Let's forget Fermi and its tessellation performance for a minute. What was ATi really thinking when they set out to implement DX11 tessellation? After all those years of pushing tessellation only to get blue-balled at every turn they should have come out with a bang. Instead we got a whimper and nary an impressive tessellation demo in sight. Were they just testing the waters and Cayman will be the hotness or are they in fact not as hot for tessellation as they claim to be?
Tessellation
- Thread starter trinibwoy
- Start date
Perhaps they wanted tessellation to be a useful option for enhancing games as opposed to some epeen inflating synthetic that allows you to make more triangles than is in anyway useful.
any amount of tessellation that current AMD cards, doesn't impress at all, if you want a few rounded corners we had that with trufrom and that could be done with geometry shaders too. Tessellation was added to DX11 for increased mesh detail for realism, not to have have to use POM or parallel bump mapping which look pretty much flat from certain angles or have artifacts not to forget shader aliasing.
Guess you missed the part where I said to ignore Fermi. Got an example of useful tessellation to share with us that enhances a game and only requires AMD level performance? The only tangible examples of tessellation usage I've seen are HAWX2 terrain and nVidia's alien demo. Stone Giant might be another candidate.
Hawx 2 is well above 60 fps in AMD cards... what's your point?
60 or 200 fps... what's the difference(gameplay)?
Seems like you're making a pretty good argument that more tessellation isn't important.
I would say AMD's architectures fare better on synthetic benchmarks than Nvidia's, all things considered. See, for example, the Beyond3d Fermi review, where on low-level synthetic benchmarks, Cypress repeatedly destroys GF100. Or look at floating-point performance: HD 5870: 2.72 TFlops, GTX 580: 1.5 TFlops. Clearly the 5870 has way more floating point throughput than is in anyway useful...
The tessellation capbilities of all out there are both more capable and higher performance than XBOX 360's capabilities. Improvements are easily attained over titles that have no tessellation and those that use 360 levels of tessellation.
While not a game, take a look at the differences in Heaven between off, moderate and extreme.
That just indicates the mechanism that we have implemented shader performance is relatively efficient in an area sense. Perf/mm in games is still higher, and the math capability is one mechanism for getting that (because less time per frame is spent being bottlenecked there).
Try reading it slower next time.
I'm not really interested in your efforts to portray AMD's tessellation performance as good enough or nVidia's as overkill since you weren't able to come up with any examples to support that position. For something that has been in hardware for so long you'd think somebody had a vision for what it would bring to games yet the implementations so far have been sadly lacking in visual impact. Where is the "OMG that's awesome" stuff?
@Dave, sure it's better than the 360 but what do we need to do to actually be wowed by tessellation? Everyone says we don't need more triangles so what's missing? Artist creativity?
If you were expecting tessellation to be some holy grail for gaming, you were always destined to be disappointed. It's just another weapon in the arsenal.
Why would you be "wow"ed by any single feature? Were you wowed by a unified pipeline? Were you wowed by GS? Where you wowed by PS2.0->PS3.0? PS3.0->PS4.0? etc.? Every progression is about providing more control and options to the developer, much like all the other functionality that was added to DX11 and each generation of DX before it and as developers increasingly explore the benefits of it the end results will increasingly get better as well - some will get more out of tessellation others will ger more out of DirectCompute, and others out of the general increased flexability.
All those were incremental changes to the shading/texturing pipeline over several years. Tessellation is the first improvement on the geometry side of things in a long time. Expectations should be higher and in line with the promises. If AMD's stance is now that tessellation is no big deal then that would explain their actions thus far.
That was one of the main features of DX 11, outside of compute. I agree each gen of API gives more control of options, but I think Trini is getting at AMD/ATi has had tessellation functionality alot longer and they didn't really push it when nV came out with their first version of tessellation and it was much more superior and actually more usable, to what offline renderers can provide.
Geometry Shader was implemented in DX10.
We demo'ed Tessellation and massive geometry levels per frame back with R600.
caveman-jim
Regular
In what aspect is it 'much more superior' and 'more usable'?
What's wrong with AMD tessellation?
Duck Dodgers
Newcomer
Hang about, your advertising asks us to be "wowed" by these features with every generational launch and historically it has been used to differentiate and sell competing products.
People are probably confusing tessellation hardware implementation / performance with the frustration of the lack of software implementation, because there is still isn't any huge entertainment value in it.
In an ecosystem where entertainment software development spans years, millions of dollars and multiple platforms perhaps tessellation is the red-headed stepchild, I don't know. What I do know is that I would prefer less spin and more value for money when I buy this technology instead of silicon idling or getting a workout now and then via a demo or a game or three.
I think tessellation won't that that useful on any hardware until games require a DX11 card as a minimum instead of a DX9-10. The geometry levels of models in current games are mostly good enough and tessellation just improves it slightly. There's nothing to get excited about there.
Right now, detailed models are turned into super detailed models, but that's not what tessellation should be for. It should be taking low and medium base models and turning them into super detailed ones and the substantial memory savings that come with it.
I also think AMD tessellation is probably good enough, and what Nvidia touts is just to win framerate wars with inefficient use of unneeded, extreme tessellation.
Right now, detailed models are turned into super detailed models, but that's not what tessellation should be for. It should be taking low and medium base models and turning them into super detailed ones and the substantial memory savings that come with it.
I also think AMD tessellation is probably good enough, and what Nvidia touts is just to win framerate wars with inefficient use of unneeded, extreme tessellation.
Jawed
Legend
The answer's bleeding obvious: Evergreen is a significantly feature-/performance-reduced chip from what was originally planned. AMD kludged the whole thing. That's why it's so inefficient (not just in tessellation, but pretty much everything). See Barts for a pointer on the inefficiences.
As to whether Cayman will be a minor or major improvement (on everything, not just tessellation), well, I guess my first sentence hangs in the balance
Cayman should be more than Cypress was ever planned to be (1 year of extra development), which will distort comparisons, but I don't for one second believe that Evergreen's tessellation architecture is as it was originally planned. It's just horrible.As to why Cypress wasn't Barts specification, who knows. AMD could have made much more effective use of 40nm - what turned out to be very limited supply of 40nm - for a negligible performance loss. Seems like another indication that Cypress was a hatchet job.
Similar threads
- Locked