My memory, faulty as it may be, seems to recall a two page ad in AW&ST touting the Night Hawk "supercomputer" that would be in each F-22.
In other words, my response to your statement that the avionics of a fighter jet are "computationally outclassed by a $100 smartphone today" is, "So what?" You buy and test the hardware that you believe is necessary to accomplish the given mission, not the latest-and-greatest for the sake of being the latest-and-greatest.
Which is fine because you're right: they don't need them. I'll add the older nodes were more reliable, too, as it was trivial to get chips to work vs deep sub-micron. I'll take old-school, fault-tolerant chip over modern ASIC any day if ny life is at stake.
This processing capacity could be replaced with a couple of high density MILSPEC FPGAs. Its likely they'd use a 64-bit PowerPC processor augmented by FPGAs. The combined solution would be flexible and probably offer 1000x what the F-22 (Block 5 airframe) went into service, at a fraction the cost of the CIP [1]
As for the reference to the $100 smartphone, the latest "$100" smartphones could probably handle MegaPixel cameras for Augmented Reality and with connected sensors, more commonly known as "Sensor Fusion" [2]
Note: A failure of a $100 smartphone doesn't usually result in a pilot pulling the ejection handles on a $185m (unit cost) fighter. Avionics has to meet higher standards. There are MILSPEC versions of some of these processors. FPGAs are available with MILSPEC qualification.
[0] https://en.wikipedia.org/wiki/Intel_i960#i960_variants
[1] http://www.f22fighter.com/avionics.htm#2.1 Common Integrated Processor (CIP)
[2] http://www.f22fighter.com/avionics.htm#4.0 Software
Like you said, though, they can be emulated or replaced with PPC if necessary. I encourage someone to emulate i960 anyway given it was badass enough to deserve at least one FPGA clone. :)
I'm not sure that we do. We did when the JSF project that produced the F-23 was launched, but that was in 1993. More recently, the 6th gen fighter project that would be the follow-on seems to have been abandoned by the air force in favor of a strategy of disaggregating into independent component upgrade projects; for the attack role, it looks like the real replacement for existing manned airframes will be drones (the MQ-9 Reaper seems to be taking that role now, and the air force is training more drone pilots than anything else.)
> Why not make a generation of hyper-advanced missiles? It has got to be cheaper and easier (you don't need to visualize anything, because it is a computer) and you wouldn't even need active lifesupport.
Drones carrying the same type of missiles a manned attack plane would carry are cheaper over time and more flexible in use than "hyper-advanced missiles" that fly the whole mission themselves and are consumed each flight.
Let me fix that: if done well.
Oddly enough, Fairchild Aircraft was formed when Mr Fairchild decided to build a 'decent plane' to carry-around his aerial cameras...
Back to the F-35 it is interesting to contrast how Dassault in France has been slowly expanding the mission scope of its Rafale by adding additional software, without affecting the actual flight-control systems. After 12 years of gradual updates they're also starting to update hardware, such as converting to an SDR instead of traditional tuned radio.
Ya know, I think he may have been onto something there ...
> The A-10 was designed around the 30 mm GAU-8 Avenger rotary cannon that is its primary armament.
https://en.wikipedia.org/wiki/Fairchild_Republic_A-10_Thunde...
Software Development is going to be a competitive advantage in warfare as well. From what I have heard, DoD has some really interesting practices on building software (code review each line with physical printouts, etc), which in combination with scope creep might be the cause of this slowness.