4 ms·

GameStop's lack of scruples: not surprising. GameStop fears OnLive as a substantial threat: a little surprising. Internet speeds for the majority of Americans a

GameStop's lack of scruples: not surprising. GameStop fears OnLive as a substantial threat: a little surprising. Internet speeds for the majority of Americans are just way too slow for OnLive to be viable.

tatsuke95 15y ago
But unless we have some bizarre infrastructure regression, it only becomes more viable in time. It (or something like it) is the future, there's no doubt about that.
- eropple 15y ago
  Requirements will also increase, though, and the biggest problem with OnLive can only be fixed by raising the speed of light.
  - jergason 15y ago
    Requirements increasing will only mean they have to upgrade their computers in a data center somewhere, not that the ping between the server and you needs to get faster. The delay is noticeable, but games requiring more processing power will have no effect on the time it takes OnLive to send you data.
    - eropple 15y ago
      Of course, but even that's a nontrivial task. OnLive's game machines target somewhere around "low-middle"; the quality isn't even console-level. They buy machines that are close to the first to be obsoleted by the upgrade march, so there'll always be the "damn, that plays like ass on OnLive" problem for new releases. When you couple pretty poor quality with persistent, noticeable latency, you do not have a recipe for success.
      - regularfry 15y ago
        It'll only take one AAA publisher with a deep marketing budget to sink a couple of million into dedicated OnLive servers for a new title for that problem to go away. Maybe that's what they're banking on.
        vectorpush 15y ago
        I'm not so sure. To make an impact it would have to be a very popular IP designed with OnLive's stack in mind coupled with platform exclusivity (at least at launch). The problem I see for OnLive is that their target demographic (gamers who purchase AAA titles) are the consumers who are most critical of performance issues. If OnLive were serving up content to casual gamers, the service's shortcomings would be less of an obstacle, but I just don't see how they're going to get tech-aware consumers to jump on board.
        regularfry 15y ago
        There's another angle, though: while the control round-trip might be hideous on OnLive, the inter-user round-trip is negligible; presumably everyone will be sitting on the same gigabit (or better) LAN. Games currently have to pull all sorts of tricks to make the inter-user latency appear not to be present. They've got good enough that generally you just don't notice. I think the same thing will happen here, although I admit that nobody's demonstrated that they're capable of surmounting the technical challenge, and I doubt it's as simple as tweaking #defines in the network stack.
        eropple 15y ago
        It's a lot easier for a game to let you shoot at where your client-side prediction thinks the guy is supposed to be than to allow players to react to things that happened before their input for a given time tick was processed. What you're saying is nontrivial and may not even be workable, because it's impossible to really separate "stupid slow player" from "player whose latency is high enough that he can't react".

tatsuke95 15y ago

But unless we have some bizarre infrastructure regression, it only becomes more viable in time. It (or something like it) is the future, there's no doubt about that.

eropple 15y ago
Requirements will also increase, though, and the biggest problem with OnLive can only be fixed by raising the speed of light.
- jergason 15y ago
  Requirements increasing will only mean they have to upgrade their computers in a data center somewhere, not that the ping between the server and you needs to get faster. The delay is noticeable, but games requiring more processing power will have no effect on the time it takes OnLive to send you data.
  - eropple 15y ago
    Of course, but even that's a nontrivial task. OnLive's game machines target somewhere around "low-middle"; the quality isn't even console-level. They buy machines that are close to the first to be obsoleted by the upgrade march, so there'll always be the "damn, that plays like ass on OnLive" problem for new releases. When you couple pretty poor quality with persistent, noticeable latency, you do not have a recipe for success.
    - regularfry 15y ago
      It'll only take one AAA publisher with a deep marketing budget to sink a couple of million into dedicated OnLive servers for a new title for that problem to go away. Maybe that's what they're banking on.
      - vectorpush 15y ago
        I'm not so sure. To make an impact it would have to be a very popular IP designed with OnLive's stack in mind coupled with platform exclusivity (at least at launch). The problem I see for OnLive is that their target demographic (gamers who purchase AAA titles) are the consumers who are most critical of performance issues. If OnLive were serving up content to casual gamers, the service's shortcomings would be less of an obstacle, but I just don't see how they're going to get tech-aware consumers to jump on board.
        regularfry 15y ago
        There's another angle, though: while the control round-trip might be hideous on OnLive, the inter-user round-trip is negligible; presumably everyone will be sitting on the same gigabit (or better) LAN. Games currently have to pull all sorts of tricks to make the inter-user latency appear not to be present. They've got good enough that generally you just don't notice. I think the same thing will happen here, although I admit that nobody's demonstrated that they're capable of surmounting the technical challenge, and I doubt it's as simple as tweaking #defines in the network stack.
        eropple 15y ago
        It's a lot easier for a game to let you shoot at where your client-side prediction thinks the guy is supposed to be than to allow players to react to things that happened before their input for a given time tick was processed. What you're saying is nontrivial and may not even be workable, because it's impossible to really separate "stupid slow player" from "player whose latency is high enough that he can't react".