ATI's GDC surprise [LOTS of good future info] **Updated 4/1**

Toggle sidebar Toggle sidebar
Don Vito Corleone

Don Vito Corleone

Elite
Feb 10, 2000
30,029
67
91
Veddy interesting . . . I have to think that as a general proposition the graphics-card guys must be psyched about the arrival of HL2/D3, as well as some demanding titles already available (Far Cry, Painkiller), which will finally drive demand for new hardware. To this point I have felt essentially no pressure to upgrade my 1 1/2-year-old Radeon 9700, or even my 2-YO CPU.
 
A

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
Only very recent games even push my FX5600

But battlefield Vietnam is very demanding on my system, im also looking forward to doom 3, and late life 2.

And of course the inevitable Counterstrike 2 :)
 
A

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
Originally posted by: Eug
Cliff notes for GPU hardware n00bs?
Click to expand...

ATi wants to push 512MB cards mainstream by next year.
they want to downplay the prediction branch features of DX9 because the performance hit will be huge on ATi hardware
A lot of programming advice for graphics (this was obviously intended to be presented to devs imho)
ATi Expects DX9.0c to run all the way to longhorns release sometime in 2006, when DX next will be released.
They also try to downplay NVIDIA cg over HLSL, although their own chart they show clearly shows NV cg being more successful, which confused me a bit.

Thats about all for the meat of it.
 
S

scooter1

Member
Dec 13, 2003
71
0
0
Originally posted by: Acanthus
Originally posted by: Eug
Cliff notes for GPU hardware n00bs?
Click to expand...

ATi wants to push 512MB cards mainstream by next year.
they want to downplay the prediction branch features of DX9 because the performance hit will be huge on ATi hardware
A lot of programming advice for graphics (this was obviously intended to be presented to devs imho)
ATi Expects DX9.0c to run all the way to longhorns release sometime in 2006, when DX next will be released.
They also try to downplay NVIDIA cg over HLSL, although their own chart they show clearly shows NV cg being more successful, which confused me a bit.

Thats about all for the meat of it.
Click to expand...
Is there going to be a DX9.0c for sure?

 
A

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
Originally posted by: scooter1
Originally posted by: Acanthus
Originally posted by: Eug
Cliff notes for GPU hardware n00bs?
Click to expand...

ATi wants to push 512MB cards mainstream by next year.
they want to downplay the prediction branch features of DX9 because the performance hit will be huge on ATi hardware
A lot of programming advice for graphics (this was obviously intended to be presented to devs imho)
ATi Expects DX9.0c to run all the way to longhorns release sometime in 2006, when DX next will be released.
They also try to downplay NVIDIA cg over HLSL, although their own chart they show clearly shows NV cg being more successful, which confused me a bit.

Thats about all for the meat of it.
Click to expand...
Is there going to be a DX9.0c for sure?
Click to expand...

Yes for PS 3.0.
 
S

scooter1

Member
Dec 13, 2003
71
0
0
Originally posted by: Acanthus
Originally posted by: scooter1
Originally posted by: Acanthus
Originally posted by: Eug
Cliff notes for GPU hardware n00bs?
Click to expand...

ATi wants to push 512MB cards mainstream by next year.
they want to downplay the prediction branch features of DX9 because the performance hit will be huge on ATi hardware
A lot of programming advice for graphics (this was obviously intended to be presented to devs imho)
ATi Expects DX9.0c to run all the way to longhorns release sometime in 2006, when DX next will be released.
They also try to downplay NVIDIA cg over HLSL, although their own chart they show clearly shows NV cg being more successful, which confused me a bit.

Thats about all for the meat of it.
Click to expand...
Is there going to be a DX9.0c for sure?
Click to expand...

Yes for PS 3.0.
Click to expand...
Any idea when it's going to be released?

 
K

KristopherKubicki

Golden Member
Jul 31, 2002
1,636
0
0
OK,

Well i just wanted to address my thoughts before this got too out of hand. As much as i would like to beleive this is true, approaching this cynically there are some very large inconsistancies. Primarily, the only existing mirror was editted before it was posted as a mirror. Here is my suspicion.

The original thread starts "Posted: Mon Mar 29, 2004 9:12 pm Post subject: DirectX Next - Longhorn only !?" This is 9:12PM in Englad, -1 GMT.

A "mirror" was posted here: http://pwp.netcabo.pt/0239863201/Save_The_Nanosecond.ppt Notice the Portugal host.

Looking at the document posted as a mirror (left) and two other documents posted by Richard In the last year.

http://images.anandtech.com/reviews/video/rhuddy.gif

It appears the "mirror" version of the PPT was editted a 3/39 at 3:19PM. Looking at the post on beyond3d by the mirror poster:

http://images.anandtech.com/reviews/video/beyond3d.gif

The presentation format on rhuddy.gif claims the presentation format is no longer "On-Screen Show", but "Apresentacao no ecra" (loosely). The document has been edited by a portugeuse speaking version of PPT. This concurs with the .PT host.

However, look at the world time zone map:

http://aa.usno.navy.mil/faq/docs/world_tzones.html

Portugal is 1 hour difference from England. Brazil is at most 4. It raises enough suspicion that the document was saved again. It raises further suspicion as to how did "felix" last save this document 6 hours before it was posted on Beyond3d? It seems possible his clock was wrong; but it also raises question to the authenticity.

Kristopher
 
A

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
Originally posted by: KristopherKubicki
OK,

Well i just wanted to address my thoughts before this got too out of hand. As much as i would like to beleive this is true, approaching this cynically there are some very large inconsistancies. Primarily, the only existing mirror was editted before it was posted as a mirror. Here is my suspicion.

The original thread starts "Posted: Mon Mar 29, 2004 9:12 pm Post subject: DirectX Next - Longhorn only !?" This is 9:12PM in Englad, -1 GMT.

A "mirror" was posted here: http://pwp.netcabo.pt/0239863201/Save_The_Nanosecond.ppt Notice the Portugal host.

Looking at the document posted as a mirror (left) and two other documents posted by Richard In the last year.

http://images.anandtech.com/reviews/video/rhuddy.gif

It appears the "mirror" version of the PPT was editted a 3/39 at 3:19PM. Looking at the post on beyond3d by the mirror poster:

http://images.anandtech.com/reviews/video/beyond3d.gif

The presentation format on rhuddy.gif claims the presentation format is no longer "On-Screen Show", but "Apresentacao no ecra" (loosely). The document has been edited by a portugeuse speaking version of PPT. This concurs with the .PT host.

However, look at the world time zone map:

http://aa.usno.navy.mil/faq/docs/world_tzones.html

Portugal is 1 hour difference from England. Brazil is at most 4. It raises enough suspicion that the document was saved again. It raises further suspicion as to how did "felix" last save this document 6 hours before it was posted on Beyond3d? It seems possible his clock was wrong; but it also raises question to the authenticity.

Kristopher
Click to expand...

So the question i guess would be, where is the unedited original, or is it hoaxed. (it would take someone with a strong knowledge of graphics design and implementation to create that document).
 
K

KristopherKubicki

Golden Member
Jul 31, 2002
1,636
0
0
The document itself could be real, but i am curious if the COMMENTS (which are the incriminating parts) are real.

Kristopher
 
A

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
Originally posted by: KristopherKubicki
The document itself could be real, but i am curious if the COMMENTS (which are the incriminating parts) are real.

Kristopher
Click to expand...

I would agree with that as the comments seem less professional than the presentation itself.
 
W

Wingznut

Elite Member
Dec 28, 1999
16,968
2
0
Originally posted by: Acanthus
(this was obviously intended to be presented to devs imho)
Click to expand...
Actually... It looks like an internal presentation for someone who is preparing to make a presentation. Hence, the "comments" were never intended for the actual audience.
 
zephyrprime

zephyrprime

Diamond Member
Feb 18, 2001
7,512
2
81
although their own chart they show clearly shows NV cg being more successful, which confused me a bit.
Click to expand...
I think you misread it. The chart showed HLSL owning CG.
 
J

jjyiz28

Platinum Member
Jan 11, 2003
2,901
0
0
Originally posted by: Acanthus
Originally posted by: KristopherKubicki
The document itself could be real, but i am curious if the COMMENTS (which are the incriminating parts) are real.

Kristopher
Click to expand...

I would agree with that as the comments seem less professional than the presentation itself.
Click to expand...

how do i view the comments, sorry for newb ?
 
S

Sunner

Elite Member
Oct 9, 1999
11,641
0
76
Originally posted by: Wingznut
Originally posted by: Acanthus
(this was obviously intended to be presented to devs imho)
Click to expand...
Actually... It looks like an internal presentation for someone who is preparing to make a presentation. Hence, the "comments" were never intended for the actual audience.
Click to expand...

I agree, that looks like an internal "sales talk", where the suits love to tell the rest of the company how fine and dandy everything is using everything from buzzwords and smileys to a piece of real information here and there.

Im sure you have them over at Intel too ;)
 
RussianSensation

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
2 Important quotes:
"Higher (64bit) precision is not for the near-term...."
"64 bit may well be required in the first new DX after Longhorn"

But longhorn will not come until 2006, so technically wouldn't this mean we will not be required to use 64-bit until then? or essentially the TRUE benefits of having 64-bit instructions will not take place until the whole industry slowly shifts there. If it takes 2 years for software game developers to catch up to hardware and make the software the way it was meant to be played 2 years ago, then A64 in 2004 would precisely land the 64-bit mainstream and acceptance and strong push only by 2006 which is when Longhorn will come anyway. Still A64 is an awesome processor and it is fast but the 64-bit computing strategy isn't something which will substantially sway the purchase away from Intel in my opinion (maybe i am wrong?), but in 2 years the 3000+ A64 will be just too slow anyways. Of course windows xp 64-bit is supposed to appear soon. Now, I am not saying buying A64 is a bad idea, I think I would buy it if i was going to upgrade right now, but I think people shouldn't base this purchasing decision on the 64-bit feature and praise it so much because by the time 64-bit OS (+drivers etc.) becomes more bug free, and more than 50% of games will take advantage of it, I do not think 64-bit is any advantage whatsoever. And in 2 years when it does pose the advantage faster hardware will have come out. Of course if you think this way you'll never make progress, but its just my 2 cents for TODAY.
 
D

DerekWilson

Platinum Member
Feb 10, 2003
2,920
34
81
Originally posted by: RussianSensation
2 Important quotes:
"Higher (64bit) precision is not for the near-term...."
"64 bit may well be required in the first new DX after Longhorn"

But longhorn will not come until 2006, so technically wouldn't this mean we will not be required to use 64-bit until then? or essentially the TRUE benefits of having 64-bit instructions will not take place until the whole industry slowly shifts there. If it takes 2 years for software game developers to catch up to hardware and make the software the way it was meant to be played 2 years ago, then A64 in 2004 would precisely land the 64-bit mainstream and acceptance and strong push only by 2006 which is when Longhorn will come anyway. Still A64 is an awesome processor and it is fast but the 64-bit computing strategy isn't something which will substantially sway the purchase away from Intel in my opinion (maybe i am wrong?), but in 2 years the 3000+ A64 will be just too slow anyways. Of course windows xp 64-bit is supposed to appear soon. Now, I am not saying buying A64 is a bad idea, I think I would buy it if i was going to upgrade right now, but I think people shouldn't base this purchasing decision on the 64-bit feature and praise it so much because by the time 64-bit OS (+drivers etc.) becomes more bug free, and more than 50% of games will take advantage of it, I do not think 64-bit is any advantage whatsoever. And in 2 years when it does pose the advantage faster hardware will have come out. Of course if you think this way you'll never make progress, but its just my 2 cents for TODAY.
Click to expand...

RussianSensation:

Well, your thoughts and observations on their own aside: you've made an improper assumption. In the PDF in question, ATI is absolutely not talking about 64bit wide CPU architectures, but rather a pixel/vertex shader architecture on a GPU that supports 64bit floating point internally...

doubles could be useful in GPU hardware, but they'd be massively slow, painful and not at all useful or exploitable right now. you would end up using 256bits per pixel (as opposed to 128 bits per pixel with 32bit floating point)... that would KILL bandwidth (which is already pushed to the max with any floating point pixel data (or even fx12 in some cases)).

Anyway, the point is: watch out for taking things out of context ... and we still don't have confirmation that this is real yet (to my knowledge).
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom