Re:iTerm2 3.3
by mikeash on Thursday August 01, 2019 @04:47PM
(#20588193)
Measuring the time between a simulated keypress and pixels changing in the framebuffer seems kind of inadequate, since it’s missing out on delays in the input and display hardware.
[ Parent
| on HN ]
The hardware should be the exact same between all terminals right? It seems accounting for that would only add noise.
In many cases the display lag can be one of the biggest sources of input lag, often dwarfing the sub 20-30 ms lag times the earlier post measured across iTerm and it’s alternatives.
Surely that would affect all the applications being compared equally, though?
Not necessarily. For example, if one app synchronizes with screen refresh and another does not, the latter will perform much better on this test but it’s not actually getting the changes to your eyes any faster.
It would also just be nice to know the magnitudes. If the hardware adds a constant 100ms (for example, not a real number) then a 10ms advantage on the software side looks less interesting.