Hacker News Reader

Comment by muxator | Hacker News Reader

What was the power consumption of these chips?

More interesting yet it would be to know the Power/speed ratio compared to modern chips.

Interesting question - it’s not trivial to find a metric that’s directly comparable between a 486-era chip and a modern x86 chip. MHz isn’t the whole story, as modern chips have multiple cores and greatly improved instructions-per-cycle. Synthetic benchmarks like Whetstone might provide a point of comparison, but it’s hard to be sure the different benchmarkers were running the same version.

Wikipedia says the 386SLC from the article drew 2.5W at 25Mhz. A Ryzen mobile CPU draws something like 15W. My gut says the Ryzen is a lot more than 6x faster.

The 386SLC had 815,000 transistors manufactured on a 1.2-micron process (1,200 nm).

The Ryzen 7040 has 25 billion transistors manufactured on a 4nm process.

That’s 30,000x more transistors on a much more power efficient process. There’s no real-world task where the Ryzen won’t perform many orders of magnitude better. Anything involving floating point must be millions of times faster because the 386SLC didn’t have an FPU at all whereas the Ryzen has wide SIMD and an integrated GPU.

Wouldn’t surprise me if the 386 could respond to interrupts faster.

I cycles, yes, in absolute time, no.

What about a 6502?

You might be surprised if disable Ryzen’s cache. A lot transistors are dedicated to cache.

> MHz isn’t the whole story

MHz has never been any part of the story.

Reply on news.ycombinator.com