Microsoft has started to talk in detail about Windows on ARM: what it will do, what it won’t do, and how it has been put together in its latest post on its Building Windows 8 blog. The focus of the lengthy post is the work Microsoft has done to bring Windows to ARM: building a common ARM platform that works the same way, whether using a processor from NVIDIA, Texas Instruments, or Qualcomm. This was a substantial undertaking: unlike desktop PCs, ARM systems are all wildly different.
Windows on ARM, or “WOA” as Microsoft is calling it, is substantially identical to Windows on x86/x64. The difference between ARM systems and x86 systems is more than just the instruction set of the processor. x86/x64 systems are almost all built in the same way. The system uses BIOS or UEFI to initialize hardware and hand over control to the operating system, they use ACPI for enumerating hardware and power management, major system devices like the video card and storage controllers are connected to PCI or PCIe with other peripherals attached to USB, and storage is either SATA or SCSI.
This idea of a common system platform dates back to the original IBM PC: being PC-compatible meant fitting into IBM’s idea of how the system’s components should be interconnected. The connections and technology have changed (with the BIOS being the lone hold-out against progress, though even that is finally making way for UEFI), but the basic concept of a standard platform has not.
ARM has no standard platform. ARM system-on-chips might attach devices with USB or PCIe, but they might not; those devices might be hardwired to particularly memory addresses, or they might use low-power and/or low-speed busses rarely found on regular PCs such as I2C or SD I/O. They might use SATA for their mass storage, but they often use eMMC. (Incidentally, some of Intel’s x86 SoCs also discard x86 platform conventions; as a result, they can’t currently run Windows.)
Loading comments...