Quote from: Brockshad on July 16, 2017, 07:17:50 AM

Quote from: mu2pilot on July 12, 2017, 07:21:28 PM

Thanks to everyone who has contributed to this thread!

I'm looking for help in getting my 2 XFX RX 580 GB Hynix to hashing > 30 Mhs. I'm close and getting a consistent 29-29.5, but never go above 29.5. I've made the following changes:

GlobalCore 1200
GlobalMem 2175
GlobalPowerTune 3
Globalminer Claymore (9.7)

Modded the GPU Rom to copy the 1750 Mhz Timing setting to the 2000 Mhz setting.

I've seen several instances of people getting over 30 Mhs with this card. How can I get that last 1 Mhs?

My numbers are based on single mining ETH.

TIA.

mu2pilot

I have literally spent days on this... I ordered 60 RX 580 Nitro+ 8GB GPUs over a month ago. The first round of them started showing up this week (First two that I tested with are Hynix mem). Out of the box I was getting 24Mh/s. Updated drivers and was getting 21Mh/s. I spent hours and hours looking for info on how to get my hash up and was getting no-where. I had a min panic attack but then came across the below post.

http://1stminingrig.com/best-bios-rom-sapphire-nitro-rx-580-8gb-oc-hynix-memory-30-mhs/

User Ti posted down in the comments how they got it up around 30Mh/s (Real Solution). I did what was suggested and was getting about 28-29Mh/s. I just like you wanted that extra 1Mh/s as that plus 60 GPUs is a big deal. I tweaked and tweak in TriXX and below is how I am currently getting in the upper 30 and even 31 on occation Mh/s

GPU Clock: 1275
Mem Clock: 2300
GPU Voltage: -66mV

for the most part it seems stable but only been running for like an hour. I'll post back if anything changes. Good Luck!

http://i.imgur.com/yvyX2Xs.png

EDIT: Once I started hooking up more than one GPU to the rig it would crash. Getting VM_CONTEXT1PROTECTION_FAULT_ADDR while trying to mine on SMOS. In Windows 10 once I would plug in the 2nd GPU the first would hash 1-2 Mh/s below the norm and the 2nd GPU was hashing at 4Mh/s total. There is something wrong with running these cards in windows 10. I've tried driver version 17.1.1, 17.4.3, 17.4.4 and 17.7.1 and all react the same. If I don't have an HDMI plugged into the card running a monitor the Sapphire LED goes off after boot and I can't get any temps or readouts. If I start mining with it the LED comes on but I can't see any temp/fan read out in Claymore and it gets very hot. Overall just very glitchy and unreliable in Win10. Why would running 2 GPUs cause this? Any ideas?

Right now I have it stable for about an hour in SMOS (ETH 29Mh/s - SC 585Mh/s) but I had to reflash the bios and remove all undervolting. Just have copied mem straps as suggested in the link above. Only downer is that each card is pulling 180W as before with the undervolt it was under 100W.

If anyone has successfully got these running (6 or 8 gpus) on a rig and stable around 30Mh/s I would really like to know the secret magical settings. Any help would be very much appreciated.

Do not waste your time with Windows. This system is only for small rigs.

I have 27 rx580 8G Nitro hynix, 3 rigs with 9 gpus each. Only works with Ethos. I've spent weeks of testing and more testing. (Ethos is a big shit but only works steadily at 30MH in this SO)

The Saphire rx580 8G Nitro also seemed to me the best cost benefit, but that's not what I'm seeing. It is low cost but only for ETH mining.

I have another rig with 7 Nvidia 1070 Gaming X worked perfectly in double mining with 31Mh ETH and 310Mh SC without having to change the bios or overclock, running Windows 10

Update: I discovered that all Linux-based systems (Simpleming and Ethos) can not do undervolt on AMDs, so going back to Windows with limited GPU, 8 by rig.