Like for like: fib(10,000), same algorithm everywhere

Every system proves fib(10,000) mod 7919 by fast doubling, fourteen rounds instead of ten thousand steps. The comparator guests use a hand-written doubling routine; zkFOL is handed the naive two-call recurrence and the compiler rewrites it to the doubling form. A second zkFOL row proves the exact 2,090-digit integer with no modular reduction: the harder claim, included because the reduction is not what buys the result.

Workload: fib(10,000) mod 7919, fast doubling	Prove	Verify	Prover memory
zkFOL, compiler-derived doubling	11.48 ms	2.30 ms	under 10 MB
zkFOL, exact 2,090-digit integer, no reduction	9.98 ms	3.88 ms	under 10 MB
RISC Zero, composite	3.69 s	11.8 ms	312 MB
RISC Zero, succinct	14.73 s	12.4 ms	1.39 GB
SP1, core	13.32 s	74.3 ms	9.39 GB
SP1, compressed	49.24 s	32.9 ms	17.00 GB

Prove-time ratios against the reduced zkFOL row: 321x, 1,280x, 1,160x and 4,290x for the four comparator rows in table order; against the unreduced row, which computes the whole integer while the guests compute a four-digit remainder, 370x to 4,930x. Memory ratios run from roughly 30x to roughly 1,700x. [S, benchmark record]