Question Interesting input latency comparison

Toggle sidebar Toggle sidebar
D

Doug S

Diamond Member
Feb 8, 2020
3,451
6,124
136
I saw a link to this and while it is a bit out of date I don't think the situation in 2025 would be all that different from the 2016-2017. As operating systems have become more complex with more software layers between "keypress" and "character appears on screen" latency inevitably increases - and to a greater degree than CPUs get faster.

You can see a clear progression from the really fast - Apple II is kinda cheating because 1) it ran a busy loop waiting for input with no provision for multitasking and 2) the OS supported character generation so it wasn't dealing with fonts and all the bitmap overhead. To the medium fast - SGI Indy was multitasking and used proper fonts but the "terminal" ran on a bare X server. To the more recent "slow", like the Powerspec G405 from 2017 - which was dog slow whether you ran Linux or Windows despite a 4.2 GHz clock! Both have all the overhead of window manager, on top of display servers, on top of graphics drivers, on top of framebuffer management, and that's just on the display side!

So if you're Gen X like me and have this sense that computers seemed to have faster response times when you were a kid despite the phone in your pocket being more powerful in every way than the fastest supercomputer in the world in 1985, you're not imagining things!


computerlatency
(ms)		yearclock# T
apple 2e3019831 MHz3.5k
ti 99/4a4019813 MHz8k
custom haswell-e 165Hz5020143.5 GHz2G
commodore pet 40166019771 MHz3.5k
sgi indy601993.1 GHz1.2M
custom haswell-e 120Hz6020143.5 GHz2G
thinkpad 13 chromeos7020172.3 GHz1G
imac g4 os 9702002.8 GHz11M
custom haswell-e 60Hz8020143.5 GHz2G
mac color classic90199316 MHz273k
powerspec g405 linux 60Hz9020174.2 GHz2G
macbook pro 201410020142.6 GHz700M
thinkpad 13 linux chroot10020172.3 GHz1G
lenovo x1 carbon 4g linux11020162.6 GHz1G
imac g4 os x1202002.8 GHz11M
custom haswell-e 24Hz14020143.5 GHz2G
lenovo x1 carbon 4g win15020162.6 GHz1G
next cube150198825 MHz1.2M
powerspec g405 linux17020174.2 GHz2G
packet around the world190
powerspec g405 win20020174.2 GHz2G
symbolics 362030019865 MHz390k



devicelatency
(ms)		year
ipad pro 10.5" pencil302017
ipad pro 10.5"702017
iphone 4s702011
iphone 6s702015
iphone 3gs702009
iphone x802017
iphone 8802017
iphone 7802016
iphone 6802014
gameboy color801998
iphone 5902012
blackberry q101002013
huawei honor 81102016
google pixel 2 xl1102017
galaxy s71202016
galaxy note 31202016
moto x1202013
nexus 5x1202015
oneplus 3t1302016
blackberry key one1302017
moto e (2g)1402015
moto g4 play1402017
moto g4 plus1402016
google pixel1402016
samsung galaxy avant1502014
asus zenfone3 max1502016
sony xperia z5 compact1502015
htc one m41602013
galaxy s4 mini1702013
lg k41802016
packet190
htc rezound2402011
palm pilot 10004901996
kindle oasis 25702017
kindle paperwhite 36302015
kindle 48602011
 
  • Like
Reactions: Nothingness, Hitman928, Elfear and 2 others
J

johnsonwax

Senior member
Jun 27, 2024
338
505
96
Apple Vision Pro passthrough latency (camera to screen) of 11ms sort of in this category.

But I think this is a great trend to equalize multiplayer online games. Playing on 100 ping is no big deal if your input latency is 50% higher than that.

But as an old who learned to code in the 70s, before all your fancy frame buffers and the like, you usually needed to get everything done within a single NTSC frame, and do all of your drawing during VBLANK. It was a very tight loop and you generally didn't have a lot of logic outside of that loop, and until modern operating systems you weren't competing with any other processes on the system - you owned it. Ever since with some reversion with iPhone, the goal has been to further and further abstract from the hardware mainly because a little bit of latency is worth adding multitasking, or multi-language support, and so on. iPhone yanked that back somewhat toward a 1 process system, but also Apple focused a LOT on scroll latency and the sensation of direct interaction - so the screen scrolled pixel for pixel with the user input. And I suspect some of the above data is incorrect because I know that the iPhone 5 had better scroll latency than the 4 (each iPhone improved upon that in the early iPhone era) and was closer to 55ms and the 6 is when it started to get inconsistent - I think up to 60ms and that varied a bit from then out.

It's reported that the pencil on 120hz iPad Pros is 9ms.
 
  • Like
Reactions: 511
Quintessa

Quintessa

Member
Jun 23, 2025
87
55
51
Doug S said:
So if you're Gen X like me and have this sense that computers seemed to have faster response times when you were a kid despite the phone in your pocket being more powerful in every way than the fastest supercomputer in the world in 1985, you're not imagining things!
Click to expand...
Modern stacks add tons of overhead compared to those single-task machines. Input now goes through USB polling, kernel drivers, compositor, display server, GPU driver, rendering pipelines, then finally the app. Each layer adds a few ms. Back in the Apple II or early X11 days, it was "keypress > write char to memory-mapped screen" with almost no buffering.

What's interesting is how many "performance" devices today deliberately add latency: wireless keyboards with extra firmware, compositing window managers with frame pacing, V-sync/VRR smoothing. It’s all aimed at visual polish, not raw responsiveness.
 
  • Like
Reactions: pugs, Tlh97, Elfear and 1 other person
D

Doug S

Diamond Member
Feb 8, 2020
3,451
6,124
136
johnsonwax said:
Apple Vision Pro passthrough latency (camera to screen) of 11ms sort of in this category.

But I think this is a great trend to equalize multiplayer online games. Playing on 100 ping is no big deal if your input latency is 50% higher than that.

But as an old who learned to code in the 70s, before all your fancy frame buffers and the like, you usually needed to get everything done within a single NTSC frame, and do all of your drawing during VBLANK. It was a very tight loop and you generally didn't have a lot of logic outside of that loop, and until modern operating systems you weren't competing with any other processes on the system - you owned it. Ever since with some reversion with iPhone, the goal has been to further and further abstract from the hardware mainly because a little bit of latency is worth adding multitasking, or multi-language support, and so on. iPhone yanked that back somewhat toward a 1 process system, but also Apple focused a LOT on scroll latency and the sensation of direct interaction - so the screen scrolled pixel for pixel with the user input. And I suspect some of the above data is incorrect because I know that the iPhone 5 had better scroll latency than the 4 (each iPhone improved upon that in the early iPhone era) and was closer to 55ms and the 6 is when it started to get inconsistent - I think up to 60ms and that varied a bit from then out.

It's reported that the pencil on 120hz iPad Pros is 9ms.
Click to expand...

The 11ms and 9ms you're mentioning there aren't measuring the same thing. The link explains the exact testing methodology, any numbers derived in a different way are not comparable.
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom