thanks for sharing your experience! I have some problems with my 3rd rig. 6 x Vega on a Z170 Pro Gaming Board. in Windows is all ok, after installing my Blockchain driver -> Reboot. Everything ok. Use the reg keys for softplaytable -> cant boot into windows.
i will try to disable crossfire in front of using the softplay table reg
but i have only "Enable"crossFireAutoLink" in the reg keys, did you mean thad one?
Yes, that one. Set "EnableCrossFireAutoLink" to "0" after installing the drivers but before rebooting. You have to close out of the AMD installer and not reboot; go into RegEdit. But I don't think that is your problem. Try a different SoftPowerPlay table -- find one that let's you reboot. |
|
|
|
anyone see a grey screen and crash when running more than 2 gpu on cast?
on a single gpu it works fine, but when i add more grey artifcats then crash,tried two separate motherboards
it only does thiswith the aug23 blockchain drivers , the ronbin hood ones dont crash but 1500 hs vs 1950
I hit a similar problem, but it was the same with XMR-Stak or Cast. I have a MSI X370 and could not get past three GPUs without a blue screen/crash. Changed risers, cables, Vega cards, PCIe slots, settings, on and on. Finally solved it: Used DDU to strip everything out. Went directly to Microsoft's site and downloaded the fall creators update (takes a while). That loaded the non-blockchain drivers and for the first time I could mine with 5 Vegas on that motherboard (but slow speeds). Uninstalled the newest drivers, reinstalled blockchain drivers. It's all good now. hmm i tried this several times, I have a window image that alrrady contsins the fall creators update included and did this with a fresh install still the same deal, Ill try it again i guess , thanks It is possible that your image of the Creators Update has files that have since been updated by MS. I downloaded the Fall Creators Update live from the Microsoft website. You don't have to wait until Windows updater schedules the update for your system. YMMV. Just sharing what worked for me. |
|
|
|
Does anyone knows the reason of the different hashrates per card in xmr-stak (same settings obv. 1408/1100)?
I have 3 vega 56@64 rigs. Usually I get 1900-1950/card, but there is one which gets only 1750. 1750 is like a stock or a little bit lower hashrate.
I've tried to reinstall driver, tried different HBM clocks 1100,1050, 1000, different voltages and still the same. I've tried disable-enable many times.
Any ideas what to do?
Yes. Remove all monitors and HDMI dongles from your Vegas. Then click on a window, say the AMD Relive settings program window so that it is in the foreground and XMR Stack/Cast is not. Won't completely solve the problem but the speed on the one card will bump up closer to the other cards. Other people have mentioned doing your build with an Intel processor with an internal GPU, and then connecting any monitor to the motherboard, not directly to a Vega card. Or: put a HDMI dongle on Card 6 *and* plug a monitor into Card 1 or vice versa. Don't even ask me why, just try it. Total hash rate should be higher, but there will still be differences in the hashing values across your cards. |
|
|
|
anyone see a grey screen and crash when running more than 2 gpu on cast?
on a single gpu it works fine, but when i add more grey artifcats then crash,tried two separate motherboards
it only does thiswith the aug23 blockchain drivers , the ronbin hood ones dont crash but 1500 hs vs 1950
I hit a similar problem, but it was the same with XMR-Stak or Cast. I have a MSI X370 and could not get past three GPUs without a blue screen/crash. Changed risers, cables, Vega cards, PCIe slots, settings, on and on. Finally solved it: Used DDU to strip everything out. Went directly to Microsoft's site and downloaded the fall creators update (takes a while). That loaded the non-blockchain drivers and for the first time I could mine with 5 Vegas on that motherboard (but slow speeds). Uninstalled the newest drivers, reinstalled blockchain drivers. It's all good now. hmm i tried this several times, I have a window image that alrrady contsins the fall creators update included and did this with a fresh install still the same deal, Ill try it again i guess , thanks Have you tried newest amd in compute mode? With rx 4xx/5xx there is no need of blockchain driver anymore. I tried those drivers, had problems too **update** following Romis instructions i did a clean install, this time though i did not give the default driver time to install, I ran gpedit.msc and disabled the windows updates and driver updates in Immediatley rebooted the box. <snip> I think a lot of people experience crashes with multiple Vegas because they install the driver and immediately reboot. The most important thing I found was that when you install the block chain driver, CLOSE the AMD installer at the end of the installation but don't reboot. Instead, use Regedit to go into the same directory as the SoftPowerPlay mod, and change "EnableAutoCrossfire" to "0" decimal value for every card -- e.g., folder 000, folder 001, folder 002, folder 003, folder 004. Then reboot. The pathway is listed at the top of every SoftPowerPlay mod that you download. Turn off auto cross fire in the Registry values, and also in the AMD Relive settings application as well. Second interesting thing I found was that it's best not to get too aggressive with your basic SoftPowerPlay settings. Keep P7 in the 1401-1408/900-905 range; P3 in the 1050-1070/850-900 range (Monero settings). Too far outside those values and SoftPowerPlay causes a blue screen/crash *on startup*. Once you boot, however, you can use OverDriveNTool 0.2.1 to tweak the settings card by card and get stable performance. But you can't always write those optimal values directly into the SoftPowerPlay table and get a stable reboot. Just doesn't work, I don't know why, but that has been my experience. I see people setting 5 or 6 cards to 1100 memory across the board -- no wonder your rig is crashing. Not every air cooled Vega 56 can handle P3 at 1100. Your base settings should be dialed back. Then experiment card by card once your rig is booted. You can save some time by taking one SoftPowerPlay Registry mod and copying that as several files -- RegMod001, RegMod002, RegMod003, RegMod004, RegMod005 -- edit each of those files for one particular card, e.g., edit the "000," "001," "002," so you don't have to manually do this each and every time. |
|
|
|
If you have to set the blower fan to 3,000 rpm+ and at that speed they are annoyingly loud. You need to either consider water cooling, Raijintech air cooling, or locate in a place where loud doesn't matter. You can't run these in a room that any significant other might hang out in. |
|
|
|
anyone see a grey screen and crash when running more than 2 gpu on cast?
on a single gpu it works fine, but when i add more grey artifcats then crash,tried two separate motherboards
it only does thiswith the aug23 blockchain drivers , the ronbin hood ones dont crash but 1500 hs vs 1950
I hit a similar problem, but it was the same with XMR-Stak or Cast. I have a MSI X370 and could not get past three GPUs without a blue screen/crash. Changed risers, cables, Vega cards, PCIe slots, settings, on and on. Finally solved it: Used DDU to strip everything out. Went directly to Microsoft's site and downloaded the fall creators update (takes a while). That loaded the non-blockchain drivers and for the first time I could mine with 5 Vegas on that motherboard (but slow speeds). Uninstalled the newest drivers, reinstalled blockchain drivers. It's all good now. |
|
|
|
After a bit more than 24 hours of running, I don't really see the point of using this. While my xmr stak setup was running like a champ for weeks, with the same settings this thing is just not stable. Had to restart 4 times during that time - all after several hours of working, because suddenly one card freezes and starts hashing at 0 for a few seconds, and then the program freezes. That thing happened to 3 different cards. And these same cards were running at these settings for weeks with XMR Stak with no issues ... The output hashrate looks higher, but with the sudden drops of hashrate for every card in the rig (all at different times) the average hashrate looks only around 100 H/s higher (for the whole rig). That at the time when the miner is running, the ~1 hour I lost to downtimes + the payment system of nanopool cost me far more than this. Not to mention, that on top of that xmr stak has 1 (or even 0% fee) and a slightly lower rejected shares rate, than the one I am having with the cast. The neat feature for showing temps I like a lot, but that cannot compensate all the drawbacks. So I moved back. I must mention, that it was working good on a single Vega 64 at 1407/870 1100/850 , expect the random hashrate drops that I mentioned earlier with a screenshot (same as on the 6 card rig). So it had 50 H/s more before calculating fee and higher % stale/rejects. Still - with some work on the code, the program might be worth trying again. I will defnitely follow it and give it a try with next versions, but several things just need to be improved. Starting with: 1) Adjustable intensity settings. 2) Fix or at least explanation for these random hashrate drops. Is this being cause by collection of the dev fee? This is my suspicion, and if this is the case ... Well, it looks like the fee is not 1.5% but way higher. The program gives no indication when this dev fee is being collected anyway ... 3) Lower dev fee  I don't mind paying 1% tbh, at least for some time, but 1.5% is just too high, when there is a cheaper (and probably better) client on the market. 4) Maybe a way of restarting the program, or even the rig if nothing is happening for some time. Maybe you tried it already, but looking at your settings I would bump the GPU up to 900 to 904 (keep it at 1400-1408, e.g., 1405/900), and on memory keep 850 but try dropping to 1070 (1070/850). I've tried both XMR-STAK and Cast. Both are stable in my particular setup. I think Cast gives just slightly better total speed and just a hair less power usage. 3 cards so far but adding more this weekend. |
|
|
|
How much heat wattage you need? 100% silent would be difficult, think about watercooling, and a very big radiator. Maybe from a car?  Vega 56s + Raijentech aftermarket air-cooler - almost silent. PM me if interested in details. |
|
|
|
RX470 rigs are still difficult to upgrade, tbh, as it's impossible to get as much bang for the buck currently... And they're still chugging along nicely. I've been contemplating a Vega56 rig for a while but I just can't seem to make the step. Mostly for logistical reasons... But time to breakeven is definitely a big factor.
Vega at this point only seems to be good with Monero - and only then on Win 10 as the bloody blockchain driver needed to get good Monero hashrate DOESN'T WORK on Win7. (Perhaps LINUX, as I think AMD finally got around to updating the LINUX driver). Thank goodness I didn't get my Vega 56 card with the intent of mining on it. yeah... XMR is what I ran my 470 to Vega-56 ROI scenario on as it appears Vega can get close to 3x the hashrate of my 470's on Cryptonite algo. Assuming I could get $200 each for my 470's and replaced them with 4 Vega56's, I figure my ROI at only about 60 days. That is also assuming a total at wall wattage of 1KW for the 4x Vega's. Anyone here know if that is close? I'm assuming that I cannot go any higher than 4 Vega's mainly due to lack of extra 8-pin PCIe power connectors. My 1200W PSU only has 4 x2 so if both power ports need to be filled on Vega's I think this would be limited to 4 of them. But, if those 4 will take anywhere close to 1000w then I wouldn't have enough headroom to add a 5th Vega anyway. This isn't quite a no-brainer yet but if XMR were to have a few more good days and hold, it's probably a worthwhile endeavor... at least on 1 of my 4 rigs to start. Eh, 3x 470's? Not really, but at least 2.2 - 2.4 and slightly less power draw when set up right. You do need both power ports filled on the top of the Vega, but you can get Y splitter cables that plug into a 6-pin and then branch out to (2) 8 pin connectors. Three Vegas will mine 5,600/5,700 and draw under 600 watts from the wall, so a fourth should not take that over 1K. I personally would not add a 5th to a single 1,200w PSU though. |
|
|
|
Dealing with a warranty issue right now with Parrallel Miner / Centrix. Do you think this is how you should treat a customer that have purchased over 400 of your PSU and recommended others? They push the blame / burden on the customer when they failed to put stickers on some of the PSU instead of track via Serial # like everyone else. They are the only vendors I know of that sells these 2400W PSU. My previous go to PSU was the IBM 4KW bundle but they are harder to get these days so I went with these due to the flexibility and low prices. However the 2400W PSU from Parrallel miner are very poor quality and I've had over 40 out of the 400+ failed, they are also very loud under higher than 50% load. I only requested 20 of the PSU to be replaced because I knew chances will be very low to send in the 40 and get them replaced by a small vendor. I was willing to eat part of the loss but seeing Albert's response really disappointed even more.
Look at the response from the vendor. Looks like they are trying to weasel their way out of replacing the PSU just because cost increased. When you sell a product they always factor the warranty cost into the profit margin. A warranty is only worth something if they're actually going to honor their promise. It's just wrong to say we're low on stock or price is higher so you're out of luck buddy. There was no credit offer or we'll help you this one time but "our new policy is, yadda yadda. It also took 3 emails and almost 1 month before Albert would respond to me and only after Joseph from Centrix step in.
Is this how you want to be treated as a customer? It is always cheaper to take care of your customer than to ruin your reputation and lose even more business thru 1 telling 10 and 10 telling 100 more. I asked Albert to put himself in my shoes and see if what he's doing is right. Let's hope he rethink the situation / resolutions. This is not how you treat your customer large or small. What's your take on this?
<snip>
I can't really comment on the Centrix situation, but now I am curious and will check my own Parallel Miner units to see about this "warranty sticker." One thing that does help with the loud small fans: add a cheap finned heat sink to the top of the unit (if space allows for it). Finned aluminum heat sinks from China are very inexpensive through eBay, though they do take 2+ weeks to arrive. |
|
|
|
I have some weird problems with Vega.
I had 3 of them running on a rig but couldn't get the blockchain driver working. I clean-installed everything and then the driver finally got detected by Cast-XMR.
However, power usage was through the roof and before I managed to get the powerplay mods installed, the PSU got overloaded and somehow damaged the setup when it shut down. Couldn't get the stuff running again but found no faulty components when trying them separately in other systems either...
So I moved the 3 amigos into another system and repeated the procedure. Started with 2 cards and this time, I applied powerplay. Now blockchain driver is recognized by Cast-XMR. Hashrate is around 1000 per card regardless of how much fiddling and toggling with devices or HBMC I try. Hashrate for DaggerHashimoto is a steady 44.1 so blockchain driver is definitely in use.
When I try dual mining, I get an instant hard crash no matter the settings with clocks and voltage.
When I connected the third card, the entire system bugged out and wouldn't even post. Had to tear out all GPUs and reset BIOS.
Hard to see what that has to do with the driver but I previously had all the cards running in this machine with newer drivers.
The question is why CAST-XMR gets such bad performance. I got better without the blockchain driver.
Personally, I could never get the PowerPlay modifications to work in the Windows registry. They always caused terrible hash rates on my Vega 64. (I think it is also because my Vega 64 does not overclock or undervolt well.) I ended up using Display Driver Uninstaller in safe mode to remove my AMD blockchain drivers and all of the registry modifications. After restarting my PC, I reinstalled the blockchain drivers. I did not apply any SoftPowerPlay modifications or use OverdriveNTool. In your case, HansHagberg, you probably should just disable crossfire and avoid any other registry modifications. R0mi mentioned how to disable crossfire in post #302 above. I enabled HBCC and set the slider to the max 16GB my system had available. I experimented a little with grandalph3000's (a.k.a. glph3k) recommended settings in Wattman. For my Vega 64, setting the core clock to -10.5%, setting the memory clock to 1025 MHz, and setting the power level to -15% was the best option. (-20% power or lower made my system crash.) Those settings got my Vega 64 to 1900 H/s in CastXMR. I hope it helps, HansHagberg! I wouldn't be so quick to blame Cast. I had two cards running fine but nothing could get the third to work. Then Windows snuck in an update to build 1709 (I think) and then three cards worked fine. Do disable crossfire though. |
|
|
|
This miner totally outshines the RX 4xx/5xx series as the best of them could make you up to 700 Hash/s with modded Bios. Now this GPU which costs in Ebay the RX VEGA 56 in US cheapest price of 450-490 USD brand new gives us the hashing power of 3xRX 580 with modified Bios. Well done to the creator of such thing but why am I not seeing a fall in price of the RX 4xx/5xx series yet, they continue to be 300 USD or more at ebay ?
Mmm... even an RX 470 could do 860+ Hash/s. Not sure why you list 700 H/s as the maximum. I like Vega as well, but my god what a hassle between the buggy blockchain driver, crossfire always enabled by default, and pos Wattman. |
|
|
|
that aug 23rd driver just crashed my 3 vega 56 build, only the robin hood drivers seem to work
I cant get over 1500 hash using your cast-xmr , and thats with 1400 core, 1090 mem and 20% power limit.
My garage is quite cool temps not over 70 degrees so its not throttling
help pls
Go back to the BlockChain drivers. Use DDU to clean everything out. Restart. Install AMD Blockchain driver - DO NOT RESTART. Instead, go into Regedit - HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Class\{4d36e968-e325-11ce-bfc1-08002be10318}\ Find folder 000 - turn EnableAutoCrossfire to "0" Find folder 001 - turn EnableAutoCrossfire to "0" Find folder 002 - turn EnableAutoCrossfire to "0" Then restart. That should bring your system up without crashing. I prefer to use OverDriveNTool and the SoftPowerPlay registry mod. Find a registry mod that works (meaning you can install that registry mod and restart and your system does not crash) Then save as many copies as you need with one edited for "000" another edited for "001" another for "002" and so on. Then you just need to double-click on them and restart. Try 1405 core / 905; Memory 1070 / 840 - use SoftPowerPlay mod Reset HBCC in Wattman for each of your three cards. Then launch Cast. |
|
|
|
I've RMA'd two already. This will be the third. This time I'm going to try to ask them to send me another model back. I'm putting 600W load onto them. It's within spec. They should really not die within a few months. I believe they probably use better quality components on the higher-range of their PSUs... but then they shouldn't advertise the GQ series as being "gold" efficiency (if they're only going to be "golden" for a few months that is - they are pretty dark brown in my book).
Yeah, the GQ series should be avoided. G2 is a different story, very good. I had a 1,000 GQ crap out pretty fast. EVGA did turn around the RMA pretty quickly though. |
|
|
|
Vega 56 can do it on XMR-STAK (1800-1900+) if you apply aftermarket air or water cooling. |
|
|
|
I hadn't originally planned on putting this out there, but some other asshole decided to charge a fee for this info. Fuck that guy. Here's how you run 4+ Vega at full speed while mining XMR.
Basic Users:
1. Use the Blockchain drivers.
2. Open up your device manager.
3. Open up the "Display Adapters" dropdown.
4. Right click on one of your Vega's.
5. Disable that fucker.
6. Wait a few seconds.
7. Right click on the Vega you just disabled.
8. Re-enable that fucker.
9. Repeat steps 4-8 for the rest of your cards.
10. Enjoy mining at full speed with as many Vega as you damn well please. Probably.
XMR: 42e8AWjcirkBDzCNnSjPxeGJeht71kFfWcoxxCWsxe8HZqP29NruDsxcvVSjbKw17AUDepopK7ZYCUn mRvcGS9kBT5XWhMQ
ETH: 0x74Ed2CA095Dd3aE98D88d5ca1dDb77E752152938
Interesting. I was banging my head against the wall trying to get a third Vega running on an MSI X370 Ryzen build. I noticed that if I went into the Registry and disabled Crossfire and ULPS, the third card would show up, but there were still problems. Question: the command to launch OverdriveNTool -p0Vega0 -p1Vega1... is that pulling up separately saved profiles for each Vega card and applying them, or something else? Thanks. |
|
|
|
|
Sapphire RX Vega 56 sold by NewEgg.com on eBay w/free shipping: $419
|
|
|
|
My two vegas were too loud for the office even with music playing so had to move them to the hallway. Living room for me would be a nono.
The main issue is that once you downclock the gpu and overclock the ram the card gets stupid and allows the ram to reach 65 degrees at which point it trottles thus reducing performance. To overcome this you then need to set the fan to at least 3000rpm which gets quite noisy!
So I would certainly say avoid vegas if noise is an issue...
Or try this maybe: https://www.youtube.com/watch?v=THaMlkNDzYs |
|
|
|
|
Thinking about building a new rig on one of those Onda motherboards and keeping it in the living room as heating over winter, not sure how I'm going to get the wifes buy in on that though!
[/quote]
I think Phil sent one over to Rick @ SpotsWood so that he could come up with a case. So long as you can find a way to put into something like a server case, you can get buy in.
|
|
|
|
|
I'm running 3 Vegas with 5 nvidia gpus in 1 rig without any problem. It was speed drop but I solved it by removing nVidia-related tasks from Windows Scheduler.
[/quote]
I have a question for anyone else who is trying to get 3 or more Vegas working on a MSI X370 Gaming Pro Ryzen board. Two Vegas -- not a problem; mines fine. Add a third Vega card and you can't boot. After reinstalling Windows, adding RAM, switching RAM, DDU cycles, driver installs, and so forth, I found that if one goes into the Registry folder for each card and disables crossfire (set value to "0" instead of "1"), I can then boot with 3 Vegas. They appear just fine in Device Manager, I can reboot, etc., and three Vegas show up in Device Driver. But when I try to use Radeon Settings, or OverDriveNTool, or GPU-Z the computer freezes, them reboots, etc.
To be clear, you can't get to AMD Radeon settings to disable crossfire once you install a third card because the computer freezes on startup. You can do it in the Registry settings after you get two cards working, then you install the third.
If you figured a workaround, please PM me. I think this is specific to the MSI X370. Yes, it has the latest bios, yes I have 16GB RAM, yes I set the Paging file to over 30GB.
Also, not a great card for a dual PSU type of setup. If you power the Vega from one PSU but remove the riser from the motherboard during troubleshooting/rebooting, the card will just heat up and the fan will never spin. It gets scary hot. Maybe a reason not to flash to the 56 with the 64 bios...
I think it is crazy the the beta blockchain August 23 driver for miners by *default* sets Crossfire to active. Why?
Separate question: if I can control Crossfire settings from the Registry, why can't we control the HBCC settings and values for each card from the Registry? I hate the bloated, slow, adware Radeon Settings app. I only need to do four things using Radeon Settings: reset HBCC and set the value/hit apply; change memory, fan speed, and temp to manual from automatic. It is easier to control everything else from a saved profile in OverDriveNTools. Anybody figured out how to control other Vega settings from the Registry?
|
|
|
|
|
Show Posts
