AIX is full of "hacks" or "modifications" in the TCP/IP stack to greatly improve the performance on POWER architecture on MP systems. Have any of these made it into mainstream Linux? Are they even valid on Intel architecture?

For instance, when running a benchmark on an AIX POWER system, try increasing the load, and see if your results go up. It can happen, that you increase the load, the CPU utilization climbs, but you benchmark remains the same. Well, you might be hanging in spin locks. AIX supports instrumented locks, so you can check this with the lockstat command.

Another potential problem is that two many global variables are located in the same CPU cache line. So you can pad single variables, so that they are in separate cache lines. Or, even worse, you have one global variable that is being constantly updated by all processors, and is constantly causing cache invalidation on the memory bus. Then you need to do a hardware memory bus trace, with an HP logic analyzer that looks like something out of Hentai Porn. Then you need to write up a patent or something:

https://patents.justia.com/patent/6430659

So I'm just wondering if all this poop will be done for Linux on OpenPOWER . . .

Yeah, but Linux on POWER today runs on Big Endian. See another post in this thread about IBM intentions, but Red Hat has not announced support for Linux on Little Endian yet. That one hurts.

Linux on OpenPOWER doesn't exist yet . . . or does it . . . ?

AIX was pretty cool way back when, when they introduced 64-bit support. The processor was 64-bit. However, you could run a 32-bit kernel or a 64-bit kernel. And you could run a 32-bit process or a 64-bit process on either of the kernels.

So what does some poor chump (i.e. me) who is tasked with writing a device driver for AIX need to do? Well, first #ifdef the code, so you compile different stuff, depending on if you are building a 32-bit or 64-bit version of the device driver. Then you needed to add simple "if" statements in the device driver, to check if you were running a process in 32-bit or 64-bit mode. Then according to the mix, you would have to thunk the addresses, when copying the data from user space into kernel space.

Fun stuff. I can't believe that I actually did this in a former life . . .

Unfortunately, if there is one thing that the IT Industry does not lack today . . . is a shortage of morons.

But, alas, I always seem to get called in to debug code from other folks.

Perfect, for the next "Occupy" demo crowd.

Well, if you have a shotgun that you don't care much about, you can buy some "Dragon's Breath" shells for it: http://en.wikipedia.org/wiki/D...

A simple recompile won't do it . . . some programs, like your TCP/IP stack have dependencies on Endianess. IBM's POWER has been traditionally Big Endian. Linux is mostly Little Endian. There are C macros, ntoh() and hton() that do the required byte swapping for you . . . if you remember to use them! I have seen code that would run fine on a Big Endian machine . . . but would fall over and die on a Little Endian machine.

Sorting out all these problems is painful grunt work. Although, at one point, IBM tried to port AIX to Intel64. It was called Monterrey, or something like that. So IBM does have experience in doing this. It's just a question if they will fork out the dollars required to do this. And if the people know how to do this have been laid off.

No, but he should be forced to dress like a woman, on odd or even days, depending on whether his birthday is odd or even.

That is my simple and effective cure for sexual diversity in IT. Half the time men will be forced to dress as women, and women be forced to dress like men. Hey, presto, when someone from the government comes to do a headcount, he or she will find an equal number of men and women. Problem solved.

Jesse Jackson can be placated by having a whites wear black face and blacks wear white face policy, as well.

The most fun would be the Native American Indian role: You get to turn your cubicle into a steamy, smoky sweat box, while consuming hallucinogenic Peyote.

So, is IBM going to ditch making their own POWER pSeries, and totally go for the ARM model of just licensing the technology for OpenPOWER . . . ?

Just like in the PC world, folks stopped buying IBM built PCs, when cheap clones were available. What would be the advantage of buying an IBM built OpenPOWER system, as opposed to a much cheaper Chinese built clone . . . ? Maybe the IBM system will have some kind of "secret sauce" . . . ? Like a MicroChannel (har, har).

At any rate, somebody is going to have to invest a lot of money to make sure that Linux runs well on OpenPOWER, in order for this to succeed.

. . . and if anyone wants to file an editor gender suit . . . Bill Joy wrote vi . . .

That one would be a hoot and a half in court. None of the jurors would really understand what it was all about, and the court case would be finished, before emacs loaded.

There is an antivacuum in the universe . . . more specifically, in my apartment.

At least that is what my cleaning girl claims . . . when she tries to vacuum here . . .