proud latch
When four or eight keys are pressed simultaneously, these tests show a drastic increase in latency. Is this accurate or due to a faulty testing method?
I assumed the minimum amount of latency between the first and last key to be 0.125ms times the keys pressed. It might be even less if this: “Most of the gamer brand boards I’ve seen can send more than one keystroke per poll.” is true.
Image 1: 60HE.
Image 2: 80HE.
Image 3: Comparison between the two.
fluid lodge
Seems lower than the venom by comparison though
proud latch
Both don’t make any sense to me as it seems 8KHz doesn’t affect this at all.
fluid lodge
Seems like all the HE keyboards I see tested on RTings are pretty much in the same ms range
proud latch
It’d be odd if it was due to their testing method and them still including it but maybe.
left harbor
This test is human driven and can be difficult to press 8 keys at once, especially with adjustable keyboards so it is not surprising to see higher numbers for 8 keys. It is in big part why we score the 4 keys with a more weight than the 8 keys. Keep in mind that we are not pro osu!mania players and it definitely shows 😅 Here is some additional data.
The job, graph and scoring will only use the 20 best samples. The testing methodology requires to generate anywhere between 22 and 30 samples.
For the Geonworks we had 29 samples of the 4keys, and 24 samples for the 8keys.
For the 80HE we had 23 samples of the 4keys and 26 samples of the 8keys.
So yes having to wait at least 1 report between every key does hurt the Venom.
proud latch
This test is human driven and can be difficult to press 8 keys at once, especial...
Thanks for the clarification and it’s nice to see RTINGS here as well!
Is there a reason this test is human driven? A jig for a test like this could help eliminate many of the inconsistencies. Perhaps a solenoid with a plate the size of 4–8 keys? Add conductive contacts on the plate for each key, attach some wires and you can measure the delay between keys when they were pressed as well. The oversimplification of the century of course, but you get it. 😄
I think this is one of the most important tests, especially for many of the newer analog boards, where performance is heavily influenced by the number of keys active at the same time. Most reviewers only do single-key tests, which don’t reflect real-world use at all and are subject to those boards’ manipulative optimized single-key performance.
left harbor
Mostly because not all keyboards are straight so we would need a way to adapt the setup and at that point there's a risk that we provide an unfair comparison if for some reason the "jig" performs better/worse with curved/split keyboards.. That said if this becomes a bigger issue over time we could also just increase the sample from 20-30 to 40-50 and still take the 20 best. I'll also check with the testing team just in case there could be a difference between testers.