Input latency

I am shocked.

300ms is well past the point online games need god-tier client-side prediction to pretend they're playable. If you wanted to develop games for a Symbolics machine, you'd have to use fucking multiplayer techniques to compensate for your own machine's terrible latency.
Games are a great stress test for operating systems and programming languages.

If you're interested in the sauce, it's up at danluu.com/term-latency/

Attached: libbie_otherside_cropped_rotated.png (331x331, 45.57K)

...

"Muh games" is the reason Unix exists in the first place: a Multics developer named Ken Thompson was making a video game named Space Travel for fun in his spare time, but Multics was too fat to run it properly (to be fair, GECOS ran it even worse) so he switched over to a cheaper PDP-7 and wrote a much better version. Shortly after this he began writing a lighter Multics-like operating system for the PDP-7, ported Space Travel to it, and thus Unix was born.

Figures, Unix is a toy OS.

...

...when people with a clue discuss CPU interrupt latency, "T" means the number of T-states of the instruction decoder.

With the 6502, an instruction was at most 6 T-states (= phi clocks), and when an interrupt occurred, it went through instructions (w/ variable number of T-states on each) to push PC, push S(tatus), and set the PC to (BRKVEC). It was actually the most efficient CPU for doing interrupts on in its time-- the Z80 took something like 56 T-states to do the same thing.

...

Typical. Have some spoonfeeding:
>The next two columns show the clock speed and number of transistors in the processor. Smaller numbers are darker and blue-er. As above, if slower clocked and smaller chips correlated with longer latency, the columns would get darker as we go down the table, but it, if anything, seems to be the other way around.

If your OS and programming language shit themselves when you try making games, there's a good chance they aren't flexible enough for other demanding or unusual tasks either.

Name a task for a workstation that requires sub-300ms input latency in a terminal.