Assembly language with the PXA250?

[b9_2mr]

I am interested in developing software for the Axim, or to be more specific, the xScale processor architecture. C++ is fun and all, but I know for a fact that a 200 mhz processor is capable of doing a LOT, and wasting that processing power with a high-level language like C++ or Java is far from efficient IMHO. That's why I'd like to do low-level programming for the Ax, but I can't find many places (well, none so far actually:D) that describe the process of not only developing low-level software, but allowing you to run it on PocketPC2k. If WinCE is anything like it's grandpappy, Windows 3.x/9x/2k/xp, it's gonna take a lot more work than just a simple console program, and there is probably a special way to get a low-level software to run on PPC. Can somebody give me information that might help?

And if you want to suggest that I give up and go with C++, I say: NO! I've seen how pitiful some of the programs run on the Axim that were developed with high-level code (YameCE, HPEmu49, etc.) To think that a 300 mhz processor (the low-end Ax) can't handle a SNES game at full speed is ridiculous, I've seen a 166 mhz pentium run games at FULL SPEED, so obviously, a lot of CPU overhead is being wasted because of Microsoft compilers, and I'm sure that other compilers out there are just as bad. While C++ has it's vast advantages, and maybe even Java (?), it's advantages do NOT include speed. I want to make a lean, quick piece of software, and I need ASM to do it. Heck, if I need to, I'll go down to raw hex coding, if somebody can only help me find out how it's done.