After hinting about it all this week, ARM unveiled a new 64-bit processor architecture at ARM TechCon 2011 today during the keynote speech by CTO Mike Muller, who gave few technical specifics. First, he said, this is an architecture announcement—not an announcement of a core. However, working models of an architectural-level processor are now available to ARM’s closest software developers. Products based on the ARMv8 architecture aren’t expected until 2013. The architecture includes a 64-bit and a 32-bit state so that it can run all existing 32-bit software unchanged. The key features of the current ARMv7 architecture, including TrustZone, virtualization, and the NEON advanced SIMD instructions are maintained or extended in the ARMv8 architecture.
There are two main reasons to jump from 32 to 64 bits in a processor architecture. One reason is to crunch more bits per clock. However, that’s not the main reason—a SIMD unit can do better than a 64-bit general-purpose ALU any day of the week. The real reason is memory space. Even though the ARM Cortex-A15 and newly announced Cortex-A7 have extended 40-bit real memory-address spaces, that’s not what the software sees. Because of the inherent 32-bit architecture and 32-bit addressing model of the ARMv7 architecture, the software still sees a 32-bit subset of the 40-bit address space. A 64-bit architecture breaks the barrier, permitting larger programs, yes, but more important is large data sets.
More details about the ARMv8 architecture are certain to unfold soon.