i actually think impressions of people "feeling" or "not feeling" the lag are way more useful than something like this, which is totally inaccurate and has too many variables to be useful
We need to see performance metrics for non-controlled environments like they have over there at E3. Stadia performance metrics are from real life use cases, so it's not a fair comparison at the moment.
That's a great point! I know the math doesn't work out to needing twice the power for double the frames, but you'd need some serious hardware to make those improvements. Especially when next-gen rolls around. I'm not shooting down the idea, as an engineer myself, I'm deathly curious how they plan on accomplishing this.
You rather people tell you then seeing for yourself? Because no matter how you want to spin it, thats a broken form of reasoning. We all know the experiment in the article is flawed, but we can witness and gather an idea of responsiveness that is much more accurate then impressions, there's a frame of reference that can be repeated and iterated with the exact same outcome.
Yeah. Ping tests to the Azure datacenter from my location show 60ms, then we have controller latency and game latency added to that.
In the article's comments, they say that the servers were running in the Bay Area. Bay Area to LA ping is generally 20ms in my experience (testing from my fibre connection to an LA server) so I don't believe there's only 4ms additional latency.
A show floor is a whole lot different than a controlled environment, and anyone that says it is hasn't dealt with a show floor.
By "local", I assume they're talking about playing directly on the system. I imagine the latency you get from Steam Link is probably higher than what they're advertising here. In my case, that's about twenty feet of hardwired gigabit. I can't directly test latency, though.
Wouldn't the Data travel while a frame you can't impact anymore anyways is rendering and not after? So at 60fps its 16,7ms per frame, so the 2.15ms of time it takes to go to the server would easily fit into the 16,7ms it takes to render a frame. Of course you would have to add the one way back of 2.15ms but not the one way to the server farm. Or am I missing something?
You can't travel during a frame you can't impact anymore. You push frames as fast as you can render them. If you "wait" for in-between, that's adding delay already.
With 60ms ping + controller latency etc, would it even be possible to play a platformer or any other game which require precise inputs?
