Verifies 9to5Mac findings: CPU uplift strictly within 22–27%, GPU within 40–45% in standard target envelopes.
Interactive Silicon Performance Analyzer • iPhone 18 Pro vs A19 Pro
Verifies 9to5Mac findings: CPU uplift strictly within 22–27%, GPU within 40–45% in standard target envelopes.
A one-minute painted explainer of how this simulator turns a phone chip's power budget, thread count and workload into CPU and GPU gains over last year's chip.
Every new phone chip arrives with a promise: faster than last year. This simulator checks the A twenty Pro against the A nineteen Pro, inside the power limits of a phone. Its anchor is one reported test point. At six and a half watts, eight threads and a mixed workload, the CPU is twenty-four point five percent faster and the GPU forty-two point five percent. The tool turns those gains into scores: about three thousand nine hundred single-core in Geekbench, eleven thousand two hundred fifty multi-core, and one hundred eighteen frames per second in a graphics test, up from eighty-three. Power is the main dial. Each watt moves the CPU gain by about point eight of a point and the GPU gain by one point two, since a tighter power budget means lower sustained clocks. Threads matter too. Fewer cores trim the CPU gain and scale multi-core down in proportion. Squeeze it to three watts and one thread, and the claim check flips to throttled. These are the tool's own scaling rules, not lab measurements. Slide the thermal envelope, switch to 3D Metal, and see how much headroom each claim really has.
At 6.5 W, 8 threads and Mixed, the tool returns fixed values: +24.5% CPU, +42.5% GPU, Geekbench 6 single 3,890 and multi 11,250, and 118 FPS. Anywhere else it applies simple linear rules: about 0.8 CPU points and 1.2 GPU points per watt, 0.3 CPU points per thread, clamped to 16–32% and 28–52%.
Phones have no fans, so a chip can only hold its top clocks while it stays within a thermal budget. Lower the budget and the operating system lowers frequency and voltage (dynamic frequency scaling), which is why sustained benchmarks fall below short bursts.
Geekbench single-core measures one thread; multi-core runs work across all cores. The tool scales multi-core linearly with threads from an assumed A19 Pro baseline of 9,000 at 8 threads, and single-core from an assumed 3,400. Compare those baselines with published Geekbench Browser results before relying on them.
The badge reads 'Throttled below stage claims' only when the CPU gain drops under 20% or the GPU gain under 35%. With the slider ranges available, that happens only at very low power with few threads (for example 3 W and 1 thread gives +19.6% CPU).
Background on the A19 and A19 Pro, the baseline chip the simulator compares against.
The benchmark whose single-core and multi-core scores the tool estimates.
Why lower power budgets mean lower sustained clock speeds.
The idea of a power and heat envelope that the Thermal Envelope slider represents.
Apple's graphics API, used by the 3D Metal workload profile.
The numbers come from this simulator's built-in calibration and linear scaling rules; they are not measured benchmark results for any shipping device.