I hope Onlive succeeds (they really have some amazing technology) but I would really hate to have my fate tied to the fickle finger of the big content publishers. They've already had issues with EA (Mass Effect and Assasins Creed have been pulled).
Their current library emphasizes shooters when Onlive is really perfect for casual players with lots of disposable income who may be more interested in Lego Batman than F.E.A.R. and are much more comfortable with $X per month than $X100 for a gaming rig or even $X10 for a console.
I just don't understand how they're going to conquer the network latency issue. Latency is everything to gamers. In theory this is an amazing idea with one huge gaping critical flaw -- I look forward to seeing them solve it. Quantum entanglement, anyone?
Me neither, all the coverage I have seen (and I haven't been following too closely I have to admit) has pointed to some magical prediction algorithm they've supposedly developed and having very local servers. Given on a great day I get a 30ms ping to servers in my city this would mean at best there is a 60ms delay between me hitting fire and seeing the result. Given that[1] is below the 24fps threshold you need to have to perceive smooth motion I expect you will be able to perceive the delays.
Perhaps they can colocate their servers with ISPs or telco exchanges or something but will that be enough and can that truely scale given the consistent hardware upgrade requirements they're going to have?
What does the latency between performing an action and seeing the result have to do with the FPS at which the game is streaming into my monitor? If OnLive is sending me 60FPS, but my latency is 60ms, is just means that everything I see/do will lag by 60 ms... but I'm still seeing 60FPS of action. You likely won't notice the lag, and if you do then your senses will adjust rapidly to the slight lag and then you won't notice any more.
It is for me. The round trip lag between moving my mouse and my viewport moving made it impossible for me to reasonably play Borderlands when I tried out the OnLive system. It may be ok for casual games, but lag between input and display is a far worse experience than an equivalent amount of network lag, which games typically have prediction code to deal with.
Latency is not actually the largest problem you encounter when setting up a system like this. Now, OnLive has the problem doubled due to their generally terrible datacenter model (a few very large datacenters), since this increases the average distance between customers and the server. But raw latency still not the most difficult problem to solve. A low-latency target does make a lot of already-existing problems worse, however.
(Disclosure: I work for Gaikai. Ask me anything and I'll try to answer.)
agreed.. everyone i know who has played it says latency is not an issue. my guess is the bigger issue is game content. their selection right now pretty much sucks
11 comments
[ 5.2 ms ] story [ 36.4 ms ] threadSome coverage we can all read: http://www.engadget.com/2011/02/08/htc-to-invest-40-million-...
Their current library emphasizes shooters when Onlive is really perfect for casual players with lots of disposable income who may be more interested in Lego Batman than F.E.A.R. and are much more comfortable with $X per month than $X100 for a gaming rig or even $X10 for a console.
Perhaps they can colocate their servers with ISPs or telco exchanges or something but will that be enough and can that truely scale given the consistent hardware upgrade requirements they're going to have?
[1] 1000ms/60ms = 16.7fps
Latency is not their problem here.
(Disclosure: I work for Gaikai. Ask me anything and I'll try to answer.)