undefined | Better HN

0 pointsargo_6y ago0 comments

A technology like Google Stadia will probably solve this. Nobody will need a expensive hardware to run it. The graphics will be generated in cloud computing and streamed to the VR. Keep creating hardwares to run better graphics is a dead end.

0 comments

kec6y ago

The speed of light says no. “The cloud” will always have too much latency to do acceptable vr. To meet your deadline for a 120hz video frame, your signal can travel at most 1500 miles in a vacuum, closer to 1000 miles in copper/fiber.

dragonwriter6y ago

> “The cloud” will always have too much latency to do acceptable vr.

Perhaps but you haven't established that.

> To meet your deadline for a 120hz video frame, your signal can travel at most 1500 miles in a vacuum, closer to 1000 miles in copper/fiber.

Edge computing is a feature of some modern clouds, and certainly is intended to have much lower round trip distance than that to at least major markets, specifically to reduce latency.

Admittedly, public cloud offerings don't tend currently to have compute resource suitable for hosting VR in their edge offerings, but it's certainly something a major cloud player supporting their own gaming/VR offering could do, and for that matter a feature that it is easy enough for public cloud vendors to offer if there was sufficient demand.

kec6y ago

Edge computing is a thing, but it really doesn’t help that much: That 1000 miles has to include the signal path inside your gpu to generate the frame and assumes no data loss in transit.

If it takes 4ms to generate a frame you’re down to < 500 miles from the data center. If you need to pad for packet loss you’re down even further.

1 more reply

j / k navigate · click thread line to collapse

0 comments

kec6y ago

dragonwriter6y ago

> “The cloud” will always have too much latency to do acceptable vr.

Perhaps but you haven't established that.

> To meet your deadline for a 120hz video frame, your signal can travel at most 1500 miles in a vacuum, closer to 1000 miles in copper/fiber.

Edge computing is a feature of some modern clouds, and certainly is intended to have much lower round trip distance than that to at least major markets, specifically to reduce latency.

kec6y ago

Edge computing is a thing, but it really doesn’t help that much: That 1000 miles has to include the signal path inside your gpu to generate the frame and assumes no data loss in transit.

If it takes 4ms to generate a frame you’re down to < 500 miles from the data center. If you need to pad for packet loss you’re down even further.

1 more reply

j / k navigate · click thread line to collapse