ATI - Entire family of products in first half 2004

[rwolf]

http://www.ati.com/gitg/promotions/chat/upcoming.html

We generally don't discuss unreleased products but rest assured, the ATI team is cranking away on a broad range of products. We will be releasing an entire family of products in the first half of 2004. While I can't give any specifics, ATI has been the clear technology and performance leader for quite some time and our next generation won't disappoint our fans.

The next process node in the semiconductor industry is 90nm (.09 micron). We are actively working with our fab partners on this technology and expect to have solutions in this technology in the latter part of 2004.

There is lots of talk and rumors about the R420/NV40, but it will be interesting to see what comes out in the low end and the mainstream. I wonder if the midrange and low end products will come out on .11 micron?

 

[Rugor]

Generally they have what can be considered full and half steps: The full steps have been the .18, .13, and upcoming .09 micron processes. The remaining .15 and .11 nodes are more seen as half steps rather than full process nodes. Intel and AMD don't seem to ever release anything on the intermediate processes.

 

[vb]

http://www.beyond3d.com/forum/viewtopic.php?p=201373#201373

http://www.beyond3d.com/forum/viewtopic.php?t=9400

 

[nobie]

So ATI is skipping over 110nm entirely? Interesting, considering Nvidia (supposedly) has plans for a 110nm NV41/45. I wonder if R500 or R550 will be 70nm.

 

[karlotta]

Well, this shouldn't come as a surprise to anybody. ATI has followed nVidia's business model and has "released" a full family of products with every major launch.

Hello! omg! ATI has been around years before nividia showed up.... and NO GPU card maker has ever released a full "family" of products. Not at the same time....at least ATI has the AIW...
Hmmm 420 AIW.

 

[Sage]

speaking of the 420, does anyone else find something a bit funny with their choice of code numbers to assign it?

 

[rwolf]

I think ATI will release the following.

- 12X1 or 16X1 pipe chip at .13 microns
- 8X1 chip at .13 microns
- 4X1 chip at .11 microns
- 2X1 chip at .11 microns

Just a guess.

 

[zsouthboy]

speaking of the 420, does anyone else find something a bit funny with their choice of code numbers to assign it?

I think I get that joke... is it something sexual?

 

[Tim Murray]

speaking of the 420, does anyone else find something a bit funny with their choice of code numbers to assign it?

I think I get that joke... is it something sexual?

no, he might be referring to the GF4 MX420.

or I'm on crack. wheeeee!

 

[Thowllly]

speaking of the 420, does anyone else find something a bit funny with their choice of code numbers to assign it?

I think I get that joke... is it something sexual?

http://www.snopes.com/language/stories/420.htm

 

[KimB]

Hello! omg! ATI has been around years before nividia showed up.... and NO GPU card maker has ever released a full "family" of products. Not at the same time....at least ATI has the AIW...
Hmmm 420 AIW.

Yes, but ATI has not been in the high-end 3D market (successfully) for nearly as long as nVidia has. And nVidia did release a full family of products with the release of the GeForce FX series, from the low-end (5200) to the high-end (5800 ultra). They obviously didn't come out at the exact same time, but we're not talking about that here.

 

[Dave Baumann]

And nVidia did release a full family of products with the release of the GeForce FX series, from the low-end (5200) to the high-end (5800 ultra).

That wasn't intentional though. The schedule was quite obviously compressed due to 5800's lateness.

NVIDIA's policy has previously (and really was so with the FX series) is to release a high end version on new generation technology and the low end close to or before the high end refresh.

 

[Arun]

Dave, I believe that while the schedule was compressed due to the NV30 lateness, it was always NVIDIA's goal to minimize the "latency" between high-end and low-end releases. The NV41 and NV43 (or was it NV42? I *always* take one for the other, since they're the same thing, just different process) should be in stores 6 months after the NV40 is, and that's a maximum.

It was the original plan, and the delay of the NV40 from November to February (not 100% sure if it's still scheduled for February, no sources anymore ) makes it just that much more likely now.

I agree with you that NVIDIA's strategy remains to release the mid-end and low-end cores of the original NVx0 shortly before or at the same time as the NVx5 refresh, though
---

Regarding ATI's H1 2004 releases:
1. R420: Shaders 3.0. high-end
2. ??? mid-end: RV450 ??? - or maybe same thing as with the R9500NP: blocked pipelines ( a 8 pipelines R420 at $249 would be cool IMO )
3. RV381(?not sure about that codename?): Shaders 2.0. low-end, $79.


Uttar

 

[Martin Eddy]

speaking of the 420, does anyone else find something a bit funny with their choice of code numbers to assign it?

I think I get that joke... is it something sexual?

http://www.snopes.com/language/stories/420.htm

That would be cool if that is the reason.

Sort of jams nvidia's"they must be smoking something hallucenegenic!" right back at them.

HA HA HA HA HA HA HA

Maybe i've smoked a little too much.

 

[KimB]

That wasn't intentional though. The schedule was quite obviously compressed due to 5800's lateness.

NVIDIA's policy has previously (and really was so with the FX series) is to release a high end version on new generation technology and the low end close to or before the high end refresh.

And the 5900 would probably have been the high-end part released at that time.

Anyway, similarly, the GeForce4 family was an entire family of products, though the bastard child was a bit upsetting in its features.

 

[Joe DeFuria]

Anyway, similarly, the GeForce4 family was an entire family of products, though the bastard child was a bit upsetting in its features.

the GeForce4 family may have been an "entire line", but it was essentially almost simple respins of existing products. (And there were only 2 different chips to cover the consumer market...not three like there are today.)

Geforce4 MX = "refreshed GeForce2 MX"
GeForce4 Ti = "refreshed Geforce3"

By this measure, ATI had an "entire line" when they launched the Radeon 8500 and 7x00 series.

8500 = Brand new mid to top of the line
7000 series = "refreshed" older Radeon products.

 

[Dave Baumann]

That wasn't intentional though. The schedule was quite obviously compressed due to 5800's lateness.

NVIDIA's policy has previously (and really was so with the FX series) is to release a high end version on new generation technology and the low end close to or before the high end refresh.

And the 5900 would probably have been the high-end part released at that time.

Anyway, similarly, the GeForce4 family was an entire family of products, though the bastard child was a bit upsetting in its features.

And GeForce 4 is to GeForce 3 what 5900 is to 5800 - same architecture, some silicon differences. As I said, in otherwords, as of yet, they haven't released an entire architectural family at the same time as the first high end part of that architecture, with the exception of NV3x, which wasn't the plan.

 

[Dave Baumann]

L'Inq[/url]]NEW GRAPHICS CARDS for the mainstream market are beginning to tip up in a variety of flavours, all aimed around the crucial 200 Euro mark.

We just learned that Club 3D will ship a 256-bit DDR Radeon 9800SE with just four pipelines, as opposed to the normal eight. This card will be identical to the Radeon 9800PRO, also clocked at 380MHz core and 340MHz memory, save for its crippled pipelines.

As for the Geforce FX 5900XT; this card will be clocked at 390MHz - just 10MHz less than the FX 5900 non-Ultra, with memory working at 350 times two. The regular version works at 800MHz.

Interesting - 256-bit bus cards are entering the mainstream price range. I'd say this is a good bet for 8-pipe cards on 130nm with 256-bit bus coming in at mainstream prices next year...

 

[Joe DeFuria]

Interesting - 256-bit bus cards are entering the mainstream price range. I'd say this is a good bet for 8-pipe cards on 130nm with 256-bit bus coming in at mainstream prices next year...

That is the logical next step.

I see no reason why ATI can't take something like the R350, port it to 0.13u and use a 256 bit wide bes to "relatively slow" memory, and still have decent margins. I would think that would be cheaper (and probably less power consuming?) than significantly raising the clock speed of the memory on a 128 bit bus. I have to beleive that cost and/or power is the reason why ATI did not ship the 9600 XT with faster RAM.

 

[KimB]

And GeForce 4 is to GeForce 3 what 5900 is to 5800 - same architecture, some silicon differences. As I said, in otherwords, as of yet, they haven't released an entire architectural family at the same time as the first high end part of that architecture, with the exception of NV3x, which wasn't the plan.

And this also isn't what ATI claims to want to do. They claim to release an entire line in the first half. That's six months of leeway.