AMD RX580's - The best ROI Card - IMHO


#82

I think I have the problem sorted.
It looks like the samsung memory requires more mvddc than the hynix so I bumped it to 880 mvddc.
All looked good so I bumped up the cclock to 1170 and left cvddc at 840. I know…not recommended but i’m trying it out anyway.
2200 mclock is also not recommended especially with ~860 mvddc but hey :stuck_out_tongue:
I may end up changing pools to alleviate the high rate of stale shares I was getting. I will see how it goes.
I found a comment on another forum (in the image below) that may come back to haunt me…we will see :smiley:
He is commenting on someone else’s settings that are less extreme than these.
If you want to follow my real-time result you can follow here for now if you want.
PS: Sorry for spamming up the forum :stuck_out_tongue:


#83

Asus Z370 TUF is my motherboard


#84

I’ve read about these settings as well. I’ll try these out tonight


#86

shit after reflashing with one click …i lost 9 MH/s per card :frowning:


#87

Maybe it’s because I used the one click bios update on a bios that was already flashed? Would that make a difference?


#88

Thank you. Will try it tomorrow :+1:


#91

I’ve got my previous mods saved so it’s not doom and gloom. Just a pain foing rhough all the cards again lol


#92

I think I might have just bricked a card/s


#93

You might be able to try AtiFlash.exe -f -p 0 path-to-new-vbios.rom to force flash.
Failing that there is another way to boot shorting out 2 chip pins. I’m at work now but can find it later.


#94

I tried to reflash and it fails from the ui. I try mining with the card and I get 0MHs

I’ll try the console force


#95

Forced the flash. But still getting 0MHs weird


#96

Phew I fixed that card. Running at 30.5 right. I’ll need to double check the other cards but that force flash saved my life thank you!


#97

Fantastic :smiley:
Not a nice feeling bricking a card.
I had to force the other day after trying to flash a gpu with more than one riser plugged into the Pci-e slots. Atiflash seems to fail/freeze with more than one device plugged in.

I have to say, I didn’t realize the importance of labeling the gpu’s and saving the bios individually for each one until I noticed I have samsung memory in one. I now have no idea which is which, should I want to flash back to stock. I will have to manually copy the memory straps back to each rom.

On a side note…
Nanopool seems to be going great guns :smiley:


#98

Mine are all hynix memory same make and model. So theoretically I could be able to flash any stock rom to any card ?


#99

And the same size file? If they are I would think so.


#100

They all saved when I orginally saved the rom with the same file size . Think it was 256kb range or something like that.


#101

Getting 152 MH/s with 5 cards now :smiley:


#102

Sweet :smiley:
I had a quick look around to see about getting 6 cards working on your m/board but there is not a lot of info out there. A few people are having trouble getting 6 going with some Z370 chipsets.
Maybe you could try a M.2 riser?
A video I saw showed the white one working but the green one not working FWIW.
On the hash rates… my brother has the same cards as me on a mining expert board but can’t get the same hash rates as me with the exact same claymore’s settings. His rig just kept crashing. In reality, with the core speed and volts raised, he seems to be getting a very similar actual eth return per card and similar temps. I think running hwinfo for the fine tuning is beneficial to get the internal volts etc to match each other since every card is different and also the supporting hardware varies.


#103

I like this post. I’m waiting for my Alienware R7 850W with 4 x RX580 to start mining.

Dell suppose to delivery in 3 -4 week…


#104

@Pangs Nice. Seems a little extravagant for a mining rig but hey :smiley:

So I have been busy testing out the limits of my mining gpus on eth in Claymore’s miner and decided on the following:
-cclock 1210 seems to be optimum across the board on all gpus at -cvddc 850 (all except 1 card which required -cvddc 860 to load)
-mclock 2200 seems to be the ceiling before the memory can’t load and all the hynix memory cards were optimal and happy at that, both on the msi gaming x 8GB gpus and the 8GB pulses. The samsung memory on the 8GB pulse, on the other hand, had a hissy fit above 2120 which gave the best hash rates. -mclock still only making a difference in steps of 20.
-mvddc I started at 850 and raised it 10 mv at a time until there were no errors in hwinfo64. It really is like running around blindfolded in a forest if you don’t use this program :stuck_out_tongue:
After doing all that I got a stable 31.25+Mh/s on all gpus and even though the settings on each gpu is different the end results are extremely similar with the volts and temps being almost identical.
6 x GPUs
31.25+Mh/s with a total of 187.6Mh/s
115W each from the wall
59 degrees C (fixed target temp with the -tt 60 command)
Fan speed 24-36 rpm, sometimes dropping to 0 on the MSI gaming X because of the OS heatsink.
@caerbannogkitty I wouldn’t recommend the use of the -tt -60 command because it sets your fans to 60% and throttles the card up and down to suit.

This is my current eth.bat file…
EthDcrMiner64.exe -epool eth-eu1.nanopool.org:9999 -ewal 0x03BEF224EAf3b41c18DAE95282a611879cf9c77d/munternet/chris1@gmail.com -mode 1 -etha 0 -cclock 1210 -cvddc 850,850,860,850,850,850 -mclock 2200,2120,2200,2200,2200,2200 -mvddc 870,880,860,865,855,865 -tt 60 -fanmin 24 -platform 1 -eres 3 -tstop 80 -tstart 50

When I look at my history on nanopool I really don’t know if the gains are worth it. 0.7 Mh/s total on the whole 6 gpu rig and I may lose something on my actual hash rate. Hmmm
Still…I had fun doing it :smiley: