More
referral
Increase your income with Hive. Invite your friends and earn real cryptocurrency!

Vega 56 and Vega 64 guide

First rig in video, dungeon´s is 6x rx570 ±180mh. PSU 850W.
Second rig is 7x vega´s 56/64 ± 430mh. 1600W + 750W.
Small on the floor is new, bought used mb for 60eur had all other components with 1x rx580 and bought 1xref vega56…that is still not ready. ±83mh 750W.
Fourth rig is 4x rx590 + 3x rx580 ± 220mh. 1350W.
When normal temps these produce about 930mh.
On vega and 5xx rigs there is H110 pro btc MB´s. 6x570 have some older asrock and "new"one is Asus something with 4x pcie without splitters.

Total from the wall is around 3,5kW.

Yeah, i understand youtube etc, thats fine, maybe just a photo would be good if someone want to share , it goes on the side with some random post :wink:

2 Likes

You sure? That’s about 61.4MH/s per Vega. Then you’ve found some amazing straps that you haven’t told us yet :stuck_out_tongue_winking_eye:

I’ll see what I can do

1 Like

:laughing: right…8x vega´s on that one :face_with_monocle:

@Smining570 thats sick
I just got these two little guys, the custom server box one looks cooler but runs cards about 2-3 c hotter due to limited spacing between the cards.



^ the bottom card here is the 64 with Morpheus cooler and noctua fans

3 Likes

wery clean, i love the box fans, havent found right size so ordered one floor fan for hot days. So in server box they run hotter, those are delta fans,right? must be really good rig that one, wouldnt mind to get one :wink: really nice

Ok, my turn now :smiley:

I first started small in the garage; but there is not enough ventilation so it got too hot in there when I got to 13 gpu’s, so I moved the rig to the back of our house into a toilet which we didn’t use anymore :sweat_smile:

The backdoor was always open to ventilate, even during subzero ambient temperatures :hot_face: :cold_face:

Don’t look at the (non-existing) cable management :stuck_out_tongue:

On the bottom is my test computer for 1 gpu. I’m planning for a bigger test rig but lack time lately.

Then in the basement I created some box from old metal racks. The side panels can be removed for maintenance on the rig, and I have mounted 2 industrial fans on the basement exit, one for fresh air intake and one for hot air exhaust. I can make change their speeds depending on how hot it gets. On full power they consume 230W each and make more noise than the rig itself :shushing_face:

The 13 Vega’s run on an ASRock H110 Pro BTC+ mobo and are powered by 3 PSU’s totaling 4200W, but pull only ~2400W from the wall (or in B-mode even less, see one of my earlier posts).

I still feel I can improve it a lot, like maybe a dust filter and isolate it for noise and heat, but lack time :confused:

5 Likes

@oni Thank you for the suggestion of resetting OC’s I was able to get rid of the DPM

@ss-silver I added the tweak along with setting the VDD to 825 and I am still having no luck. Number of invalid shares has slowed down but its not stable… ie shutting down every 20 mins.

What should I be trying to get this thing stable. I will worry about increasing mh later on all my vegas

Unfortunately one of them died a lot during the night, the original 56.
But the 64>56 didn’t! (But as the 56 died a lot I don’t know yet if it keeps stable by itself this long).

I now lowered their clocks to 1050 but kept their vdd high. Will see if that keeps them alive.

that metal frame with fans is insane! :muscle: mail man brought to me allso a bit bigger fan yesterday, but this your setup with doubble fans and frame that u can modify like that must be perfect for all ambient temps. Still trying to figure out best place for this exhaust

nope, both of them are dying again :sob:

thanks! :sunglasses:

unfortunately they don’t cool the ambient air; they just make sure there is more air flow, but if it’s hot outside, the hot air is blown over the gpu’s so I see their temperatures rising anyway. But the hot air from the gpu’s is also exhausted better, so overall it should be a better setup, but as mentioned above, when outside temps go >23°C I have to tune the gpu’s down to keep their HBM’s at 70°C max.

hello,

i have a gigabyte vega 64 flashed to 56 (gpu 4), i tried so many straps but it crashs with any --REF
any advise? thanks

and none of my cards can run over 950, 960 mem, invalid share or crash! :pensive:

A couple of nice set up here! good job guys.

I treated myself to a third vega 56 this week and have just installed. The price on fleabay is coming down, picked up a powercolor for £395 which I thought was a good deal.

My Vega rig currently is as follows for a nice 231MH. These cards are beasts for their age. I’ve another 56 in my gaming rig but it will stay put for the meantime ha!

1 Like

I love this GPU (no bios mod)(unfortunately it is not stable as I expected)

amdmemtweak --CL 20 --RC 38 --RP 12 --WR 14 --CWL 8 --FAW 12 --RAS 26 --REF 21000 --RFC 248 --RTP 5 --RRDL 4 --RRDS 3 --WTRL 9 --WTRS 4 --RCDRD 12 --RCDWR 12
amdmemtweak --CL 20 --RC 38 --RP 12 --WR 14 --CWL 8 --FAW 12 --RAS 26 --REF 21000 --RFC 248 --RTP 5 --RRDL 4 --RRDS 3 --WTRL 9 --WTRS 4 --RCDRD 12 --RCDWR 12

1 Like

what are the stability issues that you’re encountering? dead gpu’s or rejected shares?
if those are original 56’s, your mem is set very high, maybe downtune a bit to 1020 to see if that is already more stable?

Be careful with those HBM temperatures! You should try to lower them or your cards won’t live very long.

Unfortunately I have no experience with these cards, but some more information would come in handy;
which vbios was original and which vbios have you flashed?
and what do you mean with --REF? that’s for other cards like rx580. Vega’s use straps and not REF; which straps have you tried?
if you still see REF in the OC window instead of tweaker, then you should update your hiveos.
straps and clocks are linked; some straps will work with low clocks, some with high clocks; you might have tried straps for high clocks on your low mem clock, but the cards might run well on straps for high clocks with high mem clock too.
Vega’s are trial and error, just keep trying and keep us posted about the progress!

I solved that… GPU1 and 2 are the Gainward, the hot GPU3 is the Asus. lol

This is my rigs:

Functional :ok_hand:. I was preparing it for one fan per GPU, but it’s unnecesary, airflow is good, ambient temp. so good.

I would like to put 7cards per rig, but one Motherboard supports six cards (nvidias), M2 Risers doesn’t work, and the other MB (MSI X470 Gaming Plus Max for both) only supports five cards (AMDs), one PCIe doesn’t work properly. I have a third X470 giving also problems with some pcis… I did a bad choice there.

In the ‘test’ rig you can see an old Antec CP850, 12years already.

2 Likes

It was dead GPU

I love the vertical setup!

Dead GPU usualy need some more VDD; yours are already pretty low to start with at 812. Start at 850 and work your way down step by step. I tend to lower by 5 and let run for 24h.

Again, are those original 56’s or 64’s flashed to 56 ?