Computer networks (including the Internet) are inherently big endian. Little endian CPUs, such as Intel/AMD X86 and X86-64, have to flip the bit order when engaging in network communications. That bit flipping obviously works (and is usually performed by the network driver), but it's not totally free in terms of instructions.
ARM and Power CPUs are capable of running in either big endian or little endian mode. When ARM CPUs are deployed primarily for networking-related missions (such as embedded controllers for routers), especially in power-sensitive roles, there's some appeal to running in big endian mode. Hence, Linux (and some other operating systems) are available for ARM's big endian mode. That's the "armeb" flavor of Linux, specifically. Linux for Power always runs in big endian mode. Itanium is also bi-endian and can run in either mode. VMS, for example, runs on Itanium in little endian mode. I was merely pointing out that there are lots of big endian CPUs that are selling very well and that are running Linux in big endian mode, including System z, Power, and ARM. There's no danger that Linux will somehow forget big endian bit order any more than X86 CPUs will forget how to use the Internet. To pick another example, Solaris is available in both little endian (X86-64) and big endian (SPARC) flavors. Not surprisingly, Java is almost entirely endian-agnostic, but to the extent bit order matters it's big endian. I've known HP in its sales pitches to make a lot of fuss about endianness as reason why it would be oh-so-difficult for an HP-UX customer to move to Linux on X86, or for a Linux X86 customer to move to (or add) Linux on System z, depending on their sales situation. Then hundreds/thousands of HP customers moved without endianness difficulty, and many more will follow. The IT community figured out how to flip bit order a long time ago. Before System/360, even. That's not to say endianness isn't a problem...for HP. If they want to move HP-UX to a little endian CPU, they'll have a lot of investment to do (as Sun did for Solaris X86). For non-OS kernel/non-compiler programmers, which is the vast majority of us, it's not a real-world problem. In fact, endianness is one of the least interesting issues when porting from one CPU to another. For my thoughts on the HP Itanium meltdown, see The Mainframe Blog: http://mainframe.typepad.com/blog/2011/04/hp-itaniums-ignominious-demise.html - - - - - Timothy Sipples Resident Enterprise Architect Value Creation & Complex Deals Team IBM Growth Markets (Based in Singapore) E-Mail: timothy.sipp...@us.ibm.com ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
