flood's input lag measurements

[krp]

Can you test if Hyper-threading on/off affects input lag in CSGO? Thanks.

[Sparky]

Can you test if Hyper-threading on/off affects input lag in CSGO? Thanks.

That would be pretty hard to test, if there's a significant difference it would be in a situation where hyperthreading makes you go from CPU limited to GPU limited or vice versa. Do you know a situation like that?

[flood]

Can you test if Hyper-threading on/off affects input lag in CSGO? Thanks.

yes i can but convince me why it matters at all.

i can also test whether using a gold-plated usb cable affects input lag, but obviously i wouldn't

[krp]

Can you test if Hyper-threading on/off affects input lag in CSGO? Thanks.

yes i can but convince me why it matters at all.

i can also test whether using a gold-plated usb cable affects input lag, but obviously i wouldn't

From what I have been reading HT ON may cause lower FPS in some games. Essentially I'm just wondering if it could also affect input lag and whether one should turn it off for competetive gaming.

[Sparky]

Can you test if Hyper-threading on/off affects input lag in CSGO? Thanks.

yes i can but convince me why it matters at all.

i can also test whether using a gold-plated usb cable affects input lag, but obviously i wouldn't

From what I have been reading HT ON may cause lower FPS in some games. Essentially I'm just wondering if it could also affect input lag and whether one should turn it off for competetive gaming.

I remember that being an issue when HT was very new, due to the OS not knowing how to schedule tasks correctly on the virtual cores, but I haven't heard about any recent problems with hyperthreading. Even if there is a difference in some game, the "which is better" answer probably depends on OS version, number of real cpu cores, and on whether your GPU is outpacing your CPU in that game.

So, Which games are impacted by hyperthreading in terms of FPS? If there's a difference in any game, that would be where to look first.

[flood]

Can you test if Hyper-threading on/off affects input lag in CSGO? Thanks.

yes i can but convince me why it matters at all.

i can also test whether using a gold-plated usb cable affects input lag, but obviously i wouldn't

From what I have been reading HT ON may cause lower FPS in some games. Essentially I'm just wondering if it could also affect input lag and whether one should turn it off for competetive gaming.

as long as it's not something that
1. affects fps
2. dramatically changes the rendering or input behavior at a high level (e.g. windows 8 originally forcing 125hz mouse, vsync, sli with alternate frame rendering, etc...)
you really shouldn't worry.

if turning ht off/on increases fps, then do it, but if it's only a 2% change then who cares

[flood]

btw for csgo i have an empty dedi server you guys can use for testing if you want. pm me for info

[stirner]

flood you have a NVIDIA card right - care to test latency with the nvsvc service enabled vs. disabled?
the service launches nvvsvc.exe and nvxdsync.exe, so make sure those are not present with it disabled.
profiles are still applied without them (used inspector frame limiter) so I don't know how to convince you except that it feels less responsive without them running.
not much information about them out there, but apparently they do various things...

[flood]

yea nvidia. ill probably get around to it and everything else at some point.

there's something about my circuit that i don't quite understand

basically i have a program that measures the display's refresh period

Code: Select all

		while(!PD_READ);
		a = mytime_get_count(); //time in units of 0.5 us
		while(PD_READ);
		b = mytime_get_count();
		print_ul((b-a)/2); //x in the diagram below
		usb_debug_putchar('\t');
		print_ul((a-c)/2); //T in the diagram below
		usb_debug_putchar('\n');
		c = a;

when i display a white stripe on a black background, the crt's light emission is basically a rectangle wave convolved with the phosphor's decay function.
i.e.


the program outputs x and then T every time it detects a pulse.
for thin stripes, the output is extremely stable. for instance for a single pixel thick stripe (monitor @ 800x600@160hz)

as i make the stripes thicker (by scrolling in my program), initially the left number (x) increases and the right number (T) stay steady, as expected since the program triggers on the rising edge which is well defined since the phosphor's rise time is essentially instantaneous
the weird thing is that once x gets to the ~3800, it stops increasing, and T becomes unstable...


which makes no sense because even if there is noise the timing of the rising edges should still be consistent...

here's a paint drawing of the circuit.


pretty sure it's something to do with the behavior of the op amp(s) when they're saturated, but i don't understand what causes this

[Sparky]

Vertical blanking interval? What refresh rate are you using?