Indeed! Now we need such a magical strap for Hynix too 
EDIT: I found this one from @vsimovic really good !
But my Hynixes need higher RCDR; I’ve never had a Hynix that could go lower than 14…
EDIT 2: sorry, this post shouldn’t go in this thread as we’re talking about vega 64 flashed to 56. That is off course not applicable on Hynix.
Awesome thanks for the answer I was going crazy with this.
I have a bad Vega 64 flashed to 56 that will throw invalids or crash anytime I push Mem Clock over 960 with the good Samsung strap:
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
Just found a different setting that allowed the card to run at 1028 mem. Wanted to share incase anyone else has a 64 flashed to 56 that cant get past 960 mem.
Increased Mem to 1028, core to 1100, only change I made to the Strap is drop RCDRD to 17 (HW errors at 13,14,15 & 16). Has been running steady with no HW errors or invalids. +3 MH
New Strap:
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 17 --RCDWR 12
(same GPU, just moved it to my testing rig to mess with it)
Hello, how did you flash the BIOS? I cannot do it with hiveos, it always fails. Regards,
Hello, I have the same card. How did you flash the BIOS? I cannot do it with hiveos, it always fails. Regards,
Enable the toggles “force flash” and “reboot after flash” before selecting a vbios to flash
Very nice straps.
I made one change and I manage to boost memory clock to 1050Mhz and I have 1 HW error per every 10000 accepted shares for a single card. I can live with that.
amdmemtweak --CL 21 --RAS 20 --RCDRD 17 --RCDWR 12 --RC 37 --RP 11 --RRDS 3 --RRDL 6 --RTP 5 --FAW 12 --CWL 8 --WTRS 4 --WTRL 9 --WR 14 --REF 65535 --RFC 248
just wanted to share, thanks for this post and straps provided by @0Lucifer0
i’m able to throw in any vega cards to the rig and get to 54 mh at a low power consumption and runs rock solid for days.
the last card in the rig was newly added vega64 flashed to 56. its choking up a few invalids every couple of hours. not sure which to reduce, core or mem?
not running config B as its unstable for the mix rig.
lower mem or higher rcdrd in the straps
maybe your vdd might be too low too for that particular card
Here my vega´s Asus 56 is flashed from 64, rest oem 56. I belive i could go higher with some, but for now im happy with these.Maybe ot, but what do u guys recommend for 56 hynix mem? Looks like im getting some invalids wit 1075, perhaps should go to vdd 340. Happy with these my first vegas, great value.
i was so so tempted to buy one vega 64 and sell my two 580 8gb in exchange but i did not
as your setting vega taking actual on the wall 180 watt is not 110 or 130 watt did you check on the watt meter
and my two 580 8gb card give 64mh take on the wall 220 watt which is i find it good
Don’t believe what software show you
my whole system taking 900 watt . 80 watt system alone taking
i havent check individual gpus, but sure, i have watt meter on the wall and i have adjusted hive settings to match actual draw from the wall. im pulling with 2 rigs in total 2485W, if i keep hive settings power settings, the reading is way different. Like my rig 1 is showing 1103w combined draw of gpus on that one on gui is 688w+there will be more than 100w from motherboard and psu effiency lost, fans etc…so its may be around 850-900 w according hive, but the total draw is adjusted in hive settings with watt meter so actual draw match wath hive is showing. Havent really been thinking about single component draw, just total. I belive its well known fact that hive is showing consumption wrong…or am i missing your point here?
@0Lucifer0 Just created an account to thank you mate!
Just got myself a PowerColor RX Vega 64 and have tried numerous settings but never got it to work as well as my Asus Strix RX Vega 56…I would get close (around 50-51) but it was never stable.
I force flashed it with the PowerColor 56 firmware and applied your straps and timings and now it performs exactly like yours. Now I have to test them both on the same rig and see if they will both maintain these hash rates. Just for the record, I am using the latest version of Teamredminer (I think it’s
0.8.2.1)
EDIT: As I was writing this, the rig crashed - approx 1 hour after applying these lol.
In any case, you have set me on the right path and I am experimenting around these values. I am currently using Core: 1050 VDD: 850 MEM: 1000 to see how it goes. These settings coupled with your straps are giving me 52.90-53.00
The important thing to note - and this is something that I have seen in many videos and I have experienced first hand - is that these settings do not work from a cold boot, at least for me.
I have to:
If I do a cold start with OC’ed settings, my card goes into a frenzy and stops working for 10-15 minutes.
If someone knows a way to overcome this, please feel free to chime in. I am relatively knew to this, so any info helps.
you can now use the advanced settings to set up the amdmemtweak so do not have to command line anymore
@0Lucifer0 Yeah, I saw that a tweak → advanced config option was available but in some of the videos I had watched, I saw them enter it manually and thought maybe it worked better if you enter it like that.
That’s very helpful to know, it is however irrelevant to the fact that I cannot cold boot with these overclocks, for example after a restart.
Do you have a way to circumvent that?
Does teamredminer have a command to apply OC after the dag regeneration like TREX has for nvidia cards (“dag-build-mode”:“2”)?
And just for the record, card is chugging along perfectly with my slightly lower clocks and voltages, now @52.98MH.
My only problem would be what I asked above, so I don’t have to go through the rigmarole every time I want to reboot. Again, many thanks!
Heya, struggling with my rig, keep ending having some message like
[ 383.238611] amdgpu 0000:10:00.0: amdgpu: failed to read reg:ecc5
Is it normal ?
I can’t reset the card, and have to reboot eveyr time, is it the only way ?
Also it is 
Is it needed to flash to 56 ? and I just tech one similar from techpowerup ?
It does the tweak on startup amdmemtweak should be run on startup once not when the miner is already running
@_viktor I don’t know which model exactly your cars is but yes, cards are way more stable if you flash them with the same make and model RX Vega 56 bios. Especially if the card has dual bios, you have no risk.
At first, I got a Subsystem ID mismatch, so I had to force flash it. Waited a couple of minutes to make sure, then restarted the PC and it saw it as an RX 56.
I wrote what procedure works for me and it’s been running for a sold day with no hickups.
Good luck!
Yes, I understand that and thanks for the reply.
The problem I mentioned isn’t the straps, it’s the core and memory settings you mention in your post.
The card will not apply such high setting upon first boot / restart.
If I start it with default settings and apply the overclocks after a minute, they work just fine, as you can see below.
My only question was - for someone who might know - is if teamreadminer has an option to delay applying the overclocks.
