mwdmeyer wrote on 2021-05-10, 06:26:
I've just started benchmarking some WinChips and are interested to know if anyone had one back in the day and what they thought about them?
I have a range of WinChip 2 (pretty fast!) results that I will post once I have together. Some very interesting things really, it's not a bad chip at all.
I think a lot depends on whether you compare it to clock-for-clock equivalents, or to its period-correct competitors, as it was very late to market for an So7 CPU.
Certainly for ALU-intensive stuff it performed decently, clock-for-clock, and around 1998 FPUs still weren't being utilized a lot for the desktop applications this CPU was targeted at. But if you look at performance vs other things being sold at the same time (K6-2, Celeron), particularly in anything touching FPU, it was very low.
I'm actually looking for a cheap Winchip 1 or 2 to see if I can get one working on a very eccentric So5 board that refuses to boot with any AMD or Cyrix CPUs I've thrown at it so far. The biggest redeeming feature of the Winchip was that it ran at 3.3-3.5v, so could be used as an upgrade for boards without voltage regulators. In that sense the real comparison should be with other CPUs that run on 3.3V. That also explains why the 2A was a complete commercial disaster: it needed VRM anyway, which meant it had to compete with K6-2, M2 and P55C, which it clearly couldn't.