Comment by lysace

1 day ago

USB and VM latencies are both questionable here, I would think.

ISA motherboards still work and have been mass produced in the hundreds of millions. (I have about a dozen in my retro closet.)

The esoteric ISA expansion cards would be the bottleneck, not the motherboards.

All those layers add some latency, but 8 mhz is really slow. USB 4 can do extremely low latency, otherwise PCIe over USB wouldn't work.

  • The problem is that a lot of those old ISA drivers were peeking and poking card registers as fast as the bus would allow, which was over 1 MHz (less than 1 microsecond per transaction), not DMAing data to RAM (though some did that too).

    With a modern 4 GHz CPU, that means your budget is only 4000 clocks per interrupt. The context save alone can make it hard to hit 1MHz, and you really want to have your interrupt code pinned to the L1 cache because a few memory reads will blow your entire budget.

    As nice as USB4 is, I’ve never heard anyone claim it has single-digit microsecond latency, to say nothing of sub-microsecond latency. The typically-quoted ~20 usec latency of USB4 limits it to around 50 kHz for an emulated ISA card.

    • Often people confuse latency vs throughput/bandwidth. With streaming and batching USB4 will shuffle a lot data through on average per second. But it will never hit reliable microsecond level or even submicrosecond latencies.

      At the same time ISA card will never reach 40Gbps throughout with CPU clocking at tens of MHz only or often less.