Anybody else use GFN or Stadia in lieu of locally owning a powerful GPU?

[NewbieOneKenobi]

As per title.

I'm fresh from the paradigm-shattering experience of seeing a game in 1440p @ all settings maxed max out on my monitor's refresh rate… on a potato laptop from 2014 with an i5-5200u and no discrete graphics. So at this point I'm like… huh, a local 3080 would cost me at least 50 months of GFN subscription and obviously within 50 months we'd be looking at 5080, with 4080 in the meantime. This is probably very much like the 'buy or lease' dilemma with company cars and laptops. Welcome to HaaS. And it's ironic that gaming should be the pioneer of it — but completely not irrational given the prices of gaming equipment and the costs and risks of ownership (+300W on PSU; stronger CPU to drive the GPU; more PSU wattage with that; replacement cost if the thing breaks down the day after its 2-year warranty expires, etc.). Of course, a real, physical 3080 would have some resale value, but then the subscription presumably updating to current generations would still beat you selling the 3080 and buying a 4080 and then 4080 => 5080.

Of course, if the monthly subscription cost you 1/20 of the cost of a local card, then it could be better to buy the card. But if the ratio is closer to 1/40 or 1/60, then that changes things, and quite drastically.

In my case, being almost without a GPU at the moment (7850 2GB will run some games @ low or mid settings), I figured I could skip the GPU purchase and just buy a new 12600K (with an iGPU to support GFN, which my 9600KF doesn't have) and 420 AIO and OC the heck out of it for single-core performance I sometimes need at work when processing particularly large or complicated files through unoptimized parsers.

Of course, I could buy a nice 3070 for 620 pounds, but a year's worth of GFN costs 120. I'd rather take 5 years of streamed 3080 than a local 3070. There are factors to consider, but they only seem to weaken this result a little, not overturn it.

The next thing I can think about is, well, if GFN proves to be a reliable service and good value for the money in the long run, and if it stays around without becoming much more expensive, I suppose I could reduce my desktop + laptop setup to just laptop instead of upgrading both. (In a way, this is conceptually similar to the older idea of using a USB/Thunderbolt GPU case to hold a desktop GPU, which is still less expensive than a desktop PC, except going a step further and eliminating local ownership of a GPU at all).

What do you guys think? Obviously part of me longs for a nice little 3070 here in my rig, but at the same time I can't deny the maths, it really seems to make little sense buying a GPU vs GFN.

 

[NewbieOneKenobi]

Putting a two or three dozen GB per hour strain on your relationship with your ISP probably carries a bit of negative goodwill.

I've been calculating the economic factors these last days and while EPYC+A10G 'rented' for fifteen quid monthly, hence e.g. 360 for 2 years, appears to be more attractive than a local PC upgrade, especially if your CPU is also a problem, such as taking an old business laptop as a starting point (e.g. my Latitude 5550 w/i5-5200u and no discrete graphics). However, in the long-term, after 5 years that's 900 pounds and you begin to wonder about the resale value a local PC might have. In the short term, especially if you have a decent CPU, you realize that 90 — for half a year — is a dent in your GPU budget easily the size of a whole GPU tier and your GPU obviously wouldn't lose 180 pounds p.a. in value bar extreme events. So without the benefit of a proper, rigorous analysis, both the short term and the long term look far less attractive than the mid-term.

Then again, perhaps the mid-term could be a niche in which GFN could find its legitimate application. For example if you need to wait something out, such as the current situation, or wait for some specific future event like AM5 release or large payout. I suppose it's not unreasonable to expect 3080 prices to go down by at least 90 GBP in half a year, but then you could always be wrong — see the Q4 2021 letdown after the high hopes of July and August. Or there could be another pandemic and supply chains grinding down to a halt (although that in itself could be a situation GFN is of good help in waiting out).

If I look at my existing backlog of not very recent games, none of which bar one or two support RT or DLSS, a 1080ti would be sufficient for fluent 1440p ultra, for the price of 1.5 years of GFN's 3080 subscription (Priority costs half but is capped at 1080p; although 1080p ultra @ 60fps is still great in itself, it looks bad on my 32'' 1440p monitor). It's reasonable to presume that a 1080ti will keep at least 50% of its street price 1.5 years from now. The only risks are that it might simply die or end up being hard to sell due to age, but I can't imagine being unable to recover at least 33% of its value after 18 months, while 66% is not out of the question. The difference in GFN 3080's favour, however, is that it recently supports 4K/60fps, which is something a 1080ti wouldn't always achieve in games like The Witcher 3, Kingdom Come: Deliverance or especially newer ones like Horizon, and 'fake' (dowscaled with DSR/VSR) 4K usually looks better than normal 1440p on my monitor. GFN 3080 won't go above 60fps in 4K, but realistically speaking a 1080ti usually wouldn't either, or maybe 65 (my 1070ti achieved 55 in the best cases) but not really the 75 my monitor can do. And for pure 1440p without insisting on a huge framerate, a used 1070ti costs 2/3 of the price of a used 1080ti and is probably the best money-for-fps except for risks relating to its age and history (although 1070ti released 1.5 years after 1080 or 1060). If it survives a year or two, and it probably will, it should also retain at least a half of its current value. And its SLI-able in case it gets really cheap.

I suppose I'll have to rethink this idea. However, for the time being my subscription is valid until 16 June, and on 31 May comes out a nice DLC for CK3, which I already own and which will benefit greatly (in the late game) from GFN's EPYC CPUs compared to my 9600KF. Until 31 May, I have a company website to redo. So it's probably reasonable to give it a try, maybe extend by another month. One advantage is that I could actually buy a modern game (in which 9600KF + 1080ti would cap out at 40fps in 1440p ultra with RT disabled) while my subscription lasts and push the oldies backlog further down. Unfortunately, I have several GFN-supported games on GOG rather than Steam, which means I'd need to rebuy them on Steam (increasing the total cost of this path), but again, those could be put off till later and I could just buy and play new games. GFN 3080 does have an advantage when it comes to playing new AAA releases in 1440p, especially with RT.

What's optimistic is that the economic outcomes are so close to each other that you can't go completely wrong, whichever route you take. Kind of like owning vs leasing in some industries. There's mess with the various outcomes, a difficulty sorting out and comparing them and making your personal pick, but the actual economic loss is small in any case. Sigh. Haven't made much progress, it seems.