The timeline for fast multiplies is also a bit off, IMUL was already 4c latency / 1c throughput on the Pentium Pro (1995) and Pentium II (1997). Pretty sure IMUL was also only a few cycles on the AMD K6 (1997). Though Intel slowed it back down again in the Pentium 4, until they reverted back to the P6 architecture in the Pentium M.
Comments
The timeline for fast multiplies is also a bit off, IMUL was already 4c latency / 1c throughput on the Pentium Pro (1995) and Pentium II (1997). Pretty sure IMUL was also only a few cycles on the AMD K6 (1997). Though Intel slowed it back down again in the Pentium 4, until they reverted back to the P6 architecture in the Pentium M.