Your New Year Wish

Razor1 said:
Ogl will not be screwed over in Vista
Are you saying that because fullscreen will still be 100% functional? If so, may I remind you that a number of OpenGL applications, such as some content creation ones, benefit highly from being in windowed mode? And that this would be a major disadvantage?
If you imply something else, please do enlighten us! :)

Uttar
 
Since i´ve been pretty impressed by advance in technology in the GPU sector in particular, there currently isn´t any concrete wish i´m having, but i can still think of the magic 3:

1. INQ, esp. Fuad, getting some brains (did i just write that ? :LOL: )
2. Competition at the Ultra-High-End to settle down somewhat, cause it´s getting quite ridiculous where we are now (i won´t bet my money on it just yet, cause next year isn´t gonna be exactly the year we can expect that to change (..))
3. Better games instead of just remaking everything again and again and again... (doubtful)

PS: I´m wishing everyone a happy new year in advance.
 
Last edited by a moderator:
Uttar said:
Are you saying that because fullscreen will still be 100% functional? If so, may I remind you that a number of OpenGL applications, such as some content creation ones, benefit highly from being in windowed mode? And that this would be a major disadvantage?
If you imply something else, please do enlighten us! :)

Uttar


Well earlier rumors of Ogl being held back to dx9 specs, thats what I'm worried about. I would like to see Ogl still as it is now in fullscreen, in exclusive mode is what I mean.
 
A slight theme with my reply... but I couldn't resist commenting...

That MS realizes that they are stupid not making D3D10 backward compatible with D3D9 cards.
Huh? it's called going forwards - D3D9 (and D3D9Ex) are gonna be around for a while, so you'll get "dual" engines. Being a software engineer I'm really looking forward to dropping this mish-mash and general mess that a multi-version runtime/CAPS brings. D3D10 is a distinct breath of fresh air :cool:

Not to have to wait until 2008 before we see any SM4 games that truly take advantage of the tech.
I don't have the links to hand, but theres a demo/presentation at GDC '06 showcasing some new D3D10 (which implies SM4) game launching with Vista. Might not have to wait as long as you think :smile:

1. SM4 support in DX9.
Can't and won't happen :rolleyes:

the 4-5 years D3D10 is aimed to last
I think you'll find it's DXGI thats intended to last into the 5+ year timeframe. D3D10.0 will certainly still exist, but I doubt it'll be top/current for more than 2-3 years.

That DX9 apps run faster in Vista than they do on WinXP with the same hardware (maybe more system memory).
I'd imagine that "default" D3D9 won't really get any advantages and that D3D9Ex is where any benefits come from. Although, having said that, the new display driver model might well have a "general" speed-up. Particularly when we get hardware context switching in the +1 generation.

Oh, and my wish... that I can get my hands on some D3D10 hardware ASAP - running apps via the reference rasterizer gets a little tedious ;)

Cheers,
Jack
 
JHoxley said:
Although, having said that, the new display driver model might well have a "general" speed-up. Particularly when we get hardware context switching in the +1 generation.

Right, they've talked about overhead being much better. So I didn't think it unreasonable to wish it translates in general.
 
geo said:
Right, they've talked about overhead being much better. So I didn't think it unreasonable to wish it translates in general.
Yup, it probably is reasonable to assume that the new display driver will give some sort of universal speed up... but I think the limitations of the D3D9 runtime will prevent it from reaching the dizzy heights of D3D10.

There is a lot of "create time" validation done in D3D10, which makes for little/no overhead when the actual per-frame (aka high-frequency calls) rendering takes place. Hence the improved small-batch performance and generally improved "flow"..

However, D3D9 under Vista will probably still retain many of the basic architectural "features" of it's current WinXP incarnation. I'm not entirely sure how much they could get away with changing before it suddenly becomes a complete internal re-write...

Jack
 
It is in their own interest tho. First [insert current hot Triple AAA title name here] bench at one of the big sites showing an extra 3fps on Vista will sell tens or possibly hundreds of thousands more copies of Vista (or at least sell them earlier in Vista's lifecycle). Alternately, first 3fps _slower_ bench of same will start the wail of "never in my liftetime" in that same crowd. I mean, we've been here before with Win9x and WinXP.
 
JHoxley said:
Can't and won't happen :rolleyes:
Oh, I can believe that SM4 won't happen in DX9. But I don't believe for a second that it can't happen.

Basically, what I'm hoping for is that game devs don't need to use OpenGL to make use of SM4 functionality for next-gen cards under Windows XP and previous.
 
JHoxley said:
so you'll get "dual" engines

Exactly.
Those engines will be general enough to run on D3D9 and D3D10 so you'll lose most of the efficiency improvements of the D3D10 API.
 
Chalnoth said:
Oh, I can believe that SM4 won't happen in DX9. But I don't believe for a second that it can't happen.
SM4 is tightly bound to the new D3D10 pipeline - there are various parts of D3D9 that just don't expose the same level of information. Getting a ps_4_0/vs_4_0 on D3D9 wouldn't really have any advantage without changing the conceptual D3D9 pipeline. So, yes, I suppose you could argue that it's possible but there's no sane reason why it'd be worth doing - you'd just end up with some strange D3D9/10 hybrid API.

For example, SM4 has a much more relaxed resource binding/access/writing system - texture reads from the VS (present in SM3) are only the beginning there.

However, what I would expect to see is that the first D3D10 compliant devices will have outstanding D3D9-SM3 support. The only "if" here is whether the IHV's choose to release/support decent D3D9 drivers for their D3D10 parts.

If you translate D3D10's baseline caps down to D3D9 then you'll end up with an awesome piece of kit :cool:

Chalnoth said:
Basically, what I'm hoping for is that game devs don't need to use OpenGL to make use of SM4 functionality for next-gen cards under Windows XP and previous.
I see what you're getting at... but I'd imagine that it'll be the hybrid system that I mentioned - similar to how you get your "fallback" systems for D3D7/D3D8 hardware.

As an ISV I would honestly be wondering if the time and effort to try and "hack" D3D10 features (such as the GS) into WinXP/D3D9 is worth it. Surely most of those people that have a decent D3D10 part will probably also have Windows Vista and thus be running your engine in "D3D10 mode" anyway. That small percentage of people that have a D3D10 part and still run it under XP/D3D9 will get graphics equivalent to the latest-n-greatest we're seeing now (full SM3 glory).

Jack
 
Hyp-X said:
Exactly.
Those engines will be general enough to run on D3D9 and D3D10 so you'll lose most of the efficiency improvements of the D3D10 API.
Yeah, I'd guess that'll be a problem to start with. Purely from a software design and architecture viewpoint it's difficult to get "best of both worlds" - you have to try and strike a happy medium and that'll probably mean compromises in both directions...

Jack
 
JHoxley said:
As an ISV I would honestly be wondering if the time and effort to try and "hack" D3D10 features (such as the GS) into WinXP/D3D9 is worth it. Surely most of those people that have a decent D3D10 part will probably also have Windows Vista and thus be running your engine in "D3D10 mode" anyway. That small percentage of people that have a D3D10 part and still run it under XP/D3D9 will get graphics equivalent to the latest-n-greatest we're seeing now (full SM3 glory).
Fine. But my point is that it's another blatant monopoly bid by Microsoft. I'd be okay with upgrading to Vista if it adds new, real value. But I will be exceedingly pissed-off about being forced to upgrade due to software support.
 
Have to agree with Chalnoth here .. if i'm gonna buy a new SM3.0 part .. i'd like it to last at least 2-3 years before replacement.

That's why I think ATi giving us the R580 late, yet early is a good choice. It should've been an option 2 months ago though.

US
 
Back
Top