We can do a lot better in software these days. Audio processing is cheap, compared to video processing, and so it makes sense to go all software. Things that used to be intense are trivial now. A good example would be convolution: This is the...
I mean... Don't get too excited. While I enjoyed this game and would like to give it a go, remember that community mods often are vaporware. Don't celebrate until it actually launches. Too many Coming Soon(tm) mods seem to never come.
Look at...
I dunno if you've ever contributed to an OSS project but you'll find that it isn't uncommon to have maintainers that are dickheads and communities that will perpetually bike shed anything they don't like/want. While there are projects out there...
Even the 4090 there's quite a bit of overhead. Rated power for 8-pin is 252w, specified is 150w so 1.68 safety factor. Rated for 12vhpwr is 684w, specified power for the 4090 is 450w so 1.52 safety factor. Less, but not a ton less. Only the 5090...
Ya my bet would be that either it would have to get worse in quality compared to the tiny DLSS to match the performance, or it would offer the quality it does now but with much more overhead and thus lower framerate.
It does not appear that AMD...
Could be, or they just didn't have anything big yet. You can't have a big update all the time, at least not if you want to spend time working on code. They seem to like to do a rollup every year, they may not not have had a lot for this one.
Ya it does seem likely that there is something else going on with a 5080. Since it is literally drawing about half the wattage max, it means that you could have a total failure of 3 wires and still likely be in the spec. Short term spikes aren't...
Some games would use higher quality textures with more VRAM. However the NV1 didn't really do that well for most games because it used quadrilaterals instead of triangles. Basically sprites stretched and rotated in space. Neat idea, the Sega...
I almost always give a 2-star review when something really sucks in part because it usually isn't COMPLETE garbage, but also because 1-star reviews tend to just be whiners who can't be bothered to figure the product out, or are mad because it...
It's also just cost. A modern game engine is a hell of a thing to make so you need a lot of talented programmers who are working just on the engine, not even really on the game, if you are going to roll your own. No surprise then that many...
FYI for some of the "old modules remade" there's a few that have been done in NWN:EE. Eye of the Beholder and sequels were redone in it. It can be worth your while to check what's out there (The Neverwinter Vault is the main site) if you want...
Most likely, particularly since many of the things that make them good at one make them good at the other. There's not 100% overlap, but there's a reason why nVidia's cards do both. DLSS/FSR4 are inference, so you want a fast inference engine for...
I think part of it would depend on it if it was good for LLMs. The thing that has made the prices go extra nuts for nVidia, particularly the 5090s, is that they are good for inference so people buy them not to game, but to do LLMs. If AMD made a...
No but it is hardware that FSR wouldn't be using. Running on shaders is also not free. So unless AMD has something that is more advanced training wise then no, it isn't likely to be faster with better quality. There's no reason to believe they...
Probably neither inherently. We'd need to know more about what is in them, particularly the resistors and how high quality they are. Like if they spent the money for ones with super tight tolerances, that'd help with accuracy. Also I don't know...
No, they won't. Well unless someone else decides to start a price war and that isn't very likely. However I do think there will be more care about the consumer market after the DC boom wears off. They will still almost certainly make more on...
The problem is those two statements are kind of opposed. What I mean is if they port FSR4, and it looks good, then it will be heavy on the resources. To reduce the load it has to be a cut down network. That won't look as good. Also I believe the...
Do note they are always likely to disagree a little bit. Part of it is you only get so much accuracy with a shunt resistor measurement. They aren't always bang on the ohmage they are supposed to be, and there's thermal noise. A bigger part is...
Well, I'm actually inclined to believe nVidia in this case rather than the leakers as it doesn't make a lot of sense for them to make a statement like this if it isn't true. They could just say nothing if they are discontinuing them, they don't...