Doomtrooper
Veteran
Rican said:
There is some key differences here, you will actually be able to buy a R420...and won't require a new power supply.
first r420 review leak
- Thread starter Ulukay
- Start date
There is some key differences here, you will actually be able to buy a R420...and won't require a new power supply.
Ailuros said:
Why? there actually fairly little point to it now...
"A first german review:
http://www.chip.de/artikel/c_artikel_11828619.html"
(originally posted by Exxtreme , here: http://www.rage3d.com/board/showthread.php?s=&postid=1332817355#post1332817355 )
http://www.chip.de/artikel/c_artikel_11828619.html"
(originally posted by Exxtreme , here: http://www.rage3d.com/board/showthread.php?s=&postid=1332817355#post1332817355 )
Regarding the usefullness of Shader Model 3.0 as far as being 'new':
My thoughts on the matter exactly!
HLSL can reduce the new hardware feature
game engine use time span a whole lot. Which is a wonderful thing as games will start to use shaders to much greater extent and hardware is powerful enough to advanced shaders.
I would have preferred that we didn't have to bother with anything but pure Shader Model (1) 2 and 3 but the NV3x and R420 are only a profile and compile anyway to be put to used ... so whatever.
DemoCoder said:
My thoughts on the matter exactly!
HLSL can reduce the new hardware feature
game engine use time span a whole lot. Which is a wonderful thing as games will start to use shaders to much greater extent and hardware is powerful enough to advanced shaders.I would have preferred that we didn't have to bother with anything but pure Shader Model (1) 2 and 3 but the NV3x and R420 are only a profile and compile anyway to be put to used ... so whatever.
anaqer said:
Which means that it won't save fill-rate in the end?
Then why not reduce to 4xAF instead of 2xAF on "weird" angles? Probably the same reason why I have to force trilinear on all texturing stages in various cases and experience another performance drop.
Ailuros said:
To paraphrase one of the gems fom N.O.L.F. : "A correlation doesn't necessarily imply causality." AFAIK bandwith saving was more of an (admittedly very useful) side effect of, and not the reason for angle dependant AF. (Remember, back then on .15 micron every single transistor counted big time, but bandwith was plentiful due to the smokin' 256bit memory interface.)
I was refering to future products and pointed at the XBox2 and other related designs for the record. Even worse R420 is based on low-k 130nm.
I'm very well aware for which reasons specific design decisions were made in the past; I'm talking about the future and to finally get rid of the dreaded angle dependancy. I closed one eye in favour of the excellent MSAA algorithm in R3xx, I do not want to do it in the future too.
Especially since NVIDIA adopted a similar method, which is the worst thing that could have happened.
There's a huge difference between an IHVs design decisions and a user knowing or wishing what fits his imagery better. To get the bickering one step further I'm actually being quite modest here; I'd love feline alike AF quality too
Bouncing Zabaglione Bros.
Legend
Chisholm said:
Wow!
You certainly get a lot more for your money with NV40! Another piece of commonplace : history repeats itself. With the increased number of pipelines, added functionality, etc. they might run out of transistors with the XBox2 chip too. Not very likely, I know, but you can't just disregard past obstacles 'cause they tend to bite you in the back when you least expect it.Ailuros said:
Sorry, I don't really get what you are suggesting...Ailuros said:
I just noted that now the R420 can do 32x0 just like the NV40, so that's no longer a trump card for nVidia.
| EDIT :does the "scales with number of pixel pipelines" part refer to the "significant performance gainat high resolutions" bullet only, or the others too? |
Similar threads
- Locked
