For smartphone makers that insist on twice-a-year upgrades, MediaTek and Qualcomm typically release a midyear kicker for their premium processors. These refreshes usually offer faster CPU and GPU clocks while maintaining the same features. Both MediaTek’s new Dimensity 9000+ and Qualcomm’s Snapdragon 8+ Gen 1 take this approach. But the latter company goes one step further with a manufacturing-process change that also cuts power. Both new parts should appear in phones in the third quarter.

The D9000+ carries forward the same three-level CPU subsystem, 10-core GPU, and 6-core deep-learning accelerator (DLA) of the original version. It boosts the top speed of the high-end Cortex-X2 CPU by 5% to 3.2GHz, yielding an equivalent gain in single-core performance; multicore rises by about 2%, as the remaining CPUs run at the same speed. The company touts a 10% GPU-performance increase, which implies a commensurate rise from the previous peak frequency of 850MHz.

The Snapdragon 8+ offers the same CPU configuration as the D9000+ and raises the Cortex-X2 frequency from 3.0GHz to 3.2GHz. Unlike MediaTek, Qualcomm boosted the speed of the mid- and low-tier CPUs as well. The result is about 6% better single-core performance and 10% better multicore performance. The company also increased GPU speed by 10% to 900MHz, producing a similar graphics-benchmark gain.

The 8+ also jumps fabs, moving from Samsung 4nm to TSMC 4nm. That change may seem small, but like the D9000, the original Snapdragon 8 used a 4nm-in-name-only process. Thus, a true 4nm process from TSMC provides significant benefits: up to 30% less power, according to Qualcomm.