Try the LPC2000 series of microcontrollers.
A fast CPU clock rate does not necessarily end up meaning the ability to examine (or output to) port pins at high rates. Some of the ARM chips in particular have been dissed as having particularly poor "pin toggle" capabilities.
I you're just COUNTING pulses, there are other ways to get high speeds (a high speed counter is "easy" external logic), and there are also the external clock input to the timer/counter units.
You should be able to do better than 7us pulses even on an Arduino, if you use direct port access rather than the digitalRead() functions. But less than about 2us would probably be tough.
There was a series of microcontrollers (Scenix/Ubicom) that explicitly attacked the "fast pin access" problem, with the aim of doing away with standard peripherals and implementing everything in software (IIRC, they ran 80 to 100MHz.) They failed to survive :-(
An FPGA or CPLD might be applicable as well.