We can FINALLY Test This!!

[James]

Buy Elgato 4K60 PRO Capture Card (PAID LINK): https://geni.us/e82s
Buy Elgato HD60 S+ Capture Card (PAID LINK): https://geni.us/ySWNx
Buy Logitech G203 Lightsync Mouse (PAID LINK): https://geni.us/YHPfq
Buy AMD Ryzen 9 3900XT CPU (PAID LINK): https://geni.us/QmpH5
Buy Nvidia RTX 2080 Super (PAID LINK): https://geni.us/pGJpr

 

What happens to your gaming experience when you add a capture card, a sound bar, or another video device between your gaming PC and your TV? Let’s find out…

 

 

[Uttamattamakin]

I really like the graphs you prepared for this.  I'm assuming you built actual histograms and fit a gaussian to them  to get those bell curves.  Then you realized that would be too busy for most people  to understand.   It allowed one to understand that the "zero latency" can only mean latency that is within one standard deviation of the mean latency ... they  overlap enough that there is no significant difference between what one sees on  screen and what one hears.   

 

Of course "latency" between sound  and visual is part of life.  The thunder must always follow  the lightning and be out of sync with it. 

[N1ckFl1ghtX]

Probably a stupid question, but wouldn't a higher refresh rate display give better results?

 

The results in the video were odd with everything either being 3ms lag to 19ms lag (one frame as clamed in the video). This makes me thing that the results are limited by the display only being 60hz. If the display was a 300hz display, in my mind, that would result in more accurate numbers than what was shown here as more frames means the measuring device has more data points to look at for a change.

 

I could just also not know what I'm talking about.

[SavageNeo]

26 minutes ago, N1ckFl1ghtX said:

Probably a stupid question, but wouldn't a higher refresh rate display give better results?

 

The results in the video were odd with everything either being 3ms lag to 19ms lag (one frame as clamed in the video). This makes me thing that the results are limited by the display only being 60hz. If the display was a 300hz display, in my mind, that would result in more accurate numbers than what was shown here as more frames means the measuring device has more data points to look at for a change.

 

I could just also not know what I'm talking about.

while the monitor is 60hz, the fps can still be mroe than 60. moee fps is usually better frame times.

[power666]

I have a strong feeling that the testing methodology is a bit flawed near the end as the results appear to be limited by a device in the chain.  Essentially the change in measurable latency is dominated by the one frame latency in the chain:  as long as the 'low latency' devices are before the device that is adding a full frame, they are effectively hidden which the results show.

 

For testing and end user setups like this I'd recommend looking at the EDID tables of each link in the chain to see what they all can support.  Ditto for checking how programmable those EDID tables are as if no changes are being made from the defaults, the noted problems in the video are a reality but they don't necessary have to be.

 

Also this tested only video latency where audio latency follows the same general path plus one more for the sound bar depending on how things are chained together.  Similarly to my point above, variance here exists but if the audio falls within one frame of the video, it is difficult to actively perceive but should be easier to measure the latency difference than video.