I noticed that you are not on beta. Is your rig stable? any reboots?
hmm, I could not notice any change in mhash when changing the tref value…
Rig has been stable for 2 days so far. I switched to testing a XFX triple Dissipation card.
I’m seeing a slight increase in hash rate when changing tREF values. I also noticed that with a Sapphire Nitro+ 5700XT the power at the wall did not change from the 5700 Non XT bios to the 5700 XT bios even though it is reported drastically lower by the non XT firmware.
I’m seeing the same thing for Sapphire Nitro+ 5700XTs. Lower power reported by the firmware but stays the same at the wall.
Hi Migi, good numbers did you follow exactly the steps that 0Lucifer0 have posted above? please let me know which steps did you follow to obtain those values…thanks for your help, excellent post thread.
I was able to achieve similar hash rate values, after doing the steps described in this thread in one of my 5700 XT pulse cards. However the wattage still is more than 100W.
It has been running stable for 2 days so far, how long you guys were able to test this mod in your 5700 cards? it is safe to do this mod for a long term strategy? does it cause any potential risk to the card’s expected life?
You must fix the values… at core 1400 at vdd 750 its the safest value for this cards… at the memory side mem clock 900 not higer… (and check the temps… if its above 86 lower the clock 890…) and finally you can set the fan at 60 - 65% and then check the temps of the cards…
How did you get the mem voltage to be stable at 1275? Every time I push my PowerColor past 1150 the mem temp jumps up and the whole rig crashed.
what model of your GigaByte Card?
Great stuff will be testing this and making a video on this topic come sub so you dont miss it 
any updates so far?
Here is my 5700 XT running on NBminer 34.1
same rig with phoenixminer 5.4b
It is a Sapphire 5700 XT BE with memory strap + tRef Mod:
Memory Clock - tREF
1000 MHz - 3900 (Samsung & Micron)
1250 MHz - 4875 (Samsung & Micron)
(1500 or 1550) MHz - 6045 (Samsung & Micron)
1750 MHz - 6825 (Micron only)
1800 MHz - 7020 (Samsung & Micron)
1875 MHz - 7315 (Micron only)
2000 MHz - 7800 (Samsung & Micron)
2250 MHz - 8775 (Samsung only)
Works great.
So did some testing and made my video link here for my results. https://m.youtube.com/watch?v=Eg6JN8U35ps
1 Like
I’m still interested in that 5600… would be a lifesaver. O
I don’t think you want to push that card much more without addressing the memory temperature. In fact, it may be throttling because of the temps. I know if you look on youtube there are guides on replacing the thermal pads (I believe the MSI versions used thermal pads that were poorly placed AND too small, which might be your issue).
BUT, aside from that you will want to increase the clocks for both the memory and core to increase hashrate. See if you can do something like 1370 760/780/1300 910 and set the fan to at least 95%. You may also want to position the card so it gets the “freshest / coolest” air.
Are you running the stock vBIOS, or have you modified your BIOS? If it is stock I think you are about as good as you will get. I think HiveOS mis-identifies the PowerColor memory as Micron sometimes - because all of mine behave like Samsung memory and have a mix of the two when editing the BIOS.
I think the #1 thing that you could do is follow the directions over at Igor’s Lab for applying the Apple Memory straps on the 5700xt, or at a minimum copy the memory straps for the 1500/1550 to all higher memory functions.
Hello @SybrBlue, thank you for reply.
I have done change thermal pad, now i go around 92-94° instead of 102-104° (i confirm that MSI thermal pad are not enought and very small)
I will try your setting and tell you.
Edit : i have tested go to 55.90 mhs temp 53° mem 104°
I don’t think this is good for the cars to be so high…
Thank
Greetings … I am running stock. I will do as you stated, and see how it goes. Thanks!