OS
Pool
ASIC Firmware
ASIC Hub
More
blog
Blog
News and Articles
forum
Forum
Top discussions, announcements, help
Show all articles
Referral Program
referral
Increase your income with Hive. Invite your friends and earn real cryptocurrency!
Get your referral link

AMD GPUs and power

AMD Cards
#1

Hello,
I have 6 amd cards (5 rx570, 1 rx580) and they all run about 80w per. I have flashed the rx570s a while ago and they still work fine. Using the OC feature in HiveOS, I’ve only set CORE1100, DPM 1, MEM 2000 across all of them.
80w each seems reasonable for each?
Now the more concerning thing is the power draw total from the wall. I’m using a TPLINK HS110.
It’s saying my total power draw is over 1000w!
When I stop the mining, it’s down to 270w from the wall, and HiveOS has the gpus are roughly 210w.
So is it common that my system (mobo, psu) draws an extra ~500w while it’s in use? that seems high.
When on total: 1030w
When mining gpus: 480w
When off total: 270w
When idle gpus: 210w

So there is a 60w diff when off, but a 500w diff when running…Anyone have any ideas to help?
I’m using claymore mining ETH.

#2

GPU card it’s very complicated device
AMD Polaris cards (400/500 series) has only power usage sensor for GPU core.
So what you see on dashboard it’s only partially true.
Actual power draw for 400/500 series cards when mining ethereum 120-140W per card.

#3

Hello,
AMD gpu’s don’t display an accurate number in the power consumption menu. Generally, you can add 40W to the displayed power draw. Also, your total power consumption per card is very high. My guess is that they run very hot. Try reducing the core voltages down. My cards run stable at 815mv
Good luck

#4

I’m going to reduce to 850mv and see if it stays stable and work down for there. It’s dropped roughly 200w already. Thanks!

#5

So I lowered them but it’s seemed to hit a wall - is this due to the vbios? I set them to 820 but there was no change. I run amd-info and it’s saying they are all on 900mv
=== GPU 0, 01:00.0 /sys/class/drm/card0 ===
Adapter 0: PCI 1:0:0: Ellesmere [Radeon RX 470/480/570/570X/580/580X/590]
Core: 1100 MHz, Mem: 2000 MHz, CoreOD: 0, MemOD: 0, Vddc: 900 mV
PerfCtrl: manual, Load: 100%
Temp: 70°C, Fan: 25.8824%
Power: 59.232 W (cap: 120 W)
Core Clocks: 300 1100 980 1072 1139 1203 1245 1281
Memory Clocks: 300 1000 2000
DPM 1 CORE 1100Mhz 900mV MDPM 2 MEM 2000Mhz 900mV VDDCI 900mV REF 5 PWR 59.187 W RPM 953

=== GPU 1, 03:00.0 /sys/class/drm/card1 ===
Adapter 1: PCI 3:0:0: Ellesmere [Radeon RX 470/480/570/570X/580/580X/590]
Core: 1100 MHz, Mem: 2000 MHz, CoreOD: 0, MemOD: 0, Vddc: 900 mV
PerfCtrl: manual, Load: 100%
Temp: 69°C, Fan: 26.6667%
Power: 63.101 W (cap: 120 W)
Core Clocks: 300 1100 980 1072 1139 1203 1245 1281
Memory Clocks: 300 1000 2000
DPM 1 CORE 1100Mhz 900mV MDPM 2 MEM 2000Mhz 900mV VDDCI 900mV REF 5 PWR 63.124 W RPM 1111

=== GPU 2, 05:00.0 /sys/class/drm/card2 ===
Adapter 2: PCI 5:0:0: Ellesmere [Radeon RX 470/480/570/570X/580/580X/590]
Core: 1100 MHz, Mem: 2000 MHz, CoreOD: 0, MemOD: 0, Vddc: 900 mV
PerfCtrl: manual, Load: 100%
Temp: 70°C, Fan: 42.7451%
Power: 74.141 W (cap: 140 W)
Core Clocks: 300 1100 952 1046 1112 1174 1184 1215
Memory Clocks: 300 1000 2000
DPM 1 CORE 1100Mhz 900mV MDPM 2 MEM 2000Mhz 900mV VDDCI 900mV REF 5 PWR 74.167 W RPM 878

=== GPU 3, 06:00.0 /sys/class/drm/card3 ===
Adapter 3: PCI 6:0:0: Ellesmere [Radeon RX 470/480/570/570X/580/580X/590]
Core: 1100 MHz, Mem: 2000 MHz, CoreOD: 0, MemOD: 0, Vddc: 900 mV
PerfCtrl: manual, Load: 100%
Temp: 70°C, Fan: 25.8824%
Power: 58.198 W (cap: 120 W)
Core Clocks: 300 1100 980 1072 1139 1203 1245 1281
Memory Clocks: 300 1000 2000
DPM 1 CORE 1100Mhz 900mV MDPM 2 MEM 2000Mhz 900mV VDDCI 900mV REF 5 PWR 58.195 W RPM 982

=== GPU 4, 07:00.0 /sys/class/drm/card4 ===
Adapter 4: PCI 7:0:0: Ellesmere [Radeon RX 470/480/570/570X/580/580X/590]
Core: 1100 MHz, Mem: 2000 MHz, CoreOD: 0, MemOD: 0, Vddc: 900 mV
PerfCtrl: manual, Load: 100%
Temp: 71°C, Fan: 28.6275%
Power: 61.173 W (cap: 120 W)
Core Clocks: 300 1100 980 1072 1139 1203 1245 1281
Memory Clocks: 300 1000 2000
DPM 1 CORE 1100Mhz 900mV MDPM 2 MEM 2000Mhz 900mV VDDCI 900mV REF 5 PWR 61.182 W RPM 1119

=== GPU 5, 08:00.0 /sys/class/drm/card5 ===
Adapter 5: PCI 8:0:0: Ellesmere [Radeon RX 470/480/570/570X/580/580X/590]
Core: 1100 MHz, Mem: 2000 MHz, CoreOD: 0, MemOD: 0, Vddc: 900 mV
PerfCtrl: manual, Load: 100%
Temp: 69°C, Fan: 22.7451%
Power: 84.019 W (cap: 122 W)
Core Clocks: 300 1100 751 1019 1074 1118 1119 1120
Memory Clocks: 300 2000
DPM 1 CORE 1100Mhz 900mV MDPM 1 MEM 2000Mhz 900mV VDDCI 900mV REF 5 PWR 84.23 W RPM 1089

#6

Hello,
AMD does limit the voltages to an extent. However, yours is not lowering due to your bios. If you have 8GB cards, an easy and fast way to tune them is with OverdriveNTool on windows. They fields are very similar to hiveos. Make sure to flash the original bios back to the cards, otherwise you will again be stuck with that 900mv cap, and lower them until the PC crashes. Then flash the last, stable numbers to the bios.
If you have 4GB cards, your only option is linux but it will be very time consuming. Flash your cards back to the stock bios and start lowering the voltages in hiveos. After that’s done, flash the stable numbers to the gpu’s.
I have had this same problem on my cards where they wouldn’t go lower, but flashing to the stock bios allowed me to find lower values. After that, I used “oneclicktimingpatch” to bring the hashrate up even more.
I can also see that your cards run very hot. In the hiveos web gui, in the top right, there is an option called “autofan settings”. Change the “Target Core Temp” to something like 50C and your “Min fan speed” to something like 30%
GPU2 also looks weird. Maybe try applying new thermal paste to the gpu. (or maybe that’s just the rx580)
I hope this helps