Don't forget they charge more for Quadro cards. Gaming was about smaller margin but more volume. 980 Ti was bigger than any previous GeForce, yet still cheaper than older cards like the 780 Ti and 8800 Ultra. Increasing prices this radically is a choice, not a necessity, due to lack of competition.
If consumers stand firm, then they will decrease prices out of necessity. It's up to the market now to raise some fists.
Of course the rtx stuff isn't ready. 50-70fps at 1080p and not fully maxed settings I bet too. So their selling point is vaporware until it's implemented properly. It seems super expensive and they need time to work on it. It's not something you will get right away with every game. And those devs that don't implement it right, your game could be a slideshow.When your slow leak viral marketing campaign blows up in your face
Those previous cases are marginal differences. This 754mm2 die is massive, and the relation of cost to size increase is not linear because yield of bigger dies suffers more.
Also remember that in previous generation like Pascal, the first exposure to 102 dies was in Titan card, The Titan X was $1200.
It seems like NVidia could have just called the 2080Ti Titan T, and really there would have been little new to complain about on the high end.
"Proper RT" is a meaningless term. If you think the proper way to do this is to blast 100's to 1000's of rays at every pixel like it was traditionally done when you take a day to render one frame, that is never going to happen in real time for gaming. That isn't proper, it's just a dumb brute force, inefficient waste of resources.
What is being done here is much more efficient and intelligent use of resources, to cut down the rays by an order (or more) of magnitude, and it will never revert to the old dumb brute force method.
In a demo called "Infiltrator" at Gamescom, a single Turing-based GPU was able to render it in 4K with high quality graphics setting at a steady 60FPS. Huang noted that it actually runs at 78FPS, the demo was just limited by the stage display. On a GTX 1080 Ti, that same demo runs in the 30FPS range.
I have doubts regarding the claim that GDDR6 is THAT much more expensive than GDDR5, especially for someone like NVIDIA.
It seems like NVidia could have just called the 2080Ti Titan T, and really there would have been little new to complain about on the high end.
Please be true.
A 2080 Ti being more than 2x framerate of a 1080 Ti is what would be needed for a 2070 to beat a Titan Xp, btw. And NV did say that yesterday (though most of assumed they meant doing Ray Tracing).
We seem to have two reports of what appears to be 2X performance in existing games.Please be true.
A 2080 Ti being more than 2x framerate of a 1080 Ti is what would be needed for a 2070 to beat a Titan Xp, btw. And NV did say that yesterday (though most of assumed they meant doing Ray Tracing).
Wait they need to run an Ubisoft title and make sure the card doesn't melt first.Please be true.
A 2080 Ti being more than 2x framerate of a 1080 Ti is what would be needed for a 2070 to beat a Titan Xp, btw. And NV did say that yesterday (though most of assumed they meant doing Ray Tracing).
You don't think people would have complained about $800 RTX 2080 that's already cut-down with 2944 shaders, and no real indication of IPC increase in current games so far?
DLSS doesn't seem like it would speed anything up, though?It's probably in games using DLSS. Their secret sauce to getting performance boost in games without raytracing - A list came out somewhere that indicated over half the RTX supporting games were simply adding support for DLSS.
Whether you’re a PC gamer who was waiting in the wings for a more powerful graphics card or you truly believe in Nvidia’s vision of a ray traced future, the GeForce RTX 2080 Ti is already looking like the world’s most powerful graphics card in the world regardless.
GDDR5X was Micron only, so isn't comparable to GDDR5/6.GDDR5X is/was expensive right? I'd have to think more so than GDDR6.
From the Techspot piece.
It's a different way to achieve super sampling using Deep Learning algorithms running on the Tensor cores. Compared to standard TAA, it could potentially achieve the same degree of quality but at a significantly lower penalty but there's simply no information about it. We'll need to wait for reviews.DLSS doesn't seem like it would speed anything up, though?
The one article said these gpu's as in plural when they mentioned in excess of 100 fps at 4k. Wonder if they used 2 of them.
finally able to push 60fps 4K gaming with a single GPU.
I sense that we may be overlooking something important: Where is the Titan?
IIRC, Nvidia usually releases their Titan cards first, then non-TI, then TI.
This launch hasn't gone down that way, and we also have had some interesting rebrands with RTX/GTX being moved around, so it's possible the rage is due more to recent mining / misunderstanding the new branding.
Does the pricing structure make more sense if we relabel the 2080TI to "Titan" and move everything down 1 step?
2080TI -> Titan
2070TI -> 2080 TI
etc.
We seem to have two reports of what appears to be 2X performance in existing games.
I don't think so because why compare to the 1080ti then?Are you sure? I'm pretty sure the Unreal Infiltrator demo that had double the perf used ray tracing.
DLSS doesn't seem like it would speed anything up, though?