AMD RX 480 Thermal Benchmark – Reference Temperatures
Our thermal benchmarking for new architectures goes deep. With Pascal, we ran the GTX 1080 Founders Edition through the usual peak load (average) and thermal vs. time tests, but we added more to it – and that's been carried over. Our benchmarking now includes a two-hour minimum endurance burn-in using real games in looping scenarios. Thermals are logged during this time to uproot potential points of failure or clock throttling, which are not always discovered when using synthetic applications. Furmark, for instance, will only stress the GPU in one specific way and won't vary load in a comparable fashion to a “real” game.
This first chart shows peak (averaged) GPU temperatures. This number is averaged at the same point in the over time graph (further down) for every card. We wait until the temperature asymptotes and stabilizes before averaging the output for an 'equilibrium' value.
The AMD RX 480 runs about as warm as the GTX 1080 FE. AMD's RX 480 cooler lands the chip at 56.33C delta T over ambient when under load. Idle, we're seeing temperatures of 7.58C. The reference GTX 1080, for comparison, runs at 57.51C load and 7.96C idle. Both are handily beaten by AIB partner coolers, which often use dual-fan solutions and larger alloy/copper heatsinks and heatpipes.
Like the GTX 1080 and GTX 1070, the AMD RX 480 seems to peak around 82C absolute when under its stock (auto) settings. We'll look at fan RPM vs. temperature in a moment, along with throttle checks.
AMD RX 480 Thermals Over Time
The data used to generate the above graph is represented below as a “versus time” metric, something we've begun expanding on:
The RX 480 is represented in bright yellow. The GN Hybrid is the lowest temperature line on the chart. The GTX 1080 FE is the dark blue line that collides with the RX 480.
Endurance: AMD RX 480 Thermals versus Frequency (& Thermal Throttling)
The clock-rate is a little shaky at first as dynamic clock fluctuations range from 300MHz to the boosted 1266MHz, but this is stabilized as the card ramps into its workload.
Let's zoom-in on one of the spikes to better detail the “amplitude” of our line:
This chart is cropped to 4200-4600 seconds of the test. We see fluctuations from ~1150MHz to ~1240MHz, which is nearly a 100MHz range. That's enough to cause a very brief dip in 0.1% low and 1% low frametimes (both of which improved with our core-locked overclocking), but is otherwise easily overlooked during gameplay. These fluctuations happen so rarely that the impact is only occasionally noticed – we're talking a few times per hour, if that – and only if the user is looking for such fluctuations.
Here is a look at some of the data from the above slice:
Most of this fluctuation is a result of AMD's new dynamic clock, which adapts to pre-defined states in WattMan as workload changes. The rest is resultant of thermals – we're seeing some throttles once temperatures exceed 80~82C, but this can be overridden in Wattman by increasing the max temperature. Of course, fan RPMs can also be increased.
AMD's clock-rate stability is reliable and steady, something for which we commend the company's work on its new process and power management.
And for the curious, here is a chart comparing the fan RPM to the temperatures. Again, this is with all settings stock – so you're looking at out-of-box fan RPM & temperature target performance:
The red line represents temperature (plotted against the right axis, absolute temperature *C). The blue line represents frequency in MHz (plotted against the left axis). The orange line represents fan RPM (plotted against the left axis, in RPM).
In order to sustain the card's ~82C resting area and 1200~1300MHz stock clock-rate (boosted – so it's dynamically fluctuating), the blower fan must spin at approximately 2000~2300RPM. The specific sound of this fan is demonstrated in our video review that's embedded on the first page, but we've also got some objective metrics.
AMD RX 480 Reference Noise Levels / Decibel Output
The below formula is used to determine delta values with decibels:
Calculating dB deltas is not a simple subtraction; decibel addition and subtraction follows a logarithmic function. We extract the ambient noise (~25-27dB) from these system noise results to create linearly comparable dB readings. For more, our noise test methodology is explained on the Test Methodology page.
We're testing for Idle, Auto, 50% Fan, and 100% Fan noise levels. Idle is conducted at the desktop. For “Auto,” we run a burn-in test for 5 minutes and allow the card to ramp into its curve, then monitor fan noise and RPM from there. The AMD RX 480 fan spins maximally at 5200RPM, so our 50% speed setting plants the RPM at 2600, with 100% blasting the fan at its max speed. All extra bench fans are disconnected.
Note that the below chart is representative of total system noise. As our PSU and CLC pump make some amount of fan / vibrational noise, the quietest operation will always hinge on those two devices. Even with a passive video card, we'd still be in the ~37.1dB range.
AMD's RX 480 peak system noise is hitting ~37.3dB when idle, which is about identical to the AIB EVGA & MSI 1070 cards (represented by the same line, as it's the same performance). AMD's fan spins at just ~780RPM when under no load, keeping the card in the 7-8C (dT) range while maintaining the 37.3dB noise output. The stock GTX 1080/1070 Founders Edition cards run about 37.5dB during this test.
After some gaming and initializing our next test, the “auto” fan speed finds its resting point in the 2000~2300RPM range and produces a total system noise of 39.4dB.
50% fan speeds are going to be the most common during drawn-out gaming sessions. We hit 42.8dB on the RX 480, now overtaking the GTX 1080/1070 FE's 41.9dB. AMD's card becomes louder at this point for a few reasons, but the most obvious is raw RPM – AMD's fan just spins faster, which generally pushes higher noise levels.
100% fan speed is utterly unbearable. At this point, the rest of the system noise becomes irrelevant, as it's all drowned-out by the GPU blower fan. You would not ever want to approach this fan speed. In fact, in our endurance OC tests (discussed on P6), we found the ~70-80% fan speed similarly unbearable. The fan had to run at ~3800~4300RPM to sustain our overclocks, something that drove us out of the room a few times for a break from the noise.
Like the GTX 1080 FE in some instances, and AMD's Fury X (though that was a pump whine issue), the RX 480 produces a high frequency output at some RPMs. The high frequency output is describable most accurately as a “whine,” but is mostly detected once passing the 50% RPM fan speed. Until this point – and most gaming use cases will keep the fan at or below 50% RPM – the whine exists in some capacity, but would not be detected by any user keeping the card in a traditional PC case. On the open air bench, it is slightly noticeable, but easily ignored.
Once passing 50% fan speed, the whine becomes significantly more annoying and would be a little maddening during long run-times. This should only be a hurdle for overclockers or users with severe heat issues caused by other parameters, like poor ventilation.
If you want to hear this noise yourself to determine if it is a problem for your use case, we'd recommend checking out our RX 480 noise video that's going live at 10:00AM EST today (June 29). That video will be here.
AMD RX 480 Power Consumption & System VA Draw
NOTE: We're currently rebuilding our power benchmarking methodology and platform. We've been using a Z97 platform for some time now, but have finally migrated to the X99 platform. These cards are the first to be run through the power bench. It is scarcer right now than our Z97 platform was, but the chart will fill-in as we put new cards through it. We will be rebuilding this chart one more time in the near future, but these numbers will give us a performance baseline for the RX 480:
We're looking at VA (apparent power consumption) here. Total system load with the RX 480 (stock clock) sat at 239.76VA, the lowest of the tested cards (to include the GTX 1070). The GTX 1070 FE reports 263.29VA, with the 1080 at ~280.63VA. The GTX 970 SSC has an extra power header and a max board power of ~202W, pushing it past the 1080, 1070, and RX 480 in power draw (289.42VA). Overclocking the RX 480 does see a fairly large power jump, but nothing abnormal.
Continue to the next page for DOOM & Dx12/Vulkan benchmarks.