Hacker News Reader

Comment by birdyrooster | Hacker News Reader

I’ll lead with my peak cynical take, apologies in advance, but I think Google just wanted to entice people with cheap centralized streaming games while it was viable to get them married to a platform they will ultimately have to buy a local console to render games with anyways.

Can they even get close to DisplayPort 2.0 performance in the next couple years? I don’t see how this is ever going to scale as bit depth, resolution and frame rates continue to increase. This is a losing strategy because a lossy, slow connection between your display and your controller will only get worse (quadratically so) as those three factors (bit depth, resolution and frame rate) increase.

They are going to have to beef up the client hardware and do more of the game engine and rendering compute locally, which will undermine the cost savings they offered being centralized.

View on HN

My guess is that they're testing the waters for enterprise software. Think Google Docs but for CAD. There's a lot of money in that market and people are used to paying $1000+ monthly subscriptions anyway. So if you can move that onto a remote desktop server, you'll have amazing performance and healthy profit margins.

And if it works well for games, it'll surely work well enough for regular software.

I agree that is a good fit for a use case, but I’m speaking about Stadia in particular. They are going to disappoint a lot of consumers.

Why would you get better performance from CAD in the cloud? Doesn't a cloud server have essentially the opposite performance profile from a CAD workstation, with many more cores but much slower clock rates? CAD vendors want your workstation to have a few fast cores and several GPUs.

Clouds are starting to focus on HPC machine types which can crank out CAD renders

I see. I guess it's technically possible for a cloud vendor to temporarily attach several GPUs to your CAD render, and afterward reassign them to someone else, therefore bringing the utilization up and the price down, it just doesn't seem like such a big market opportunity when a $20k workstation isn't that important next to the cost of the software and its operator.

dont forget that you're not tied to one workstation to do the job. makes WFH(or anywhere really) much easier, and much higher potential for easy collaboration

> CAD vendors want your workstation to have a few fast cores and several GPUs.

https://www.nvidia.com/en-us/data-center/quadro-virtual-work...

I think premise is either "floating workstation" or that if you please you can use laptop instead of being tied to workstation.

I imagine in the enterprise space there could be a desire to have cheap simple to manage machines to give to employees or BYOD with a subscription service to manage cloud compute resources for things like CAD. Lots of enterprises don't like being in the internal IT management and compute hardware hosting business, they're in the business of making whatever they're designing in the CAD software. We've already seen this kind of migration to cloud services for other kinds of business processes.

I wouldn't expect better performance, but you usually license software by node. Having it run on a shared server is an easy way for employees to share and work together.

Reply on news.ycombinator.com