It probably mainly means he's as underwhelmed as most reviewers seem to be. What with the big fps loss for a handful of not that noticeable graphical gimmicks, and all.
- Status
It probably mainly means he's as underwhelmed as most reviewers seem to be. What with the big fps loss for a handful of not that noticeable graphical gimmicks, and all.
FrameBuffer
Banned
so that pretty much sums up PhyX .. I guess shouldn't be too long before he tweets about PhyX being a failure too ??
Heh sure. Actually, Nvidia's party line for the next few months is quite likely to be that Physx delivers more than DX11 right now. Which isn't even all that outrageous a suggestion either. Dirt 2 is rather pretty in DX9.
Yeah Nvidia really has problems if they're criticizing (and alienating) developers now to distract from their own failures. Wouldnt you want devs on your side if you're peddling GPUs?
PhysX still has an upper hand on noticeable changes though. DX11 by design will be much, much subtler.
PhysX still has an upper hand on noticeable changes though. DX11 by design will be much, much subtler.
But STALKER is a lot more than that
Jawed
Legend
Eh? A complete physics and AI engine could be created in D3D11. Compute is part of D3D11.
Jawed
I like that Dirt2 has AMD and Intel splash screens side by side (in alphabetical order).
It also runs equally well on Phenoms and Core i5, i7 and even Penryns.
As to comments regarding negligible visual difference between DX9 and DX11 - we are at already high gfx fidelity, so next API's will bring subjectively small improvements, but for programmers and physical accuracy they will be huge!
It also runs equally well on Phenoms and Core i5, i7 and even Penryns.
As to comments regarding negligible visual difference between DX9 and DX11 - we are at already high gfx fidelity, so next API's will bring subjectively small improvements, but for programmers and physical accuracy they will be huge!
Comparing a middleware package to a compute API? Whether a PhysX equivalent "could" be written in DirectCompute is sorta irrelevant to my comment. You might as well say the same thing about C++.
What's going to stand out more to the average gamer? Tessellated water in Dirt2 or the Scarecrow level in Batman? Like I said, DX11 effects are by definition more subtle so it will be harder for devs to produce whiz-bang differences in comparisons to DX9/10.
Saying that you "could" code something isn't really an argument. That goes for any programming language or API. Fact is that PhysX exists and DirectCompute based physics libraries do not. Right now what we'll see from devs is tessellation and CS based post-processing. That is DX11 for the forseeable future.
Silent_Buddha
Legend
It will depend on the implementation. A game where you fly by objects at 50+ MPH (80+ KPH) you really are going to have a hard time noticing tesselation even if it's blatantly over-exaggerated. Although even there I could see some area's where it could shine. Deeply rutted dirt roads for example, but you aren't likely to see something like that in a racer.
I'm waiting to see if anyone will be able to implement anything as eye-popping as what's available in the Unigine demo/benchmark for a FPS/TPS/RPG/etc... Where the action is not only orders of magnitude slower, but scenery and objects can be viewed much closer.
Additionally from what I've seen, the biggest "wow" factor I've personally had with tesselation so far is the way light and shadows interact with tesselated geometry compared to how light and shadows interact with clever but still flat texture tricks...
In other words, the potential to wow a customer is absolutely huge, IMO. It's just a question of who is going to be the first to overdo it, and then who will be the first to truly do it justice.
Likewise with better shadowing techniques made useable (IE - not performance crippling) by Dx11 (Stalker CoP for example). Able to appreciate them in a slower paced game, but would be extremely hard to notice in a racing game.
Regards,
SB
Last edited by a moderator:
Jawed
Legend
Er, no, you're saying that D3D11 "by definition" is inferior: "DX11 by design will be much, much subtler". D3D11 by design is less limited than CUDA, so it can't be inferior if someone chooses to implement similar stuff as seen so far in PhysX. You're confusing what people choose to do, with the inherent capabilities and performance of such usage. "The foreseeable future" is a strawman.
Jawed
Jawed
I wasn't aware that "subtle" translated to "inferior".
I made no comparison between the inherent capabilities of CUDA and CS so you're having that argument with yourself at the moment. The comparison I made was between the effects that we see in games between PhysX and DX11.
If you want to argue that somebody one day in the indeterminate future will code a physics library in DXCS that's fine, but it's pure speculation.
I made no comparison between the inherent capabilities of CUDA and CS so you're having that argument with yourself at the moment. The comparison I made was between the effects that we see in games between PhysX and DX11.
If you want to argue that somebody one day in the indeterminate future will code a physics library in DXCS that's fine, but it's pure speculation.
You read something that wasn't there and it's my fault?
Yep, it'll be subtle things like that and having real doorknobs and better weapon models etc. Things that make the overall scene better that won't necessarily jump out at you immediately but will be sorely missing if you go back and play older non-tessellated games.
Yep, it'll be subtle things like that and having real doorknobs and better weapon models etc. Things that make the overall scene better that won't necessarily jump out at you immediately but will be sorely missing if you go back and play older non-tessellated games.
What? I can't understand...
December 2006: nVidia presents DX10 on Conan. The presentation says: "more polygons, better shadows":
December 2009: ATi presents DX11 on Dirt 2. Review says: "better shadows and more polygons":
What am I missing? What is ATi doing wrong with Dirt?
Why would I need to explain something I never said?
Of course, if someone codes a physics library in CS it will be the same as PhysX. Thought we covered that a few posts ago. Around the same time that I pointed out that DX11 is basically tessellation and faster post-processing for the forseeable future.Once your fabled CS based physics engine is available then maybe there'll be a useful comparison to be made.
Jawed
Legend
Good, you agree your posting was moot. You were somehow artificially restricting what developers will choose to do with D3D11, asserting it's the design of D3D11 that creates this restriction, making for only subtle improvements in rendering.
For what it's worth, I reckon most of the D3D11 stuff in Dirt 2 is not very exciting and visually less compelling than physically-interacting smoke effects or objects flying around - it's definitely subtle stuff tacked-on to the game. The physics of rocks that go flying when the car hits them are particularly bad.
None of those failings/shortfalls are some restriction of D3D11's design. The most you can say is that its newness means we'll be waiting a while. Perhaps shorter than the years and years it took for anything remotely interesting to happen in games because of PhysX.
Jawed
What am I missing? What is ATi doing wrong with Dirt?
They should of put more polygons first
- Status
Similar threads
