Large compiled applications demand more, and the translations and associated tables are typically several times bigger than the code they replace. When you're running tiny benchmark loops, a small translation buffer will suffice. If you can afford to allocate megabytes you'll probably find the emulation faster because it can spend more time running pre-translated code, and less re-compiling sections. SheepShaver - WikipediaĪt least K is needed. The experimental compiling emulator needs more memory to store the x86 code it generates as it encounters and translates blocks of original 68K instructions.
One is a conventional interpreter, and runs on any processor supported by GCC, though as with all interpreting emulators the overhead is large - typically 50 host instructions for every one emulated.