How the hell are they going to manage streaming 120 very high quality frames per second 👀 I have a pretty good fast fibre connection and even I can't imagine that going smoothly on my connection.
How the hell are they going to manage streaming 120 very high quality frames per second 👀
I have a pretty good fast fibre connection and even I can't imagine that going smoothly on my connection.
"At least 35Mbps" for 1440p120, according to the system requirements. That sounds low to me, but not absurdly so: Netflix pushes 4K content* at half that, although they've been accused of...
"At least 35Mbps" for 1440p120, according to the system requirements. That sounds low to me, but not absurdly so: Netflix pushes 4K content* at half that, although they've been accused of sacrificing video quality to do so. I'd probably want a connection that could handle at least 60Mbps rock solid in order to be happy I was making worthwhile use of this over the lower tier, but I think you'd get into diminishing returns towards 100Mbps, and I imagine Nvidia aren't going to be dedicating quite that much of their upstream bandwidth to you anyway.
Obviously the ISP market differs significantly from country to country, but I'm seeing more and more places with a straight divide between gigabit fibre and everything else now. If you've got FTTP, you can handle this, plus the rest of the family each watching Netflix/YouTube/TikTok without even breaking a sweat. If you've got anything else still within the realm of "decent connections" you'll still probably be fine but there's going to be some potential clipping of quality at the top end. If you're on a significantly older connection, yeah, you're probably out of luck on this one.
*Roughly twice the pixels of 1440, but roughly half the frame rate we're talking about. They get the advantage of doing two-pass compression in advance rather that low latency real time compression, so for a back of an envelope calculation it'd track that Nvidia needs at least double to account for that plus a modest increase in quality.
I think the cheapest 3080s are about $700 MSRP, so that would already cover three and a half years of this tier of GeForce Now by itself, without even looking at all the other parts you need to...
I think the cheapest 3080s are about $700 MSRP, so that would already cover three and a half years of this tier of GeForce Now by itself, without even looking at all the other parts you need to build a PC to support that level of video card.
It's really not a bad option, especially if you only play graphics-intensive games occasionally (assuming you have a good enough connection and the added latency won't be a big deal).
Maybe when GPU prices are back down to earth. A 3080 alone has a market price of ~$1500, which alone would pay for 7.5 years of the higher tier. Add on the rest of the PC and it'll be 10+ years...
Maybe when GPU prices are back down to earth. A 3080 alone has a market price of ~$1500, which alone would pay for 7.5 years of the higher tier. Add on the rest of the PC and it'll be 10+ years before you start saving money.
I'd also assume the gear on this tier will get upgraded while your parts are gonna stay the same. If I had the internet to use cloud gaming, I'd really think hard about getting this sub. I'm stuck...
I'd also assume the gear on this tier will get upgraded while your parts are gonna stay the same. If I had the internet to use cloud gaming, I'd really think hard about getting this sub. I'm stuck on decent LTE for now, but the Xcloud streaming barely works still.
A combination of market segmentation and good old timesharing, I imagine. I doubt this'll cannibalise that many real sales of 3080s, simply because users of a card that high-end probably also care...
A combination of market segmentation and good old timesharing, I imagine. I doubt this'll cannibalise that many real sales of 3080s, simply because users of a card that high-end probably also care more than the average about latency, and image quality, and compression artifacts, and all the other bits that streaming will never quite match local for.
Plenty of people who don't care too much about those things being absolutely 100% optimised, wouldn't use the card enough to justify it, or just straight up couldn't afford it, can now give Nvidia money, though - and since those cores are being used by probably 6-10 people each day, that's more profit per core than just selling them in a standalone card. Less in total than selling 6-10 standalone cards, sure, but I'm assuming only 1-2 of those people would otherwise have been a buyer.
Plus, obviously, we are in the midst of a multi-year chip shortage, so profit-per-chip suddenly becomes relevant because supply is constrained even for otherwise-willing consumers.
How the hell are they going to manage streaming 120 very high quality frames per second 👀
I have a pretty good fast fibre connection and even I can't imagine that going smoothly on my connection.
"At least 35Mbps" for 1440p120, according to the system requirements. That sounds low to me, but not absurdly so: Netflix pushes 4K content* at half that, although they've been accused of sacrificing video quality to do so. I'd probably want a connection that could handle at least 60Mbps rock solid in order to be happy I was making worthwhile use of this over the lower tier, but I think you'd get into diminishing returns towards 100Mbps, and I imagine Nvidia aren't going to be dedicating quite that much of their upstream bandwidth to you anyway.
Obviously the ISP market differs significantly from country to country, but I'm seeing more and more places with a straight divide between gigabit fibre and everything else now. If you've got FTTP, you can handle this, plus the rest of the family each watching Netflix/YouTube/TikTok without even breaking a sweat. If you've got anything else still within the realm of "decent connections" you'll still probably be fine but there's going to be some potential clipping of quality at the top end. If you're on a significantly older connection, yeah, you're probably out of luck on this one.
*Roughly twice the pixels of 1440, but roughly half the frame rate we're talking about. They get the advantage of doing two-pass compression in advance rather that low latency real time compression, so for a back of an envelope calculation it'd track that Nvidia needs at least double to account for that plus a modest increase in quality.
At that price ($99 per six months), I feel like the economics make it more feasible to just build a gaming rig.
I think the cheapest 3080s are about $700 MSRP, so that would already cover three and a half years of this tier of GeForce Now by itself, without even looking at all the other parts you need to build a PC to support that level of video card.
It's really not a bad option, especially if you only play graphics-intensive games occasionally (assuming you have a good enough connection and the added latency won't be a big deal).
Maybe when GPU prices are back down to earth. A 3080 alone has a market price of ~$1500, which alone would pay for 7.5 years of the higher tier. Add on the rest of the PC and it'll be 10+ years before you start saving money.
I'd also assume the gear on this tier will get upgraded while your parts are gonna stay the same. If I had the internet to use cloud gaming, I'd really think hard about getting this sub. I'm stuck on decent LTE for now, but the Xcloud streaming barely works still.
I wonder what the motivation for this service is, if it is in fact cheaper for the consumer. Shortages notwithstanding...
A combination of market segmentation and good old timesharing, I imagine. I doubt this'll cannibalise that many real sales of 3080s, simply because users of a card that high-end probably also care more than the average about latency, and image quality, and compression artifacts, and all the other bits that streaming will never quite match local for.
Plenty of people who don't care too much about those things being absolutely 100% optimised, wouldn't use the card enough to justify it, or just straight up couldn't afford it, can now give Nvidia money, though - and since those cores are being used by probably 6-10 people each day, that's more profit per core than just selling them in a standalone card. Less in total than selling 6-10 standalone cards, sure, but I'm assuming only 1-2 of those people would otherwise have been a buyer.
Plus, obviously, we are in the midst of a multi-year chip shortage, so profit-per-chip suddenly becomes relevant because supply is constrained even for otherwise-willing consumers.