Laptop variant of Arm Cortex-A78 supports configurations with eight ‘big’ cores
Arm unveils its Cortex-A78C cores, which are based on the regular A78 cores. The A78C is intended for laptops, among other things. The A78C also supports clusters with up to eight ‘big’ cores; manufacturers do not need to incorporate low-cost cores into their A78C socs.
The Cortex-A78C cores can be used in socs for on-the-go devices, according to Arm. The company cites, for example, laptops as a usage scenario. Arm is basing the new cores on his current Cortex-A78, which was announced in May. The company is, however, making various changes to the A78C variant. For example, the maximum amount of L3 cache at the A78C is increased to 8MB. With the regular A78 cores, this was a maximum of 4MB.
Most striking, however, are the supported core configurations. Arm’s soc designs use big.Little configurations, providing a chip with some powerful cores and some energy-efficient cores. Current A78 cores support configurations with up to four “large” cores and four low-power cores.
The A78C also allows manufacturers to process clusters of up to eight ‘big’ cores in their socs. This means that manufacturers no longer have to process energy-efficient cores in their chips. Arm speaks of higher performance in multithreaded workloads. The company does not yet mention clock speeds for its A78C cores. The regular A78 series has an ‘optimal clock speed’ of 3GHz and is made on a 5nm process.
The cores also receive support for the Mali-G78 GPU, just like the regular A78 series. Arm also incorporates some security functions into the A78C. The company calls, among other things, Pointer Authentication. According to Arm, this should ‘minimize the attack surface of the chips’. For example, the company claims that the cores are less prone to Return-Oriented-Programming exploits and Jump-Oriented-Programming exploits. It is not yet clear when the first devices based on the A78C cores will appear.
The Arm Cortex-A78C supports configurations with up to eight ‘big’ cores. Image via Arm