Reverse-engineering the Intel 8087's tangent algorithm: more than CORDIC

The article shows how Intel’s 8087 math coprocessor achieved a major performance leap by using a hybrid tangent algorithm that blended CORDIC-style shifts/adds with polynomial techniques, delivering a tangent in 90 microseconds versus 13,000 on the 8086. For CIOs and technology leaders, the strategic lesson is that hardware-software co-design and algorithmic optimization can create outsized gains in speed and accuracy without brute-force compute, a principle still relevant to modern accelerator, AI, and edge-computing investments.