What Pool is best with Teamredminer runing vega 56 & 64, I have been using Hiveon but is there a more profitable setup.
Hi,
Is mining on egpu slower than direct on the mobo? My guess is it shouldn’t be but the highest I could get is 31Mh/s for my gigabyte vega 56 reference. The pool hashrate is higher though. Using team red miner.
I have oc the card to 1030 core, 960 mem. Tried all different straps too and currently using:
–CL 19 --RAS 22 --RCDRD 9 --RC 36 --RP 14 --RRDS 3 --RRDL 3 --RTP 6 --REF 17000 --RFC 249
The miner is set to eth_config=B mode.
Am I missing something?
Oh and compute mode is set already. Power draw is about 110W.
It’s a MUST to flash your vega 64 to 56 no questions asked.
See my vega 64s flashed by vega 56 bios.
start
core 1020 dpm 0 vdd 830 mdpm 0 mem 960 pl 150
amdmemtweak --CL 20 --RC 37 --RP 11 --WR 14 --CWL 8 --FAW 12 --RAS 20 --REF 65535 --RFC 248 --RTP 5 --RRDL 6 --RRDS 3 --WTRL 9 --WTRS 4 --RCDRD 12 --RCDWR 12
If that works, with one card, try to increase mem by 5 if no luck, go up with RCDRD to13 and try again.
If 960 is max mem, back to RCDRD 12 and try to go up with core by 10 at the time and give gpu some time in between with new core value and see if u start to get invalids, if not, another 5-10 until its crash or gives invalids.
My 64 flashed to 56 is maxed out on mem 980
I have had good luck with these straps on 56, but not with same brand gpu, these allows me to go mem 1045 that is unusual, some say its likely that this card is actually running with 1028 that is “max” for 56… core 1130 +56mh.
amdmemtweak --RC 37 --RP 12 --WR 13 --CWL 8 --FAW 12 --RAS 22 --REF 65535 --RFC 239 --RTP 6 --RRDL 5 --RRDS 3 --WTRL 9 --WTRS 4 --RCDRD 12 --RCDWR 12
Here is many guys with same brand 56 vega´s, im sure they can be more help to u with strap´s&oc´s.
good luck
Thank you for the help, I started to buy your straps 48,77. How can I push it further?
What should I do with my frequency values?
the one you are using now
CORE 1010
VDD: 830
MEM:950
I’m determined with these issues, but I couldn’t see 50mh / s 
( amdmemtweak --gpu 0,1 --RAS 26 --RCDRD 12 --RCDWR 12 --RC 38 --RP 12 --RRDS 3 --RRDL 5 --RTP 6 --FAW 12 --CWL 8 - WTRS 4 --WTRL 9 --WR 13 --REF 24000 --RFC 239
amdmemtweak --gpu 2 --RAS 23 --RCDRD 18 --RCDWR 12 --RC 36 --RP 13 --RRDS 2 --RRDL 3 --RTP 6 --FAW 12 --CWL 8 --WTRS 4 --WTRL 6 --WR 13 --REF 20600 --RFC 244
hi. i on my 56 hynix: amdmemtweak --CL 20 --RC 37 --RP 11 --WR 14 --CWL 8 --FAW 12 --RAS 20 --REF 65535 --RFC 248 --RTP 5 --RRDL 6 --RRDS 3 --WTRL 9 --WTRS 4 --RCDRD 14 --RCDWR 12
With mem 965 vdd 830 core 1020 these gives +52mh. These are powercolor and Sappphire. They can go above 53mh, but in the long run there will be issues with higher oc´s.
1 Like
Best bios for this card? 
Has a switch near ports labeled silent, standard, and oc, also has a switch near 8pin outlet, is that the dusl bios switch?
1 Like
Found a Wikipedia with some info on memory timings but there are some equations that don’t seem to apply on the stock Vega memory timings, mainly RAS = RCD + WR
I assume this is meaning the max of rcdrd or rcdwr when it mentions RCD, but the stock settings for the Samsung and Hynix don’t add up to the RAS.
I did notice that the RC = RP + RAS for the stock Samsung and Hynix Settings you posted, so I’m going to test out RAS 26 on those two cards throwing invalids to see if it helps.
VDD 812 on the reference Vegas IS very stable, as long as the mem & rcdrd are at stable levels.
2 Likes
Hi! I’ve tweaked memories and configured over&underclock in my Vega 56, so performance is decent, but there’s no way to make power limit effective… It’s just being ignored.
–enable_compute --eth_config=B
mems @ --CL 20 --RC 37 --RP 12 --WR 13 --CWL 6 --FAW 12 --RAS 26 --REF 65000 --RFC 239 --RTP 4 --RRDL 5 --RRDS 3 --WTRL 9 --WTRS 4 --RCDRD 12 --RCDWR 4
No invalid shares, but if temperature goes up sometimes the miner crashes (GPU 0: Dead). Any way to make power limit work? I see most of you have much better wattage with higher clocks and voltages.
Thank you!
If it crashes your vdd may be too low. 60 deegrees for mem is very low. If you are reaching 70+ you can push fan up to 100%, dont worry about it. Gigabyte has really good cooling.
Yeah, my problem is not the temperature now, my problem is that the GPU is consuming 50W more than necessary, because the power limit doesn’t work. Even if the cooling is good, those extra Wattage still so innecesary…
Got Sapphire Nitro+ with hynix. and get tiny amounts of HW errors with 960 memory (lowering it to 950 didnt helped).
Can straps cause HW errors?
They do. Go to a safer configuration (you can try samsung straps, for example), then I would search for the best mem clock, an finally work with the straps again, to find the best configuration for you card. This last part require time.
Step by step. 
Got a new one for yall.
Picked up a couple of “Vega 64 Engineering Samples.”
They are marked as AMD Radeon Vega 64 Engineering samples on the cards.
After playing with them a bit I am able to get about 41MH/s.
Tried flashing BOIS to both Vega 56 and 64, but flash tool always stops and says “P/N” mismatch. Not sure if anyone knows about these things or if it’s possible to even flash to a Vega 56 BIOS, but I would like to if possible?
Also, when changing any OC settings (including straps, even fan speed) my card will report 0MH/s and I need to reboot the rig for it to work again. Is there a faster way yall are doing this? Thanks
1 Like
If you don’t use those straps and you have your cards at over 1000 mem, what do they hash? Do you get same 8mh like me over 960? Also I have tried B mode but never notice any difference in watts or hashrate.
Ive never tried to run them with mem so high with out straps, but i would assume the crash pretty quickly with no straps.
I am trying out right now on a 64 I got today and flashed to 56 AMD/ATI
Samsung HBM2 · 113-D0500350-102
is this happening on multiple mother boards or just that one? and does that SSD have the stable image? its not showing mem temp on the vegas
1 Like
I could have swore I put the stable on but I am loading it on again right now just in case.