GPU Core Clock Fluctuation Identified
GPU Core Clock Fluctuation Identified
It seems a bit unusual. However, I'm unsure if this applies to Radeon cards. Maybe consider installing MSI Afterburner. Launch AB and navigate to the graph view within AB. Apply GPU stress (such as using Heaven) and check for temperature, power, and voltage limits. I believe your GPU operates at around 95-100% load when running Heaven? That should be sufficient.
Higher voltage means greater power consumption. Raising the voltage would cause you to reach the power ceiling more quickly; you should lower the voltage instead. The goal is to keep the voltage as low as feasible while maintaining stability for optimal performance.
Power limit defines the maximum power the card can draw. Exceeding it triggers throttling, reducing core speed to stay within bounds.
The outcome reflects data from a single test cycle. (Performance: 1 benchmark, efficiency 0.931v +5% power limit)
Thermal readings stayed within safe limits at 63°C.
Power consumption reached up to 89.0 watts.
Voltage remained stable around 0.956v.
Usage fluctuated between 72% and 100%.
It seems you're trying to understand the relationship between clock speed, voltage, and power limits. Essentially, if you require a higher clock speed, you'll need more voltage, and when you need more voltage, you also need a higher power limit. The goal would be to find the highest clock speed with the lowest voltage. Regarding your power limit, it's usually best to keep it at 0% or a small percentage increase. Also, considering a GPU with a -50%/+50% power limit could be a good idea.
Yes, it seems you're getting the idea right. Boosting just the frequency also boosts power, though raising the voltage has a much bigger impact (theoretically, higher frequency raises power linearly while power grows with voltage squared). You can keep the power cap at +5%, but it will only activate the extra allowance if necessary. You might be able to adjust the BIOS settings for a higher limit, but if the card is set to default a +5% increase, it's unclear if it could support something like +50% given the VRMs and cooling system.
I’m considering upgrading with another RX470 model from the same brand. From what I observe so far, there are several variants of the RX470 from Powercolor. I checked their bios and power requirements. The Red Dragon (1x6 pins) has a 5% power limit, Red Dragon v2 (1x6 pins) also 5%, Red Dragon Mining (1x6 pins) 5%, and Red Dragon Mining (1x8 pins) 5%. My card is the Mining version with 1x8 pins, so the only other one with that pin type is the Red Devil. The main differences are the higher power limit on the Red Devil and its cooling solution.
Looking at the power pins shouldn’t imply my card draws the same amount of power as the Red Devil, right? That suggests my card might be built to handle that power level like the Red Devil.
Is there a way to compare the designs without physically examining them? I’m currently reviewing the bios and specifications from both cards, and so far I don’t see any differences except the higher power limit on the Red Devil version.
Power pins only define the maximum power the card can draw through the connector, they don’t indicate the actual consumption or details about the VRMs or cooler. Where are you checking the BIOS information? Do you have a reference to the specifications for your card? I’m having trouble finding the 8 pin mining version. I wouldn’t update the BIOS of another card. There are various Polaris BIOS modification tools available (created by or for miners, though they can be used for other purposes), so it’s better to modify your own BIOS. You can use ATIWinFlash to back up your BIOS from the card, make the necessary changes (keep a copy in its original form), and then re-flash the modified BIOS with ATIWinFlash again. Proceed at your own risk.