Speculation has been flying that Google’s Pixel 10 comes with a graphics core locked well below its advertised maximum.
The rumour: its PowerVR GPU hasn’t left the ~400MHz mark, far from reaching its 1.1GHz peak. The truth: that takeaway misinterprets the way modern mobile GPUs delegate power and performance. Well, after tracing out the Pixel 10’s hardware in real gameplay, the evidence is in — the GPU isn’t underclocked; it’s pursuing an aggressive, intentional “race to idle” approach instead.
How we tested real-world GPU clocks on Pixel 10
To put some data behind that, we profiled a Pixel 10 Pro XL with Android system tracing (Perfetto) while playing Call of Duty: Mobile’s Battle Royale mode with the 90Hz option enabled. The tale can be read in the GPU frequency and “GPU busy” graphs: longer, constant periods around ~396MHz when facing open terrain, succeeded by rises stepping up their clocks repeatedly to 1.1GHz whenever scenes get more involved — which means plenty of buildings, effects, or firefights.
Intermediate hop frequencies such as ~512MHz, ~576MHz, ~633MHz, etc., do crop up but only for minute time intervals — tiny fractions of a second. In reality, the residency curve is bimodal: a low-power base clock and full-speed boost. Despite that trend, frames were delivered seamlessly during the stress test and gameplay never reflected a downclocked bottleneck.
Why the GPU sits and waits, then sprints on Pixel 10
This is straight from the book: dynamic voltage and frequency scaling. It doesn’t hover at a middling frequency; the governor aims to be efficient: burst and do work with maximum GPU clock, then return quickly back into a cool, low-power state. That’s the principle Arm and Qualcomm outline in their power management recommendations, and it is what Android energy-aware scheduling seeks to enable at a system level.
Imagination Technologies’ newer PowerVR designs can even include hardware ray tracing, but silicon capability isn’t a guarantee that a particular feature — or frequency — runs all the time. Governors, voltage rails, and thermal limits are all conservatively tuned by OEMs with the realities of a slim smartphone chassis in mind. What appears to be “stuck @ 400MHz” is the GPU idling between bursts of real work.
How power and thermals validate this tuning approach
The trace also yielded estimates for power: under sustained boosts, GPU power leapt from around the mid-200-milliwatt range to just below half a watt. That doesn’t even count the CPU, memory, or display. The GPU does not want to hang out at 1.1GHz for yet more seconds; it would heat up quickly and activate thermal throttling — again, probably netting higher long-term frame time. After about 10 minutes of non-stop gaming, the phone was warm to the touch — just as you’d expect when a governor is doing short sprints to maintain temperature.
This is in line with the directive from Android Thermal HAL, and also typical OEM tuning: prefer to be responsive, otherwise back off. Running flat-out is rarely efficient in a phone unless a workload literally requires power at all times.
Benchmarks, drivers, and misread telemetry explained
So why the underclock claims? A few culprits. Some apps poll frequency levels instead of GPU busy time, and if they sample when the board is idle, they’ll “see” ~396MHz and think there’s a cap. Others misunderstand the governor’s wish for a base state as evidence that there is a cap. Whether the instantaneous clock readout is 7 percent off isn’t nearly as important in controlled runs of commonly used tests like 3DMark Wild Life and GFXBench as frame pacing and thermal consistency. Those remained steady even as the GPU bounced in and out of base and boost states.
As for drivers, yes, vendor stacks and compiler paths can affect efficiency or feature exposure. The Pixel 10 doesn’t enable hardware ray tracing in in-game configuration, but Imagination’s documentation supports it for their latest IP. That’s policy and software, not an underclock. Frequency availability and residency are controlled by the DVFS tables and scheduler, both of which permit 1.1GHz when necessary for workloads.
How it stacks up against competing mobile GPUs
This approach isn’t unique. Qualcomm’s Adrenos and Arm’s Malis lean on flinging things into high gear and then coasting back under mobile thermal budgets, as seen from vendor white papers or AOSP docs around power hints. The Pixel 10’s pattern just feels more binary — low or max — versus some competitors that hang around at mid bins longer. That doesn’t mean it has less performance headroom; it means a different power curve and tuning philosophy.
Bottom line: what the Pixel 10 GPU tests show
The Pixel 10’s GPU is not underclocked. On-device traces have it hitting the maximum 1.1GHz under load over and over again, then simply dropping down to a ~396MHz baseline to sip power. That bimodal behavior is intentional, contemporary, and efficient. If anything, the scheduling approach helps keep the phone from cooking itself while delivering smooth gaming — which is exactly what you want out of a flagship tuned to real life versus lab fantasies.
For readers interested in diving deeper, the Android Open Source Project’s Perfetto and power management documentation, vendor power-efficiency briefs from Arm and Qualcomm, and Imagination Technologies’ public architecture notes offer valuable context on parsing GPU clocks, residency, and scheduler decisions on mobile devices.