Discussion in 'Hardware' started by The_Crapman, 7 Aug 2018.
Not DXR, but it should still be accessible via Vulcan RT, and via NVAPI.
nvidia have done you a favour by delaying the 2080 so you can cancel your orders now the bench's are out showing its on par with a 1080 ti
Wrong way around: the 2080 is shipping (FE pre-orders should have already been dispatched), it's the 2080 Ti that is shipping this week/early next week.
I'm not a fan of his but IMO Linus nails it in this video...
Same with any brand new tech really. Needs support..
The things is with most new tech you expect to be able to use it, maybe not all the time and in everything but you at least expect for it to be usable, currently it's just not.
Yup. Silly really. They would have been better off making RT cards separate for Pascal owners to shove in. Ah well, going for the big ticket prices I suppose.
I spent three years and god knows how much money chasing 4k. Not doing it again. Will wait and see what happens. What does worry me is that this is a thing on one card set (so not on AMD for example) as well as being unique to the PC. It's all about whether devs will go to that extra aggro for that % of PC users. If past numbers are anything to go by? yeah, not very hopeful. GPU Physx was really cool, but never got a foothold. Shame, but that is what happens when you have something that is limited in users.
Like with Ageia-era PhysX, putting the RT cores on a seperate card (even ignoring the impracticality due to lack of bandwidth with the GPU. No NVLink isn't close to cutting it) would have killed adoption right fast in a hurry.
To get things implemented, you've got to ship the hardware. That's what's happened with past advances in GPUs (e.g. hardware T&L, programmable shaders) and will continue to happen.
I mean... unless you count being in the majority of games ever released? It's the default physics engine in Unity and Unreal Engine. Pierre Terdiman goes into detail in this blog post from August. He does say 'it is fair to say that the GPU stuff is not used much (although a bunch of projects do so),' but also that 'PhysX 3.4 (the most recent public version of PhysX) even introduced a brand new GPU rigid body codepath - the most efficient we ever built. So it is the opposite of dead: that new version was basically just born yesterday.'
I meant GPU physx, using the PPU accelerator aboard Nvidia GPUs. That is why I said GPU Physx. Or, being a little bit more specific, games that have OTT Physx usage like Mafia 2 and Mirror's Edge that ran like dog turd when put onto the CPU. Nvidia used it as a sales pitch for many years.
Let's face it RTX has a rather large hill to climb. It is specific to one set of big brand GPUs (Nvidia, not AMD) and therefore will not be a feature on consoles (in the RTX guise at least). This means devs need to spend lots of time and money on a set of individuals who are not going to make up a very large % of all GPU owners. Oh, and lest we forget the RRP for entry level right now is £749 for a toss card with a toss cooler on that will probably fry when Metro comes out. I predict Fermi all over again.
If it makes it and if it succeeds? cool beans, I will get it. Until then? ain't touching nothing that doesn't work on a console, as that is what games are primarily coded for.
As Millhouse would say WHAT A RIP!
The technology won't die because it has uses outside of gaming where there is a shed ton of money, ie for cgi in films. I also don't think it will be ignored by developers because it's what they want; real shadows from real light sources in real-time. Nvidia are helping them build the tools and they've even got Microsoft on board with DXR and Vulcan support will soon follow.
Nvidia were staring at a full on chicken and egg situation, so they bit the bullet and shat out an egg. The problem a lot of us have is that this egg stinks. It's covered in ****, it's inedible, the scrawny-ass chick that will hatch will likely never get enough meat on it for even a happy meals worth of McNuggets and they're charging as much as Mr Fabergé which has done nothing to 2nd hand prices. But in 2 or 3 generations we'll have big ****-off fighting cocks and the consoles will follow suit when technology allows, but whether amd (or Intel) can fill that need is a big uncertainty.
NVIDIA RTX Marketing Tear-Down.
LOL, as expected by anything distributed by Digital River, shipping is cocked up and you can ask 3 different people and get 5 different answers as to what is going on. No money down (never use Paypal with Digital River, or at all if yo can help it) until anything actually ships, so meh.
Supposedly 5-9th Oct is the new date.
Anyone want to speculate why the dates slipped twice now, availability would seem an odd one as reviewers have them and it's not like you need X number of cards before you can start shipping, software problems would also seem odd as again reviewers have reviewed them.
Some clause about priority of chip allocation in their contracts with Asus, Gigabutt, MSI etc.
Maybe to entice people to buy more 2080's over a 1080ti, without there being a better card about.
My guess is containerised sea freight still being fscked up globally due to hurricanes hitting multiple international ports. Sending out a handful of cards direct from factory via courier is acceptable for a handful (one or two hundred at most), but to ship multiple thousands that isn't feasible.
Worst case is a ISO container of cards en route to a distro centre is currently sat at the bottom of the ocean near south-east Asia after being blown off of a ULCV during a storm, and TSMC are scrambling to fab and QC new dies to create replacement cards. Or in the general chaos the container is sitting incorrectly tagged in a port somewhere (after being snap-offloaded by mistake in the rush to finish turnaround before one of the various tropical storms closd port access and trapped the host vessel) while a shipping company scrabbles to find it.
If that's happened i want to know where as I've heard magnet fishing is a thing now.
Separate names with a comma.