(later in the day). New challenge. The current P-256 point multiplier with TFM clocks in at 2.75M cycles on my AMDX2 box. I want to beat 1M cycles before I release (using less than 32KB of tables). I think it’s doable as fixed point alone can get me a 2.6x improvement without changing anything (using less than 16KB of memory). Stay tuned. Expect numbers within the next week or two.