Topic: VEGA 64 mining FREE WISDOM - page 8. (Read 10803 times)

very nice haveatank!!!

cool photos and nice setup
moving rigs to a dedicated space with proper infrastructure is any miners inevitability beyond a certain size threshold usually governed by electrical sufficiency
as I found out for myself with the current setup
been designing standard 10-20 containers purposed for this but different ventilation and power supply system from the company that already does this (forget their name), they can be placed in industrial parks, parking lots, or anywhere you can lease the spot and get suffient power supply, i included an electrical to heat to electrical feed back loop system to compliment the power intake thus increasing per H/W efficiency
--------------------------------------------------

i have gotten clear indicators of electro magnetic vunrelability associated with the Vegas, seems some part of them needs better shielding like in a mobile phone but for you to be able to affect with your own bioelectromagnetism is a stretch, maybe it was bad timming, but i do share your sentiment about these GPUs behaving in a very perculiar manner ie same settings on different occasions produce different resuts, seems there is a correct mix of OS Striping/MinerSoftware/NetworkSignaling/Driver/Registry/StartupFiles/BusingIssues/PowerSupply/etcetera etcetera last of which are the settings of the GPU mining

-------------------------------------------------

been talking to people about terms of contract based employment for the development of advanced Vega specific mining/compute drivers, more to follow.../

Ok so after running for like 3 days, I'm starting to think I'm personally the problem. I went to put my hand over the blower (About 5-8in away from the card) to see how hot it is and like 1 mins later it just stopped mining (frozen no screen output or anything it just stopped consuming about 230watt) Strange. After a reboot everything seems to be working fine. Can i be electrically charged or something I'm a little confused. I currently took it back to 875m/950mv from 900m950mv. I run the core at 1432c/1000mv. Also changed XMR-Stak to 1932 for one thread and 1532 for the second one on each card. I'm really curious if this GPUs are hyper static electricity sensitive. Based on something RIGER mentioned earlier in this thread. Thinking about looking for materials to help with the static electricity just in case this holds up. It funny because i have no temp problems whatsoever and it seems people problem seem to be temperature related.

I'm just going to start from square one when i move. I was even thinking of upgrading the harddrive to 240 SSD from the 120 i have now or something to make a crazy huge page file (is this wont help please shime in), going to start with stock settings and then start working my way from there. Leave it running for a week then do change, leave it running for other week and do more changes and so on. Regarding the CPU I get about 605-620h/s alone. It an Ryzen 7 1800x 8 core 16 thread CPU, 4 cores get used for mining which is 8 threads 0,2,4,6,8,10,12,14 and the rest of the core/threads are left to run the system. Core 1 handles the GPUs. In Stak you can actually assign each thread of the GPUs to a Core but i leave this at default. I have a Water cooler for the CPU, Preparing for summer. I think i posted the specs for this system in my first post.

I will admit this has been the most pain in the A$$ rig i have ever build!!!!!!. Trying to get this 4 working for at least a week or 2 like the first time and see how it goes. my goal is to make it a 6x Vega 56 Rig (not a fan of mixing cards or name brands ect Like to stick with the same thing in this case MSI Vega 56 Air boost). I think i will be returning all my Vega Watercooling equipment i purchased and use that money to get more cards or use that money to start the 1050TI rig I'm planning on building.

I will be moving all my rigs to a new space ill be renting soon. I'm going to be moving All rigs into this 24u Server rack I have from when i use to host files for my clients. I made it to support 6 rigs and got some Inspiration from a guy in FL on YT Link below on how to keep them cool and dust free. Tired of having to dust them off every 2 months or so. Going to enclose the rack with Wood Panels on the side and just make cut outs for filters and add a 24" fan on top of it to have air flow.
(https://www.youtube.com/watch?v=59I1KvEpMco&t=203s)

Picture of server rack and Rigs
https://www.dropbox.com/sh/yn5k8929ilzfa0p/AAB41dg2K5fZN2maftdFM9O0a?dl=0

Currently looking into getting some 8x GTX 1050ti for Zcash and also looking to get a system from this guys selling a completely running rig with
8x XFX RX580 8gb
Asus Primes Z270
Celeron G3930
250w Pico Power supply for MB and SSD
2x 1200w HP server PSU With break out boards and cables
120gb SSD
for
$4,200.00 doing about 240-246Mh/s on Eth with a 990w power consumption. after doing the math is about what you would pay for all this equipment.

I'm trading electrical cost at home for office space at a flat rate with electricity included (it about $100 more then what i would pay at home, but when summer comes and AC usege i can see the electric going crazy high and me not being able to support the AC and all this miners all at once) and the place knows what i want to use the office for. which means all my calculation would be easier now since i don't have to worry about two things ROI and electrical cost. I'm pretty close to maxing out the electricity in my house. So the more rigs i get in the office the more profit i can make in a short period of time. I plan on putting 4+ Outlets with 20 amp Breakers each in that office, which will let me run at least 2 rigs at 1100w per rig. My total farm right now Maxs out at about 3,900W (minors only). I figure adding 2 more vegas to the Vega rig will bring my usage up by 300-400w for a total of 1200-1300w for the vega rig. If i add the RX 580 rig which take about 1000w and the 8x1050ti which can't be more then 600w, plus about 1000w for all the other supporting stuff (fans/AC unit/Router/monitor/ect) would bring my total to 6,900w with a total of about 58 amps, so even just 4 outlets with 20 amp each should be enough.

Current farm is a total of 36 GPUs and Growing

3 Rigs Mining ETH at 550Mh/s
2500W = 4.55w per Mh/s
 - 1060 6 GB 24 GPUs

2 Rigs Mining ETN at 10.5kh/s
1,370w = 7.66w Per Mh/s
 - RX 550 4gb 8 GPUs
 - RX Vega 56 4 GPUs
    - Ryzen 7 1800X 1 CPU

if you are having to restart them more than once every 5-10 days (or even longer) there are conflict issues you need to fix,

are you using Win10 as the OS?

i have noticed that the longer a rig runs not only does it run cooler after time it also uses less wattage
for me, after following my own advice in the previous pages, bad risers and more importantly too much voltage casused the instabilities

turn down the voltage to GPU and MEM in 5mV increments until you find the sweet spot which is as individual as your genetic code,
mine vary between 900 and 905mV on the 3 rigs with the FEs being more power hungry but all three are running smooth now
mining with the CPU in conjunction is something i dont do as i generaly believe its a strain on the sys with creation of mobo bottlenecks, but thats my unsubstanciated opinion

yes , here you can see my setup.
https://bitcointalksearch.org/topic/m.29059801

I'm going to try this after it does other reboot. Funny thing is that after i wrote this it has been working non stop. Maybe it knows and got scared lol. I will prepare this once again i tried that dev tool with limited level of success, but i will give it a try again, I was thinking of just reinstalling Windows and start again. and by the way tanks for the help.  been hanging around this forums but this is my first well second time posting. Do you run vegas?

like this? got it from here https://www.reddit.com/r/MoneroMining/comments/74hjqn/monero_and_vega_the_definitive_guide/



@havetank try to run without afterburner active and look if it stays stable
and get rid of the AMD software. No need for Wattman. Install drivers by deviceManager.

@RIGID

Great minds think alike I was thinking the very same thing. I'm excited about the possibility. Again would love to get involved in some capacity with the project.

I have been mining since Sept. Currently have 4 rigs and this 5th one that I'm trying to get stable. Current farm consist of 36 GPUs 24x 1060 6GB split between 3 rigs running about 550Mh/s on ETH, 8x RX 550 doing about 3.4kh/s on ETN and trying to make this 4x RX Vega 56 Rig stable, so i can have a total of around 10-11kh/s on ETN. The 1060s are super stable. 1 of the rigs has a TDR problem, but that is solved by a reboot by windows and then everything comes back up and starts mining like nothing ever happen. sometime Nanopool doesn't even register the drop on there site since it happens so fast.

I'm currently trying to get 4x MSI Vega 56 Air boost stable. Purchased EK watercooling equipment, but I'm thinking about sending it back if i can get this stable. I can't seem to get it working more then 4-8hrs at a time now. It ran for about 3-4 weeks with only one reboot in between and after that it was never stable again. I couldn't care less if the thing rebooted every 4-8hr as long as there was a way to automate the restart disable the cards, re-enable the cards start Afterburner and then start XMR-Stak. My other rigs which mine ETH just restart once in a while with a TDR error (did DDU and reinstalled drivers but still not fix dont really care about this since the restart takes care of it), but at least this rig just start up the miner starts afterburner starts and it back to business like nothing ever happen it takes a total of 1-2 mins tops. I don't even notice when it happens. I tried that Guardian software (from vega.miningguides.com), but it almost felt like the developer made it to reboot every 10 mins, so he gets his 5min of mining. i put most of the values at a ridicules rate like 2000hs and for it to wait 7 mins if the miner doesn't hash before it restart and it just kept restarting every 7 mins. I also have a RX550 x 8 using XMR-Stak and that runs for days before XMR-stak either stops responding or the whole computer just freeze and does nothing, but with that Guardian software it just kept restarting every 10-12 mins, so i stopped using it and it back to normal. I have tried using that Dev software, but it a hit or miss. Cast-XMR does the same thing, just reboots. Temps say at around 41-47c (fans are cranked up to 4k). so i know it not temp problems. Any idea what it can be. I have tried switching risers, power supplies since i have 4 other rigs and trying other drivers you lose about 2.5k hs. This is really frustrating. I mean has anyone figure out how to automate this. I cant care about monitoring. Just a way to cut power to the rig let it restart have it disable enable the cards start the OC program then start XMR-Stak or Cast or Claymore. Reason why i want to stay with XMR-Stak is because CPU miner never seems to work together with Cast-XMR together or at least it has never worked for me or i don't know how to set it up. Is there anyway you guys have figure out how to make the computer reboot and just automate the disabling enabling the cards this is the parts i don't know how to make happen. everything else is easy for me. If i was to have this ability i would just reboot the Rig every 4hr. i wouldn't mind losing 5 mins every 4hr. better then it randomly rebooting and nothing happening till you realize it or are sleeping and the machine just took a dump and you restart everything up ran for 10 mins you went to sleep and 10 mins after everything took other dump and wake up to realize you just lost like 6-8hr of mining because something went down. There is no logs or anything for you to look into either.

Here is my XMR-Stak Set up
2 thread per card ( at 1932 intensity (i have tried going as low as 1600 on both threads per card with the same results), Also which there as a real explanation for this intensity thing.
I also Mining with my CPU  all the even numbers cores and every tried to assign the odd numbers cores to each video card with the same results. but for the most part i have it set up for core 1 to handle the GPU's which is the default value.

I use afterburner since very time i use overdrive the cards won't say running longer then 20-30 mins. AB settings
-10 Powerlimit
85 tempature
1407 Core clock
900 memory (tried 825, 850, 860, 875, 900) Left it at 900 because it almost feels like it doesn't matter)
75 Fan
Total watts at the wall is 920w

My Hardware
BioStar TB350 BTC Motherboard
Ryzen 7 1800x With Corsair H100 V2 Water Cooler (not Overclocked since the MB i have would suck anyway)
8gb Ram
120gb SSD
HX1200i 1200W Corsair Power supple
4 x MSI Vega 56 Air Boost
1 x HD5770 (monitor output)

I have used this guide http://vega.miningguides.com and i know most of you on here us a 64 and not a 56, but I'm very well aware that i can't run the card at this crazy number. I even tried the setting someone mention on here a couple post back. One thing at least with my other rigs that no one seems to mention is the Memory manufacture. does it make a difference for this HBM2 chips. just in case it does my cards all have Hynex and they are not modded. I tried some power play settings for power savings, but same things.

thanx mate, you know I use the thread myself to back track on settings and procedures

been thinking of ways to remove the scam/pyramid doubts by issuing an ERC20 of limited supply with an ICO to fund the idea.

VEGATOKEN is born...

Each contributor would buy tokens at a fixed initial price during the ico constituting his/her donation for the project which would mean each participant would hold their contribution/donation with the ability to cash it out for any reason.

thereafter the token would be traded freely on exchanges

the vega software developing team, paid in the token/BTC would work on a one off project type basis, with specific goals and time lines constituting a road map for the token itself

each advancement in HASH SPEEDS and VOLTAGE savings would mean a DRIVER UPDATE for all those invested in the Token, which each previous driver version released as freeware

a minimum ammount of held tokens would constitute participation via a snapshot of the blockchain wallet addresses at the moment of new driver release

seems better than just accepting donations for the project
what do you folks think?

I have to say that this is probably been one of the most interesting threads on bitcointalk. Thoroughly enjoyed the reading.

And @RIGID Sounds like an awesome idea. I'm just starting out in the crypto world but would be happy to make a small donation in money and time.

if you own a Vega you need this link:

https://www.cryptunit.com/

all the cryptonight / cryptonote minable coins with their relevant info/links/downloads/pools/charts etc

if they decide to revert to a non HBM2 Memory its not a Vega in the use of today sense, no matter if their marketing department labels and promotes it as such
the whole point with this very particular exhausted since launch GPU IS the ability to double (maybe quadruple in the future?) thread on cryptonight and reach 48.2MH/s on ETH amongst others

im currently stocking up on Turtle Coin at a rate of around 350,000 of them a day, no other card can boast that and lets hope future hard/soft architecture will be geared or at least balanced towards mining

been thinking of starting a gofundme to headhunt a Vega driver developer from their core team (current/past)  to compile open source freeware mining drivers  for the 64/56/FE
i for one would be willing to part with some ETH for the possibility of reaching the fabaled 100MH/s and/or 4 threads with the FE

what do you all think?

I said this a while back, AMD have stopped retailing the VEGA card themselves.   They make the chips and parts available to partners, Vega is going into all sorts of machines including some fancy Apple models.   Of course in PC stock we see the very expensive custom boards.

There isnt enough production to go around all the various sources using this top end gfx solution.    From what I can tell they intend to revise the Vega design to a more efficent process, possibly 12nm instead of 14nm I think.    At that point I expect AMD to rerelease the Vega just like rx480 turned into the rx580.   Theres a couple youtube channels that go over all the rumours and grade how likely it is to ever happen, some of is just made up rumours but I think its fairly solid they'll bring back more Vega production some time in 2018

Last I heard on Nvidia is they are 6 months behind expectations and wont bring an update to 1080 till very near end of 2018.  I'm not that surprised, the tech is there but only for people paying thousands and using it for AI.  The whole industry is going through a fairly epic revolution from little games chips into something declared as large as the next industrial revolution which is deep learning.    Could be Hype or could dwarf miners and the blame crypto takes for using the best parallel cpu technology.

after much trial and error thought and work,
the best solution by far, considering thermal dissipation is the architecture shown in the vid below

when you situate a card air intake fan so close to the hot neighbour card you are creating a thermal feedback loop that heats the cards on average 10-20C hotter than when they have room to breath and expel the heat without affecting the neighbour cards

https://www.youtube.com/watch?v=YdXUdZDNB-s

 

nope
we got 11 running on the Asrock H110BTC Pro but it was unstable
even 8 are a pain to keep running long haul

settled for no more than 6 per mobo, works well with the 1600w PSU also

HBM Temperature at 80°C ?? Isnt that to hot for 24/7 ?


I got another question for the group. I just read this @Vegaminingguides:
Is that true? No one have a rig with 8+ Vegas? Just wondering...

yep it does quite well. vega fe(the only driver that supports overdriven tool proper for me is: 2017.Q4) hashes NeoScript @2MH stable with moderate power consumption:

It's actually almost never true. There was a problem with polaris refs that were getting more power from pcie slot that the spec allows...   it was fixed in non refs later and vegas do not have this issue at all. even at the peaks of 300+W it gets almost all it needs from12v 8pins

Vegas are in stock, they are just 200% over msrp.