Shader Models 2 and 4
- Thread starter Reverend
- Start date
Technically not, really.
As has already pointed out in this thread, newer Shader Models encompass older so the profileration of SM3.0 boards will at the very worst stagnate the pool of SM2.x capable boards, but most likely increase it.
DaveBaumann said:
Oh. So every 3.0 board is also a 2.0 board, so "technically" the 3.0 boards don't "catch up" in absolute numbers (tho they do in percentage).
Hmm hmm hmm. There also isn't any reason to think that 2.0 performance will take a hit at some point on newer boards, right? Well, I guess we'll have to see if ATI lives up to their oft-stated contention that FP24 wasn't a performance issue for them.
BOOMEXPLODE
Regular
All those "SM3" consoles won't really facilitate anything in the PC space IMO. They're all using CPUs from IBM and the games will require major re-working to function in the DX/WGF world. They will use shaders, but the platforms will be so different from a PC that it hardly matter what shader model they're using.
Perhaps what Dave is saying is from the aspects of board prices and developers doing extra bits of work (fallbacks as well as separate SM paths).geo said:
Personally, I think when the next-gen consoles are out and ATI debuts their SM3 boards, SM2 dev support will go down quickly, possibly even vanish within a year (of the debut of the next-gen consoles). And SM4 may likely suffer the fate of SM1.4 -- an in-between-SM1.1-and-SM2.
Reverend, IMHO you can not compare the DX8 situation with today. In this past days you have to write each shader with the assembler like language. Unfortunately PS 1.4 use a complete completely different model as PS 1.1-1.3. If you want 1.1 shader and 1.4 shader you have to write at least two shaders for each effect.
Today we have high level languages and this makes it much easier to write one shader and use it for more than one shadermodel. OK there is a high probability that a shader that compile for SM3 will not compile for SM2/2.X. But it should generally easier to change the high level code to make it fit with an older SM than a whole rewrite.
This highlevel languages finaly conquer the console development. This will make it much easier to convert a game to the pc. The shadermodels that this games used on the pc will IMHO highly depend on the shaders that the game used. If this shaders compile for a SM2/2.X model the game will use it. If the need a SM3 profil to make them fit it looks bad for a SM2 path in the near future. But on the pc market publisher always care about customer base. The SM3 base is still to small. This makes it necessary to support a shadermodel below SM3 to improve the customer base. At the moment there are still many SM1.X hardware out there. Because of this pulisher will prefer that a developer include shader for this model. But as time goes by more and more SM1.1 hardware will disappear and SM2 will get the postion of the lowest technology you have to support as a developer. Technology support is allways a moveing window. You have to support at least the technology on both sides. The lowlevel for a large customer base and the highlevel for the eyecandy to impress the gametester. Everything between this two is luxury.
Today we have high level languages and this makes it much easier to write one shader and use it for more than one shadermodel. OK there is a high probability that a shader that compile for SM3 will not compile for SM2/2.X. But it should generally easier to change the high level code to make it fit with an older SM than a whole rewrite.
This highlevel languages finaly conquer the console development. This will make it much easier to convert a game to the pc. The shadermodels that this games used on the pc will IMHO highly depend on the shaders that the game used. If this shaders compile for a SM2/2.X model the game will use it. If the need a SM3 profil to make them fit it looks bad for a SM2 path in the near future. But on the pc market publisher always care about customer base. The SM3 base is still to small. This makes it necessary to support a shadermodel below SM3 to improve the customer base. At the moment there are still many SM1.X hardware out there. Because of this pulisher will prefer that a developer include shader for this model. But as time goes by more and more SM1.1 hardware will disappear and SM2 will get the postion of the lowest technology you have to support as a developer. Technology support is allways a moveing window. You have to support at least the technology on both sides. The lowlevel for a large customer base and the highlevel for the eyecandy to impress the gametester. Everything between this two is luxury.
I suppose the keyword you chose to use is "is". I didn't use "is".Demirug said:
Everything else you wrote needs more clarification from you regarding the topic of "timeframe" (i.e. your "is"). I took the risk of providing a "date" of when I foresee the death of SM2 dev support. Would you take this risk and give me yours?
The next generation consoles are going to take a number of years to ramp up to significant volumes - there are far more DX9 SM2.0 enabled boards in PC's than there will be next gen consoles for a few years. Again, I would say that it would be highly premature to conclude that next gen consoles = current PC SM3.0.
http://www.beyond3d.com/forum/viewtopic.php?p=483920#483920
http://www.beyond3d.com/forum/viewtopic.php?p=483920#483920
DaveBaumann said:
Obviously it will take some time, but at least we wont be in a situation like today, when major games like Doom3 and HL2 are still not using SM2 properly, even though SM2 has been out for years.
Once the consoles come out, LOTS of games will be made around the consoles specs, which happen to be SM3 more or less, and that can only be a good thing.
This generation, there was one console that was compatible with PS1.3 maximum. And that console isn't even the market leader, therefore it's normal that games in this generation look like they were built around a system that doesn't even have pixel shaders.
Next gen, there will be 3 consoles, all compatible with SM3 (or more). Ignoring this fact sounds a bit short-sighted.
Whether the CPUs are different or not, that never stopped Mac conversions of games from working just fine (more or less). What's important is that SM3 will be the lowest common denominator, even though SM4 will be out already by the time the dust has settled. But at least we will have games with advanced shaders, and LOTS of geometry, unlike today where most games take advantage of 5-year old standards
Whats the definition of "properly"?
No, its short sighted to assume that somehing that makes full use of next gen consoles will equate to SM3.0.
This whole thread is full of assumuptions and generalities that bears very little reflection on reality - even when the next gen consoles come out there are going to be titles that or ported from the old to the new consoles, which will take many years to shake off, and the current titles being coded specificall for Xenon have been done so on SM2.0 hardware.
IMO, one of the things that we will certainly see is that as the coding for the next gen consoles take prevalence the base level of geometry for cross ported (PC <-> next gen consoles) will increase, but effects and other utilisations will al be down the timings, priorities, targets, etc., etc. of the publisher and developer at hand.
DaveBaumann said:
Meaning the engines out today are not built with SM2 in mind, they are built around older models, and then get SM2 features added on top, like icing on the cake. And AA and AF to make people happy about their recent purchases. Whereas the cake should be all a triple chocolate brownie.
Today, most multiplatform games are built around the market leader, PS2, which doesn't have pixel shaders at all. Engines are built around it, then maybe get higher res textures and some bump mapping when they get converted to PC or Xbox. And it shows. Next gen, whatever console is the market leader, it will be compatible with SM3. Engines will be built around it. Sounds pretty good to me.
DaveBaumann said:
Of course, no one knows what will happen, all i'm saying is that having SM3 as a lowest common denominator is A Good Thing.
The devs can do whatever they want, in the end there wil still be a lot of games taking advantage of SM3, even if some will just go the lazy way.
Reverend said:I suppose the keyword you chose to use is "is". I didn't use "is".Demirug said:
Everything else you wrote needs more clarification from you regarding the topic of "timeframe" (i.e. your "is"). I took the risk of providing a "date" of when I foresee the death of SM2 dev support. Would you take this risk and give me yours?
Sure I will takeing this risk because it can not cost me much.
The general customer that still build the the largest part of the gamebuyers updates the hardware in intervals around 2 years. As there are still sold SM2 hardware in large quantitys today developers can not phase out SM2 support before most of this customers have update their systems. Because of this I predict that the games for holiday season 2006 will still contains SM2 support. Even in 2007 most games will support SM2 hardware. PC gameing has a long tradition in supporting old technologies. How long did it take before developers drop support for non shader hardware?
Similar threads
- Locked
