, 2 cards are running at about half speed. I'm ready to return these at this point..
When you restart your computer, always restart the cards, too. See my post in the Cast thread: https://bitcointalk.org/index.php?topic=2256917.msg24518280#msg24518280Also make sure you're explicitly running the August blockchain drivers. Anything newer does not play well with Vega. Yep been using that. Usually one card will refuse to work after being cycled with that though. Yes, using Aug 23rd blockchain drivers. |
|
|
|
Got Cast XMR "working"..  |
|
|
|
{ "index" : 0, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 0, "intensity" : 1700, "worksize" : 8, "affine_to_cpu" : false },
Are these Vega 56 or 64? When I was using Stak on my 56s, I had my second thread worksize at 1600. I might suggest testing out CastXMR, even if just long enough to rule out a configuration issue versus a system issue. Hmm, I have 5 vega 56s running no problem on a 64GB SSD. Virt size is 32GB, and I have 16GB of physical. Getting 1800+ on ea Interesting. What's your physical RAM usage when you have a workload? You prompted me to look at my own system, and I am noticing that with 4 GPUs, my VirtMem commit is only 32GB... maybe we only need 8GB Virtual per card? Edit: Actually I'm even more confused. Maybe I'm doing something wrong, but following this guide to check my pagefile usage reveals 0 actual use. I wonder how we got here to thinking that we needed a high Virtual Memory counter for things to work properly. I tried cast but it refuses to acknowledge that I have the correct blockchain driver installed (even though I do). Using those settings above I was able to max out (2) Vegas for a couple hours. Added the third in with the same settings and now I'm getting about the same hashrate as before, 2 cards are running at about half speed. I'm ready to return these at this point.. |
|
|
|
Holy cow 16 per VEGA? I think I set like 40 at the moment for 2 but I'll check. Now I'm on trying to get the second one to work..
On a dedicated mining rig, that's easily done with a 128GB M.2 chip, like my favorite <$60 ADATA. It's only 118GB usable, but after Windows itself, and setting my VirtMem to 80GB (adding a 5th GPU this week), I still have 7 GB left. You'd have to be a little creative (stripping out Windows components) to get enough for 6, I suppose, but I'm sure it could be done. Of course, the 256GB is only $90 (and traditional SSDs are $80, but I really love the compact 'package' by using M.2). Hasn't fixed my problem sadly. Have the Virtual Memory set to 48GB with 3 Vegas. Set my threads to: { "index" : 0, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 0, "intensity" : 1700, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 1, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 1, "intensity" : 1700, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 2, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false }, { "index" : 2, "intensity" : 1700, "worksize" : 8, "affine_to_cpu" : false } And am getting "cl out of host memory" errors. |
|
|
|
I'm about to tear my hair out. I have been fighting with these damn VEGAs all weekend!
After nearly a dozen hours I have a single Vega card working in my board.
Fresh install of Fall Creative Update
Aug 23rd Blockchain Drivers
XMR CAST 0.6
HBM: 990MHz
Mining XMR at 1,400h/s
Does anyone have any ideas why I'm so low?
In device manager disable and enable cards one by one. Tried that repeatedly.. Maybe it was just coincidence, but I had a 1300 hash rate at one point until I upped virtual memory to 32GB.
This is possible. Current community understanding is that you should have 16GB Virtual Memory for each Vega. I made the mistake of getting 32GB of physical RAM, and I'm only using 10% (more if I have other programs running like HWinfo64) - never more than 8GB so far. So I just ran out and grabbing 16GB of physical RAM (up from 4GB) and now I'm getting 1 running steady at 2000+. Holy cow 16 per VEGA? I think I set like 40 at the moment for 2 but I'll check. Now I'm on trying to get the second one to work.. |
|
|
|
|
I'm about to tear my hair out. I have been fighting with these damn VEGAs all weekend!
After nearly a dozen hours I have a single Vega card working in my board.
Fresh install of Fall Creative Update
Aug 23rd Blockchain Drivers
XMR CAST 0.6
HBM: 990MHz
Mining XMR at 1,400h/s
Does anyone have any ideas why I'm so low?
|
|
|
|
Oh man I'm in hell. So Amazon had Vega 56's in stock with 1 day shipping and I had a bunch of 0% interest free on my store card so I snagged 4.
I have been trying to set this up ALL DAY LONG. Windows utterly refuses to see all 4 cards. Linux dies with the newest drivers and just generally does not seem to run with Vega even though I've found plenty of reports saying it does.
I'm using an MSI M5 which I have confirmed supports 8 RX 570s.. but Vega so far has been a nightmare.
From my experience: with windows vegas are finicky and sometimes you'll have to uninstall the blockchain driver with DDU, then install it again for the card to become 'workable'. On Linux 16.04, you need newest amdgpu-pro driver and sometimes some extra work, although i am not sure that vega's would work-did not try (580s work fine). Finicky isn't the word I would use. They're an abomination. The AMD drivers are so unbelievably buggy and unstable that I cannot believe this is the case months after release. Simply throwing a single VEGA into a normal computer causes tons of issues! |
|
|
|
|
Oh man I'm in hell. So Amazon had Vega 56's in stock with 1 day shipping and I had a bunch of 0% interest free on my store card so I snagged 4.
I have been trying to set this up ALL DAY LONG. Windows utterly refuses to see all 4 cards. Linux dies with the newest drivers and just generally does not seem to run with Vega even though I've found plenty of reports saying it does.
I'm using an MSI M5 which I have confirmed supports 8 RX 570s.. but Vega so far has been a nightmare.
|
|
|
|
Bah cryptonight is so worth right now I was about to go to my local Microcenter and grab a couple Vega's and try them out but of course they have none in stock.  |
|
|
|
The Bittrex wallet appears to be stuck Next release will be soon, this will (hopefully) fix bittrex's downtime issue, and everyone else's. Please be patient. Matrix No worries, just letting ya'll know. Good to hear. |
|
|
|
|
Show Posts
