Nvidia, the semiconductor designer best known for its graphics processors and, increasingly, ARM-based mobile microprocessors, has unveiled its first 64-bit chip based on the ARMv8 architecture.
The part combines both Nvidia's 192-core Kepler architecture-based graphics processor with a 64-bit "Project Denver" microprocessor, which is fully pin-compatible with the 32-bit Tegra K1. The pin compatibility is intended to help OEMs to get the 64-bit parts to market more quickly and easily.
It continues: "Each of the two Denver cores implements a seven-way super-scalar micro-architecture (up to seven concurrent micro-operations can be executed per clock), and includes a 128 kilobyte (KB) four-way level-one instruction cache, a 64KB four-way level-one data cache, and a two megabyte (MB) 16-way level-two cache, which services both cores.
"Denver implements an innovative process called Dynamic Code Optimisation, which optimises frequently used software routines at runtime into dense, highly tuned microcode-equivalent routines. These are stored in a dedicated, 128MB main-memory-based optimisation cache. After being read into the instruction cache, the optimised micro-operations are executed, re-fetched and executed from the instruction cache as long as needed and capacity allows."
While 64-bit ARM is being touted as a cheaper alternative to Intel on servers and in the data centre, the Denver CPU being touted by Nvidia is intended for mobile devices that need more addressable memory and more power for the next generation of mobile apps.
"For hard-core Android fans, take note that we're already developing the next version of Android - 'L' - on the 64-bit Tegra K1," promises Nvidia.