New device: GILT Game Input Lag Tester

[Q83Ia7ta]

I found a V1 for cheap. Will buy and measure the latencies again properly.
Btw I'm thinking about adding a button to my tester and the LED. In the software you will have an option to assign a key stroke or a mouse button, so anyone curious about it's accuracy will be able to test it using a camera. What do you think about that additional idea?

btw if I understand your method right then your software needs to analyze results with V-Sync OFF since frame draw starts at random part of screen with high FPS.

w3.jpg (90.14 KiB) Viewed 6879 times

With camera you can capture entire screen to view drawing of new frame at any part of screen.
How I did it:
regular mode:
https://www.youtube.com/watch?v=hFpnim0N6zE
benq blur reduction mode:
https://www.youtube.com/watch?v=2K6q3zBxhUU

[Oomek]

With vsync off i draw to the front buffer and analyze many frames and select those with the smaller latency. That's why the accuracy is not 0.1ms like with the vsync on. I still limit the framerate to the monitor's refresh. I bought myself that 1200fps camera so I will post my measurements soon.

[Q83Ia7ta]

LED and resistor choice shouldn't matter much. What do you mean by not lighting up immediately? As in you can see physical movement of the button before the thing lights up? That would be caused by a combination of switch actuation point, the time it takes the switch contact to fly across the gap, and possibly bouncing, depending on how long your shutter is open.

LED lights up to full state in more than one frame (0,833 ms).
Let me show you:
start of LED lighting up (condition A):

a.png (14.65 KiB) Viewed 6878 times

the next frame - LED ok (condition B):

b.png (17.84 KiB) Viewed 6878 times

As you can see little red dot at "condition A" is LED starting to lights up.
iirc most of time with my setup there is only "condition B" but sometimes "condition A" appears.
You can see videos at my prev post.

[Oomek]

Luckily my device when used as a part of the camera test has the led controlled by software, so the problems with switch spuroius on/off transitions will not occur.

[Oomek]

Btw, I'm a skilled shader hacker, so I can make any game to flash the screen on keypress using a simple shader injection if you still insist on in-game testing.

[Q83Ia7ta]

Btw, I'm a skilled shader hacker, so I can make any game to flash the screen on keypress using a simple shader injection if you still insist on in-game testing.

What if I play quake?

[Oomek]

Btw, I'm a skilled shader hacker, so I can make any game to flash the screen on keypress using a simple shader injection if you still insist on in-game testing.

What if I play quake?

Um, sorry that "any" was a little exaggeration dx11,12 only dx9 also, but it requires messing with asm and cannot be universal.

I had no idea though that people are still playing Quake. I'm a retro lover myself, but Quake for me is not oldschool enough

[Sparky]

LED and resistor choice shouldn't matter much. What do you mean by not lighting up immediately? As in you can see physical movement of the button before the thing lights up? That would be caused by a combination of switch actuation point, the time it takes the switch contact to fly across the gap, and possibly bouncing, depending on how long your shutter is open.

LED lights up to full state in more than one frame (0,833 ms).
Let me show you:
start of LED lighting up (condition A):

a.png

the next frame - LED ok (condition B):

b.png

As you can see little red dot at "condition A" is LED starting to lights up.
iirc most of time with my setup there is only "condition B" but sometimes "condition A" appears.
You can see videos at my prev post.

That's probably the result of a bouncing switch, or the LED turning on just at the very very end of the camera's exposure, If the mouse uses double throw switches you can use an SR latch to drive the LED, and eliminate any bounce issues.

[Oomek]

GILT WIP v0.4b

[Q83Ia7ta]

It looks neat. Is black to white transition time used?