R
Radeon350
from Deano Calver on Beyond3D:
http://www.beyond3d.com/forum/viewtopic.php?t=7875
"Dave is the journo so I let him to the real reporting but there were
a few things that caught my eye/ear. As a non-NDA event the
presentations were public infomation (of course, any gossipping over a
pint isn't for public consumption ).
ATI gave some infomation on video cards market pentration data, and
had a quick gloat about being in both Microsoft's and Nintendo's next
gen console, no real info on them (obviously) except for one tiny
snippet, that we should expect at least Dx9 level shaders on both.
Wether or not its the same basic core being used in both they didn't
say.
Microsoft showed off early version of there new graphics tools. GPA
(Graphics Performance Ananlyser) is a tool that can sample whats going
on in D3D. Its meant for performance monitoring while developing, but
it may also be a handy way of looking inside other D3D applications.
Also a D3D Helper library should make developing easier, by shadowing
the entire D3D state, when debugging you can effectively 'see through'
the COM interface to what D3D is actually doing. Its also includes
stats and simple routines to dump states to screen, warning and
erronous states are checked and generally alot of things to make D3D
nicer to develop with. Also they are about to release the SDK update,
a non-runtime update that includes lots of new documentation, D3DX
functions and exporters. 3D Studio MAX 6 has been upgraded to
integrate into D3D effects closely, combined with a new exporter it
should improve artist work flow.
ATI did more presentations on image-space post-processing. Re-covering
some of the same ground they've done before, it did have some
extensions of Masaki Kawase light steaiking filters and I hadn't seen
the heat haze filters before. Microsoft gave a quick overview to
Precomputed Radiance Transfer and Spherical Harmonics, while the
technology is amazing convincing people to use it in real games seems
to be the real problem. Lots of people were very negative on any
actual non-demo use, personally I can see it being very useful but
initially just the spherical harmonic irradience not the full on
precomputed radiance transfer (though PRT for subsurface scattering
objects would seem to make a lot of sense...). Intel did some talks on
SSE, PNI (Prescott New Instructions) and Hyper threading, Intel were
pushing HT as the way forward. A possible 20% extra performance on HT
P3 today, 30% when Prescott is released, next one after that 40% and
the long term aim is for 2x/4x performance.
The most interesting talk in Beyond3D terms was Mike's very general
talk about the next Direct3D. I'll let Dave do it properly but in
general it was primitive programmability and single resource model
(texture and vertex data interchangable etc)."
also, discussion on Gaming-Age:
http://forums.gaming-age.com/showthread.php?threadid=56064
So obviously we should expect DX9 3.0-level shaders in both the next
Nintendo and Xbox. Vertex Shader/Pixel Shader 3.0 at least. This does
NOT mean the next Nintendo will use DX9 or any MS API. It means that
the shaders in the next Nintendo will be DX9-level or better. The
Flipper in GameCube was beyond DX7 but somewhat short of DX8, although
Flipper had its own bag of tricks to do some of the things DX8 could
do. I'm guessing the next Nintendo will do anything DX9 3.0 shaders do
and more, but perhaps will not *quite* be upto DX10 like the next
Xbox. That does NOT mean the Nintendo console will be weak. It just
means the two consoles are being designed with different goals. I am
certain that the Nintendo and XBox 2 VPUs (I'll call them Flipper2 and
XB2-VPU) will each have certain strengths over the other, much like
Flipper and XGPU each had their own strong aspects over the other.
I'd still like to see Nvidia involved in one of the consoles, in some
way. if only for the fact that Nvidia has sooo much IP and many
engineers from the best in the industry.
http://www.beyond3d.com/forum/viewtopic.php?t=7875
"Dave is the journo so I let him to the real reporting but there were
a few things that caught my eye/ear. As a non-NDA event the
presentations were public infomation (of course, any gossipping over a
pint isn't for public consumption ).
ATI gave some infomation on video cards market pentration data, and
had a quick gloat about being in both Microsoft's and Nintendo's next
gen console, no real info on them (obviously) except for one tiny
snippet, that we should expect at least Dx9 level shaders on both.
Wether or not its the same basic core being used in both they didn't
say.
Microsoft showed off early version of there new graphics tools. GPA
(Graphics Performance Ananlyser) is a tool that can sample whats going
on in D3D. Its meant for performance monitoring while developing, but
it may also be a handy way of looking inside other D3D applications.
Also a D3D Helper library should make developing easier, by shadowing
the entire D3D state, when debugging you can effectively 'see through'
the COM interface to what D3D is actually doing. Its also includes
stats and simple routines to dump states to screen, warning and
erronous states are checked and generally alot of things to make D3D
nicer to develop with. Also they are about to release the SDK update,
a non-runtime update that includes lots of new documentation, D3DX
functions and exporters. 3D Studio MAX 6 has been upgraded to
integrate into D3D effects closely, combined with a new exporter it
should improve artist work flow.
ATI did more presentations on image-space post-processing. Re-covering
some of the same ground they've done before, it did have some
extensions of Masaki Kawase light steaiking filters and I hadn't seen
the heat haze filters before. Microsoft gave a quick overview to
Precomputed Radiance Transfer and Spherical Harmonics, while the
technology is amazing convincing people to use it in real games seems
to be the real problem. Lots of people were very negative on any
actual non-demo use, personally I can see it being very useful but
initially just the spherical harmonic irradience not the full on
precomputed radiance transfer (though PRT for subsurface scattering
objects would seem to make a lot of sense...). Intel did some talks on
SSE, PNI (Prescott New Instructions) and Hyper threading, Intel were
pushing HT as the way forward. A possible 20% extra performance on HT
P3 today, 30% when Prescott is released, next one after that 40% and
the long term aim is for 2x/4x performance.
The most interesting talk in Beyond3D terms was Mike's very general
talk about the next Direct3D. I'll let Dave do it properly but in
general it was primitive programmability and single resource model
(texture and vertex data interchangable etc)."
also, discussion on Gaming-Age:
http://forums.gaming-age.com/showthread.php?threadid=56064
So obviously we should expect DX9 3.0-level shaders in both the next
Nintendo and Xbox. Vertex Shader/Pixel Shader 3.0 at least. This does
NOT mean the next Nintendo will use DX9 or any MS API. It means that
the shaders in the next Nintendo will be DX9-level or better. The
Flipper in GameCube was beyond DX7 but somewhat short of DX8, although
Flipper had its own bag of tricks to do some of the things DX8 could
do. I'm guessing the next Nintendo will do anything DX9 3.0 shaders do
and more, but perhaps will not *quite* be upto DX10 like the next
Xbox. That does NOT mean the Nintendo console will be weak. It just
means the two consoles are being designed with different goals. I am
certain that the Nintendo and XBox 2 VPUs (I'll call them Flipper2 and
XB2-VPU) will each have certain strengths over the other, much like
Flipper and XGPU each had their own strong aspects over the other.
I'd still like to see Nvidia involved in one of the consoles, in some
way. if only for the fact that Nvidia has sooo much IP and many
engineers from the best in the industry.
