RX 560 hashrates

[Vlizzjeffrey]

guys, thx for all the work, im minning with sgminer now.

this rx550 elpida is nuts, it does 500+ hash constantly, its better then my hynix 460/micron 460/550.

But i have 1 problem, after a while, my screen turns black, and then into  the grey and black dots ( like snow) and then my whole system reboots.

i think the asrock 970 pro4 is not the best board to connect 4 gpu's on it, i think it has weak power phasing ( 4+1) and there is no molex connector on the board like on the btcpro boards.
But still these cards only  consume about 30 watts, so i thought it wouldn't of been a problem, hence i bought this bord for 50 euro, the btcpro boards are selling for 5x msrp now lol.

[wacko]

i think the asrock 970 pro4 is not the best board to connect 4 gpu's on it, i think it has weak power phasing ( 4+1) and there is no molex connector on the board like on the btcpro boards.

Do you plug your cards directly into the board? Or do you use old ribbon risers? Tried googling the board but it doesn't seem to exist, could only find 970 extreme4 or 970 pro3, but not 970 pro4.

[Vlizzjeffrey]

i think the asrock 970 pro4 is not the best board to connect 4 gpu's on it, i think it has weak power phasing ( 4+1) and there is no molex connector on the board like on the btcpro boards.

Do you plug your cards directly into the board? Or do you use old ribbon risers? Tried googling the board but it doesn't seem to exist, could only find 970 extreme4 or 970 pro3, but not 970 pro4.

Hey, ya sorry i made a mistake, its an asrock 970 extreme 4.


Im using oldskool ribbon risers, i got 4 cards to work now!
I needed to put on 16x to 16x  riser on the 16x slot, i think that did the trick. Other 3 cards are on 1x-16x risers.

I think the system shutting down must be my mem or gpu clocks, since everytime after it crashes and i restart the pc it says wattman settings restored due to a system failure.

Pretty annoying that i dont know wich card is the one crashing, or more then one
It's just a matter of finding stable gpu and mem clocks now for every particilar card, the MICRON cards don't even seem to run stable at above 1750 mhz.( sgminer)
Do you play with the memory voltage? or just leave it at 0.85v? Does gpu core/voltage have any influence on memory clocks stability?
My 2x550+2x460 rig now gives me 1.8 kh/s avarage 5 seconds reading, so im already very happy.

[wacko]

Do you play with the memory voltage? or just leave it at 0.85v? Does gpu core/voltage have any influence on memory clocks stability?

I keep both the gpu and memory controller voltage at 0.875v. I don't think core voltage affects the memory chips, not directly at least. And I don't recall seeing any correlation between the voltages and memory overclock on polaris cards. I think some people figured out how to increase the actual memory voltage on some cards, and that might lead to higher clocks possible, but then again it's not really advisable since there are reports out there about memory degradation on Polaris cards even at stock voltage. Most of these cheap cards don't have the memory chips cooled in any way, even the airflow from the fan(s) is barely there, so they're cooking as they are already.. increasing the voltage will speed it up even more.

Are you seeing any memory errors (in HWInfo) when the rig is crashing?

[Vlizzjeffrey]

Do you play with the memory voltage? or just leave it at 0.85v? Does gpu core/voltage have any influence on memory clocks stability?

I keep both the gpu and memory controller voltage at 0.875v. I don't think core voltage affects the memory chips, not directly at least. And I don't recall seeing any correlation between the voltages and memory overclock on polaris cards. I think some people figured out how to increase the actual memory voltage on some cards, and that might lead to higher clocks possible, but then again it's not really advisable since there are reports out there about memory degradation on Polaris cards even at stock voltage. Most of these cheap cards don't have the memory chips cooled in any way, even the airflow from the fan(s) is barely there, so they're cooking as they are already.. increasing the voltage will speed it up even more.

Are you seeing any memory errors (in HWInfo) when the rig is crashing?

Im getting 0 hw errors, but now im also getting crashes when its not even mining, so something is really instable here. Sometime only amd drivers crash, sometimes total system crashes, just shuts down and restarts itsself.
I did notice it was my rx460 2gb hynix that was crashing the miner, i unlocked shaders on that one, so i locked them again hoping that was causing the problem, but still crashing.

Then i look in bios editor what the timings are of this card and i have no clue where i got them from, they dont correspont to the numbers in this thread lol, looking at that right now, gonna take timings from this thread again and reflash.

[MagicSmoker]

It has been my experience - albeit, let's keep my member level in mind - that too high of a mem clock just causes bad shares and artifacts on the screen; it's too high of a core clock or too low of a core voltage that causes actual crashes.

That said, I noticed improved stability on my RX 570 rig after upgrading to the latest AMD driver - 18.2.1 - but haven't yet upgraded the RX 560 Cryptonight special just yet and it has been running fine ever since I lowered the rawintensity in SGMiner from 416 to 408.

[Vlizzjeffrey]

Do you play with the memory voltage? or just leave it at 0.85v? Does gpu core/voltage have any influence on memory clocks stability?

I keep both the gpu and memory controller voltage at 0.875v. I don't think core voltage affects the memory chips, not directly at least. And I don't recall seeing any correlation between the voltages and memory overclock on polaris cards. I think some people figured out how to increase the actual memory voltage on some cards, and that might lead to higher clocks possible, but then again it's not really advisable since there are reports out there about memory degradation on Polaris cards even at stock voltage. Most of these cheap cards don't have the memory chips cooled in any way, even the airflow from the fan(s) is barely there, so they're cooking as they are already.. increasing the voltage will speed it up even more.

Are you seeing any memory errors (in HWInfo) when the rig is crashing?

i reflashed with your hynix strap ( hynix) and it's been stable so far,  arround 480 hash 1100/1850
Weird thing is my rx550 4gb micron is also doing 440 hash now, while i was cursing at it for 2 days because i couldnt get it over 400 hash.
What do you get from your micron 560's/550's? Im not even sure if im using your strap, gonna check that if/when my miner crash again xD

It has been my experience - albeit, let's keep my member level in mind - that too high of a mem clock just causes bad shares and artifacts on the screen; it's too high of a core clock or too low of a core voltage that causes actual crashes.

That said, I noticed improved stability on my RX 570 rig after upgrading to the latest AMD driver - 18.2.1 - but haven't yet upgraded the RX 560 Cryptonight special just yet and it has been running fine ever since I lowered the rawintensity in SGMiner from 416 to 408.

Good you say that, my intensity is at 432 with 2 threads, so i guess thats kind of pusing it? will try 416 next if i get a crash again.

Also what is REALLY weird, i kept adding more voltage to the cards hoping for it to get stable, till a point i was at 1050 on them all, and still not stable, so i figured that cant be the problem.
Now i put them all at 900mv and its been mining for over an hour now without crashing ( new record lol )

[MagicSmoker]

It has been my experience - albeit, let's keep my member level in mind - that too high of a mem clock just causes bad shares and artifacts on the screen; it's too high of a core clock or too low of a core voltage that causes actual crashes.

That said, I noticed improved stability on my RX 570 rig after upgrading to the latest AMD driver - 18.2.1 - but haven't yet upgraded the RX 560 Cryptonight special just yet and it has been running fine ever since I lowered the rawintensity in SGMiner from 416 to 408.

Good you say that, my intensity is at 432 with 2 threads, so i guess thats kind of pusing it? will try 416 next if i get a crash again.

I didn't see any difference in hashrate going from 416 to 408, but I also haven't had any crashes since then, either; when rawIntensity was set to 416 my rig could barely make it 8 hours without crashing (the kind of crash that requires pressing the reset button, not just ctrl-alt-del").

[hack4love]

any one try this cards

Gigabyte Radeon RX 560 4GB OC GDDR5 GV-RX560OC-4GD
singel fan

ASUS Radeon RX 560 4GB GDDR5 RX560-4G dual fan

what is your hashrate

[wacko]

What do you get from your micron 560's/550's?

~ 443 H/s at 1100/1750. All 560/4 cards, I don't have any 550s.

[Vlizzjeffrey]

What do you get from your micron 560's/550's?

~ 443 H/s at 1100/1750. All 560/4 cards, I don't have any 550s.

Ya thats what im getting, i think im using your strap you posted in page 7 or 8 i think.

any one try this cards

Gigabyte Radeon RX 560 4GB OC GDDR5 GV-RX560OC-4GD
singel fan

ASUS Radeon RX 560 4GB GDDR5 RX560-4G dual fan

what is your hashrate

arround 440-460 hash cryptonight algo

The 2GB versions seem to come with Hynix and Elpida and they hash a little faster in cryptonight ( ~470-500 hash)

I also have a fx8320 as cpu in the rig, and i was thinking i could let that mine also while my gpu's are mining, but that was just a dream  i guess, that cpu is known for using a lot of power, the power delivery of the board probally isnt sufficient to run the cpu at full load while giving juice to  the 4 gpu's. I think a mobo with a molex connector to feed the pci express ports is really a nice feature, i see it on more expensive boards and ofc the ( madly overpriced) btc pro boards.

[playfast]

My MSI Aero 2gb hynix seems to like a slightly higher power limit. I get 5% higher hash with +10% power limit.

[Vlizzjeffrey]

My MSI Aero 2gb hynix seems to like a slightly higher power limit. I get 5% higher hash with +10% power limit.

I think that also could differ between hynix cards, since the power limit has more to do with gpu core then the memory.
I tried higher power limit, still getting same hash.

I also found out i get same hash with intensity 400 as with intensity 432, but on intensity 400 it just never crashes, its been going for over 20 hours now, previous record was 2 hours.

[MagicSmoker]

My MSI Aero 2gb hynix seems to like a slightly higher power limit. I get 5% higher hash with +10% power limit.

I think that also could differ between hynix cards, since the power limit has more to do with gpu core then the memory.
I tried higher power limit, still getting same hash.

I also found out i get same hash with intensity 400 as with intensity 432, but on intensity 400 it just never crashes, its been going for over 20 hours now, previous record was 2 hours.

That matches my experience, so I think there really is a correlation here. When I did my binary search on rawIntensity I got almost immediate crashing at 512, no crashing at 384, some crashing at 416, and (so far) no crashing at 408. However, I haven't been running for more than 2 days continuously because of various other issues (e.g. - yesterday Windows 10 decided to update 3 computers at once, even though I disabled updates...  )

[Mattthev]

My MSI Aero 2gb hynix seems to like a slightly higher power limit. I get 5% higher hash with +10% power limit.

I think that also could differ between hynix cards, since the power limit has more to do with gpu core then the memory.
I tried higher power limit, still getting same hash.

I also found out i get same hash with intensity 400 as with intensity 432, but on intensity 400 it just never crashes, its been going for over 20 hours now, previous record was 2 hours.

So the configs are working, that's great.
https://mining-bios.eu/guides-and-miners/cryptonight-xmr-etn-sumo-aeon-mining-guides-and-miners-configs/

[Vlizzjeffrey]

My MSI Aero 2gb hynix seems to like a slightly higher power limit. I get 5% higher hash with +10% power limit.

I think that also could differ between hynix cards, since the power limit has more to do with gpu core then the memory.
I tried higher power limit, still getting same hash.

I also found out i get same hash with intensity 400 as with intensity 432, but on intensity 400 it just never crashes, its been going for over 20 hours now, previous record was 2 hours.

So the configs are working, that's great.
https://mining-bios.eu/guides-and-miners/cryptonight-xmr-etn-sumo-aeon-mining-guides-and-miners-configs/

Yes, i got everything working, your topic about the cryptonight mining is awsome, i think you are helping so many people, because when you google about cryptonight mining your topic is one of the first hits on google bro.

[Vlizzjeffrey]

My MSI Aero 2gb hynix seems to like a slightly higher power limit. I get 5% higher hash with +10% power limit.

I think that also could differ between hynix cards, since the power limit has more to do with gpu core then the memory.
I tried higher power limit, still getting same hash.

I also found out i get same hash with intensity 400 as with intensity 432, but on intensity 400 it just never crashes, its been going for over 20 hours now, previous record was 2 hours.

That matches my experience, so I think there really is a correlation here. When I did my binary search on rawIntensity I got almost immediate crashing at 512, no crashing at 384, some crashing at 416, and (so far) no crashing at 408. However, I haven't been running for more than 2 days continuously because of various other issues (e.g. - yesterday Windows 10 decided to update 3 computers at once, even though I disabled updates...  )

Mine still crashed, after about 20 hours, system stays running but not mining anymore and black screen, have to restart pc manually.
I think for my rx550 4gb and rx460 4gb with micron, even 1750 mhz seems to be to high for them, wtf? I put them at 1700 mhz now.
the hynix and elpida can run 1850 mhz easily.

[MagicSmoker]

Mine still crashed, after about 20 hours, system stays running but not mining anymore and black screen, have to restart pc manually.
I think for my rx550 4gb and rx460 4gb with micron, even 1750 mhz seems to be to high for them, wtf? I put them at 1700 mhz now.
the hynix and elpida can run 1850 mhz easily.

All of the 2GB XFX RX 560's I've received so far have Elpida memory so I've gotten lucky in the silicon lottery. A 4th is on the way and I hope I didn't just jinx myself 

Once I get all 6 slots on my Onda D1800 filled out I am going to revisit xmr-stak and claymore because SGMiner is just so finicky to get working right and it has been abandoned for about a year now. The supposed replacement - gateless gate sharp - is still very much in beta testing (lots of crashes being reported).

[Vlizzjeffrey]

Mine still crashed, after about 20 hours, system stays running but not mining anymore and black screen, have to restart pc manually.
I think for my rx550 4gb and rx460 4gb with micron, even 1750 mhz seems to be to high for them, wtf? I put them at 1700 mhz now.
the hynix and elpida can run 1850 mhz easily.

All of the 2GB XFX RX 560's I've received so far have Elpida memory so I've gotten lucky in the silicon lottery. A 4th is on the way and I hope I didn't just jinx myself 

Once I get all 6 slots on my Onda D1800 filled out I am going to revisit xmr-stak and claymore because SGMiner is just so finicky to get working right and it has been abandoned for about a year now. The supposed replacement - gateless gate sharp - is still very much in beta testing (lots of crashes being reported).

Are you also still getting crashes? I keep fiddling with the memory/gpu frequenzies trying to ind out wich card is causing the crash, since i can't tell wich card it is, when it crashes i get blackscreen.
I might also give xmrstak another shot.
I also bought 1 more gigabyte rx550 2gb, hope it has elpida memory.
Gonna sell my gigabyte rx460 4gb micron, since for the money i can get from it i can buy 1.5 rx 550.

[MagicSmoker]

All of the 2GB XFX RX 560's I've received so far have Elpida memory so I've gotten lucky in the silicon lottery. A 4th is on the way and I hope I didn't just jinx myself 

Once I get all 6 slots on my Onda D1800 filled out I am going to revisit xmr-stak and claymore because SGMiner is just so finicky to get working right and it has been abandoned for about a year now. The supposed replacement - gateless gate sharp - is still very much in beta testing (lots of crashes being reported).

Are you also still getting crashes? I keep fiddling with the memory/gpu frequenzies trying to ind out wich card is causing the crash, since i can't tell wich card it is, when it crashes i get blackscreen.
I might also give xmrstak another shot.
I also bought 1 more gigabyte rx550 2gb, hope it has elpida memory.
Gonna sell my gigabyte rx460 4gb micron, since for the money i can get from it i can buy 1.5 rx 550.

But will that half of an RX 550 do at least half the hashrate of a full one? 

Yeah, I had another crash after about 36 hours of operation. That's pretty good, but my 6x GTX 1060 Equihash rig will run for a week without a reboot; maybe longer, but Windows Update rarely lets me go more than a week without forcibly rebooting. This despite manually disabling its service... One of the things I absolutely despise about W10.