5 ms·

A big enough gap that you'd pay hundreds of dollars to cross it?

Retric 3y ago
I’m in no hurry. Tossing a low end graphics card into a new PC unlocks significantly better back catalog every time with the added benefit of having the most patched version at discounted prices.
paulmd 3y ago
Is playing starfield really such a big deal that you'd pay hundreds of dollars, when you have a perfectly fine iGPU? If you don't buy DX12.2 capable hardware you don't get the DX12.2 features, it's really that simple. buying fermi or terascale or some other ancient hardware will always be cheaper, but studios aren't just going to not release games because someone is clinging to their 4870. and the situation is not different in kind when we're talking about DX12.0 to 12.2 transition instead of DX9 to DX11 or DX11 to DX12. If you've upgraded to a current NVIDIA gpu in the last 5 years, or an AMD card in the last 2 years, you got the capability. And when you next upgrade, you'll get the capabilities on AMD too, unless you go out of your way to buy older cards without support, as many people did with DX12.2 (because reviewers lost their minds over Turing). It is, again, super funny to look back at how uncontroversial this all used to be, when it was AMD who was ahead on DX feature support. Like yeah if you buy a DX11-focused card like Maxwell then your DX12 experience is going to be shit. Doctor it hurts when I do this. https://www.youtube.com/watch?v=fqLea0QUW1k&t=176 https://www.youtube.com/watch?v=fqLea0QUW1k&t=176 https://www.youtube.com/watch?v=3qiQH29KAXg https://www.youtube.com/watch?v=3qiQH29KAXg Developers aren't going to not release games just because some people insist on running 10+ year upgrade cycles, or run esoteric hardware with some different tradeoffs because they don't want to pay for newer stuff. I literally don't understand why it's an issue or a confusing topic, other than people having a bit of a rough time with the realization that Turing wasn't a pointless waste of a generation after all. Reviewers kinda did it dirty, you can argue that it was overpriced at first (2070 was still pretty good tho), but buying a used 1080 Ti instead of a 2070S to save $50 was a mistake, and buying a 2080 Ti for $950 was an outright bargain. There were plenty of deals that got shouted down in the whole "turing BAD!" tantrum from reviewers and gamers, and under the general lack of understanding that indeed the cost-trend aspect of moore's law was over and the price trend was gonna be upwards. https://youtu.be/tu7pxJXBBn8?t=273 https://youtu.be/tu7pxJXBBn8?t=273 These sorts of videos are just shameful and utterly incorrect to look back on, and it's absurd to think it's the same Steve who made the above videos 2 years before. Nuclear reaction from tech media aside, almost everything tom's said in their article was correct and has come to pass, it's interesting to re-read the article with 5 years of hindsight. This is exactly the sort of "cost to buying outdated tech" that Steve mocks, and the people who followed Steve's "long-term value doesn't real, buy the older cheaper thing without DX12.2 support" will inevitably have to upgrade sooner as a result. Pascal and Polaris have already received a lot of grace from developers, in the abstract it is crazy that tuning for 8-year-old cards is still such a significant aspect of gamedev today. But there will eventually be a point when Polaris and Pascal fall off, and then eventually can't even launch the game, just like older generations before them. And it seems like it's going to be a very rapid offramp over the next 2 years as studios move onto true current-gen titles and finally stop supporting base-tier PS4/XB1 hardware, because these cards are very very overdue for replacement. The transition to upscaling being a standard part of the pipeline is going to be another of these "gamer moments" too. People are dead-set on it being "just to boost your framerate", but even if you go back to the original launch presentation, it was always framed as making it possible to run much more intensive effects than otherwise could be. Being behind on this part of the pipeline is problematic for AMD, because it is going to be more and more directly relevant to actual game performance and not just an optional extra. Rendering 4x as many raw pixels is going to throw off all the optimizations etc.
- Nullabillity 3y ago
  > If you don't buy DX12.2 capable hardware you don't get the DX12.2 features, it's really that simple. buying fermi or terascale or some other ancient hardware will always be cheaper, but studios aren't just going to not release games because someone is clinging to their 4870. and the situation is not different in kind when we're talking about DX12.0 to 12.2 transition instead of DX9 to DX11 or DX11 to DX12. Again, the question isn't whether they should release new games (of course they should!) but whether those new games are meaningfully improved by relying on this new tech (hint: they aren't). > It is, again, super funny to look back at how uncontroversial this all used to be, when it was AMD who was ahead on DX feature support. Like yeah if you buy a DX11-focused card like Maxwell then your DX12 experience is going to be shit. Doctor it hurts when I do this. It was uncontroversial because back in the day the new tech lead to real visible improvements that you could tell without needing a side-by-side on/off comparison, not just spec wankery. GPU brand has nothing to do with it. The law of diminishing returns kicks in, and the S-curve slows down. It happens to every category of tech eventually. > buying a 2080 Ti for $950 was an outright bargain. Again, you're talking about a $950 purchase that for new tech that maybe improves the level of eye candy by maybe 1%. There is no market for that. There is no possible world where that tradeoff is worth it. But you're never going to understand that, because all you do is look at is a spec sheet, where "bigger number must obviously be betterer".

I’m in no hurry. Tossing a low end graphics card into a new PC unlocks significantly better back catalog every time with the added benefit of having the most patched version at discounted prices.

paulmd 3y ago

Is playing starfield really such a big deal that you'd pay hundreds of dollars, when you have a perfectly fine iGPU? If you don't buy DX12.2 capable hardware you don't get the DX12.2 features, it's really that simple. buying fermi or terascale or some other ancient hardware will always be cheaper, but studios aren't just going to not release games because someone is clinging to their 4870. and the situation is not different in kind when we're talking about DX12.0 to 12.2 transition instead of DX9 to DX11 or DX11 to DX12. If you've upgraded to a current NVIDIA gpu in the last 5 years, or an AMD card in the last 2 years, you got the capability. And when you next upgrade, you'll get the capabilities on AMD too, unless you go out of your way to buy older cards without support, as many people did with DX12.2 (because reviewers lost their minds over Turing). It is, again, super funny to look back at how uncontroversial this all used to be, when it was AMD who was ahead on DX feature support. Like yeah if you buy a DX11-focused card like Maxwell then your DX12 experience is going to be shit. Doctor it hurts when I do this. https://www.youtube.com/watch?v=fqLea0QUW1k&t=176 https://www.youtube.com/watch?v=fqLea0QUW1k&t=176 https://www.youtube.com/watch?v=3qiQH29KAXg https://www.youtube.com/watch?v=3qiQH29KAXg Developers aren't going to not release games just because some people insist on running 10+ year upgrade cycles, or run esoteric hardware with some different tradeoffs because they don't want to pay for newer stuff. I literally don't understand why it's an issue or a confusing topic, other than people having a bit of a rough time with the realization that Turing wasn't a pointless waste of a generation after all. Reviewers kinda did it dirty, you can argue that it was overpriced at first (2070 was still pretty good tho), but buying a used 1080 Ti instead of a 2070S to save $50 was a mistake, and buying a 2080 Ti for $950 was an outright bargain. There were plenty of deals that got shouted down in the whole "turing BAD!" tantrum from reviewers and gamers, and under the general lack of understanding that indeed the cost-trend aspect of moore's law was over and the price trend was gonna be upwards. https://youtu.be/tu7pxJXBBn8?t=273 https://youtu.be/tu7pxJXBBn8?t=273 These sorts of videos are just shameful and utterly incorrect to look back on, and it's absurd to think it's the same Steve who made the above videos 2 years before. Nuclear reaction from tech media aside, almost everything tom's said in their article was correct and has come to pass, it's interesting to re-read the article with 5 years of hindsight. This is exactly the sort of "cost to buying outdated tech" that Steve mocks, and the people who followed Steve's "long-term value doesn't real, buy the older cheaper thing without DX12.2 support" will inevitably have to upgrade sooner as a result. Pascal and Polaris have already received a lot of grace from developers, in the abstract it is crazy that tuning for 8-year-old cards is still such a significant aspect of gamedev today. But there will eventually be a point when Polaris and Pascal fall off, and then eventually can't even launch the game, just like older generations before them. And it seems like it's going to be a very rapid offramp over the next 2 years as studios move onto true current-gen titles and finally stop supporting base-tier PS4/XB1 hardware, because these cards are very very overdue for replacement. The transition to upscaling being a standard part of the pipeline is going to be another of these "gamer moments" too. People are dead-set on it being "just to boost your framerate", but even if you go back to the original launch presentation, it was always framed as making it possible to run much more intensive effects than otherwise could be. Being behind on this part of the pipeline is problematic for AMD, because it is going to be more and more directly relevant to actual game performance and not just an optional extra. Rendering 4x as many raw pixels is going to throw off all the optimizations etc.

Nullabillity 3y ago
> If you don't buy DX12.2 capable hardware you don't get the DX12.2 features, it's really that simple. buying fermi or terascale or some other ancient hardware will always be cheaper, but studios aren't just going to not release games because someone is clinging to their 4870. and the situation is not different in kind when we're talking about DX12.0 to 12.2 transition instead of DX9 to DX11 or DX11 to DX12. Again, the question isn't whether they should release new games (of course they should!) but whether those new games are meaningfully improved by relying on this new tech (hint: they aren't). > It is, again, super funny to look back at how uncontroversial this all used to be, when it was AMD who was ahead on DX feature support. Like yeah if you buy a DX11-focused card like Maxwell then your DX12 experience is going to be shit. Doctor it hurts when I do this. It was uncontroversial because back in the day the new tech lead to real visible improvements that you could tell without needing a side-by-side on/off comparison, not just spec wankery. GPU brand has nothing to do with it. The law of diminishing returns kicks in, and the S-curve slows down. It happens to every category of tech eventually. > buying a 2080 Ti for $950 was an outright bargain. Again, you're talking about a $950 purchase that for new tech that maybe improves the level of eye candy by maybe 1%. There is no market for that. There is no possible world where that tradeoff is worth it. But you're never going to understand that, because all you do is look at is a spec sheet, where "bigger number must obviously be betterer".