HYPOTHETICAL: R420 two months earlier than nV40?

[digitalwanderer]

My point is that NV40/45 is still in development and can benefit from improvements. R420 is finished and is effectively standing still - it isn't going to get any better sitting around in warehouses waiting for NV40 to launch.

Do you know that the R420 is "finished and effectively standing still" or are you just assuming/speculating with that bit-o-data there? :|

 

[Doomtrooper]

1st to market is crticial for any product...and as usual most titles are CPU limited so a huge difference in frames is not going to differ from ATI and Nvidia. It will again be the feature set that 'should' sell the card, be it better AA, better pixel shader support etc...

 

[Bouncing Zabaglione Bros.]

My point is that NV40/45 is still in development and can benefit from improvements. R420 is finished and is effectively standing still - it isn't going to get any better sitting around in warehouses waiting for NV40 to launch.

Do you know that the R420 is "finished and effectively standing still" or are you just assuming/speculating with that bit-o-data there? :|

I'm speculating based on the supposed late April/early May ship date. As JR says, you can't just chop and change designs at the last minute.

Even if R420 is finished in one month and NV45 is one or two quarters later, the same principle applies. You can't leave the finished R420 sitting around waiting for it to be superceded by NV45 six months later with an additional half year of development time under it's belt in order to try to beat R420.

I don't underestimate Nvidia enough to think that they wouldn't be able to do anything useful with an extra 3-6 months development time.

 

[digitalwanderer]

Thanks for the clarification, it's appreciated.

I don't underestimate Nvidia enough to think that they wouldn't be able to do anything useful with an extra 3-6 months development time.

I think one of the major reasons/arguments to push back the R420 release would be to allow 'em to get top-dollar for their R3xx products for the max time and clear out the stock on them, but I agree more with you to release early.

 

[martrox]

I think one of the major reasons/arguments to push back the R420 release would be to allow 'em to get top-dollar for their R3xx products for the max time and clear out the stock on them, but I agree more with you to release early.

Except the 9800Pros are down to $250 everywhere and the AIW 9800 is $300..... RETAIL! at CompUSA!

 

[PurplePigeon]

First to market (by a non-trivial amount of time) would also make sense if ATI introduces new product in all price ranges. Enthusiasts could wait until both nV and ATI has their high-end parts out, but otherwise the world keeps on turning. Other individuals may not be willing or able to wait. OEMs still need to refresh their product line.

 

[Bouncing Zabaglione Bros.]

I think one of the major reasons/arguments to push back the R420 release would be to allow 'em to get top-dollar for their R3xx products for the max time and clear out the stock on them, but I agree more with you to release early.

If you look at the recent price drops on the 9800, I think that people are already clearing stock out of the distribution channel in preparation of the R420 launch. If they don't ship R420, ATI will have to find fab space (and scheduling) to make more R360 chips (as will their card maker OEMs) or else be left with no product to sell.

I think that's unlikely to happen, especially given that you have to run your own plans centered around what you need to do, not what you think your competitors might be doing.

 

[StealthHawk]

1st to market is crticial for any product...and as usual most titles are CPU limited so a huge difference in frames is not going to differ from ATI and Nvidia. It will again be the feature set that 'should' sell the card, be it better AA, better pixel shader support etc...

Then ATI's in trouble, aren't they?

 

[Daliden]

Mature drivers on launch wouldn't go amiss either.

Aaaanyway . . . I would think that at least one of the IHVs concentrated on getting additional IQ for "free", i.e. aniso, trilinear, AA, whatever, without any drop in framerate. Wouldn't that be a good place to search for additional FPS? But is that kind of hardware even theoretically possible?

 

[Natoma]

I agree. Personally I'm beginning to think the R420 and NV40 won't be released until Half Life 2 and Doom 3 are literally a month or so away. There's certainly no compelling software reason today to release these cards when you've got super powered 9800XTs and 5950 Ultras on sale that can run games at high res, AA/AF and other eye candy. Hell, the main reason I upgraded my entire system last June was because I thought Half Life 2 was coming in September. Little did I know.

With the system I've got today, there's no way I'm going to upgrade again anytime soon. But if R420 is indeed the same jump over the 9700/9800 as the 9700 was over the 8500 as that ATI rep said a few weeks ago, then I'll get it, but only because I want to play Half Life 2 with all the eye candy. Software pushing hardware and all that.

 

[AlphaWolf]

My point is that NV40/45 is still in development and can benefit from improvements. R420 is finished and is effectively standing still - it isn't going to get any better sitting around in warehouses waiting for NV40 to launch.

Do you know that the R420 is "finished and effectively standing still" or are you just assuming/speculating with that bit-o-data there? :|

I'm speculating based on the supposed late April/early May ship date. As JR says, you can't just chop and change designs at the last minute.

Even if R420 is finished in one month and NV45 is one or two quarters later, the same principle applies. You can't leave the finished R420 sitting around waiting for it to be superceded by NV45 six months later with an additional half year of development time under it's belt in order to try to beat R420.

I don't underestimate Nvidia enough to think that they wouldn't be able to do anything useful with an extra 3-6 months development time.

Well you don't necessarily drop and change the design, but usually every respin improves yields and performance. While I am sure that maintaining the performance crown is one goal of ATI, another goal is making their oems happy. Right now they can do nothing and accomplish both. I am sure there are dates and times where it will make sense for them to release a new product regardless of what nvidia does, but I don't doubt they have a couple months of flexability

This may upset a few enthusiasts, but it's not like they can run out and buy a faster card from the competition.

What they could do if they want to improve on their lead is release a Radeon(2) now as the new performance leader and follow it up with a pro/XT version to best (compete with) nv40/45. This would probably require that they turf the 9800pro or 9800xt as that segment would shortly be filled by Radeon(2).

I dont really know what their limit on lowering the price of these products are, can they sell 9800pro's for $150 to move them into mainstream market and then sell 9600xts as a $100 card?

 

[Hyp-X]

When ATi released the original Radeon they did and SDR version and the VE.

Together with R8500 they released the R7500

Together with R9700 they released the R9000

So I'd be surprised (and disappointed) if R420 arrived alone.

 

[Tim Murray]

Yeah, see. There's working 16-pipe NV40 silicon. Devrel has apparently been sending out benchmarks of the new chip, too.

 

[Dave Baumann]

ATI are gearing up for PCI-Express, hence I would be surprised if there is anything else, chipwise, that will be announced / released until the big push for PCI-Express.

 

[Fodder]

So I'd be surprised (and disappointed) if R420 arrived alone.

There's RV380/M24, and later on RV410. PCIE RV380 will be one of the main public drawcards for partners like Sapphire at CeBit, or so it appears.

 

[digitalwanderer]

Yeah, see. There's working 16-pipe NV40 silicon. Devrel has apparently been sending out benchmarks of the new chip, too.

Link? 8)

 

[Tim Murray]

Yeah, see. There's working 16-pipe NV40 silicon. Devrel has apparently been sending out benchmarks of the new chip, too.

Link? 8)

You'll see it on L'Inq in two weeks, probably.

 

[digitalwanderer]

Yeah, see. There's working 16-pipe NV40 silicon. Devrel has apparently been sending out benchmarks of the new chip, too.

Link? 8)

You'll see it on L'Inq in two weeks, probably.

Oh, it'll just be FUaD then...

 

[Joe DeFuria]

Yeah, see. There's working 16-pipe NV40 silicon. Devrel has apparently been sending out benchmarks of the new chip, too.

I don't consider nVidia provided benchmarks any indication at all of actual silicon existing. This doesn't mean it doesn't exist, of course, but given nVidia's history, it certainly doesn't convince me.

 

[Tim Murray]

Yeah, see. There's working 16-pipe NV40 silicon. Devrel has apparently been sending out benchmarks of the new chip, too.

I don't consider nVidia provided benchmarks any indication at all of actual silicon existing. This doesn't mean it doesn't exist, of course, but given nVidia's history, it certainly doesn't convince me.

No, I'm not using the existence of PR benchmark scores as the proof of the actual sillicon's existence.