Faster ATI 9700 Coming?

[sbp]

http://www.nvmax.com/#1032537674

Word came in from an anon emailer that has relations with some ATI folk that ATI will introduce a faster version of its 9700 chipset sometime around January.

Because TMSC has switched wafers to .13 micron ATI will be able to ramp up the clocks and use faster memory to cut short NVIDIA's NV30 right at launch. The emailer said ATI folk sounded 'optimistic' about this new version being as fast as NV30.

ATI already have the R300 core rolling so will be little design overhead where-as NVIDIA are busy trying to be in time for the Christmas rush.

[thcdru2k]

this is bad news for nvidia.

[Lezmaka]

The guys at ATI must have changed their minds because I remember reading that the CEO said they wouldn't do a .13 micron refresh of the R300. Unless the chip they are talking about is the R350

[LORD-eX-Bu]

wonder how long they have been working on a .13micron process. I mean, Nvidia has been doing it for a while and have gotten all the problems out of the design and it is in production now, but I wonder how far ATI is in doing the same.:eek:

[Uttar]

Originally posted by sbp
http://www.nvmax.com/#1032537674

Word came in from an anon emailer that has relations with some ATI folk that ATI will introduce a faster version of its 9700 chipset sometime around January.

Because TMSC has switched wafers to .13 micron ATI will be able to ramp up the clocks and use faster memory to cut short NVIDIA's NV30 right at launch. The emailer said ATI folk sounded 'optimistic' about this new version being as fast as NV30.

ATI already have the R300 core rolling so will be little design overhead where-as NVIDIA are busy trying to be in time for the Christmas rush.

Actually, i think this is VERY unlikely.

But there may be some trust here - a launch in January.

I'd guess it's the R350 paper launch. But if that's true, woah, ATI is getting real fast those days! But then again, we might have to wait until march/april for the real cards since they mostly want to spoil nVidia fun.

Oh, and did you note the fact that they are optimistic it'll be faster than the NV30? Great, now we actually got from some inside-ATI source that the R300 is slower than the NV30. LOL!

Uttar

[LuckyNV]

ATI stepping up Hardware launches?

I mean.... look at the difference between R8500 and R9700, that was like a very very long time.

Would be odd for ATI to launch a new card any time soon.

[thcdru2k]

looks like ati doesn't want to be screwed everytime they release a radeon :)

[Fotis]

I think Ati should first correct their faulty agp8x radeon 9700's and then go for 0.13.

[budd_wm]

The AGP issue was not the fault of ATI's, though they are working with mobo manuf's to get it solved. The fault is actually on the mobo end because the manuf's did not specifically follow the AGP 8x spec. There have already been some BIOS revisions released on some mobo's that fix the issue.

[Philibob]

Originally posted by Uttar
Oh, and did you note the fact that they are optimistic it'll be faster than the NV30? Great, now we actually got from some inside-ATI source that the R300 is slower than the NV30. LOL!
They didn't use the word faster they used as fast as :eek:

[Bigus Dickus]

Originally posted by LuckyNV
ATI stepping up Hardware launches?

I mean.... look at the difference between R8500 and R9700, that was like a very very long time.

Would be odd for ATI to launch a new card any time soon.

8500 to 9700 launches was less than one year. NVIDIA takes about that long between their new cores as well.

[Bigus Dickus]

Originally posted by Uttar
Actually, i think this is VERY unlikely.So you doubt the validity of the "inside source" then?

Great, now we actually got from some inside-ATI source that the R300 is slower than the NV30. LOL!So you don't doubt the validity of the "inside source" then?

Erm... :rolleyes:

[Stavros]

I think its a bit odd

You have AMD, nvidia etc all having problems with the move to the .13 micron process, then ATI come along and say "we can't meet demand for the R300, lets just swap over to 12" waffers and .13 micron process". Now I'm no hardware expert but from what I've been told through reading these forums that moving from .15 to .13 is not just a case of 'zooming in a bit'.
If this is true then ATI must have had the .13 micron part taped out about the same time, if not before the NV30.

Its either total crap, or ATI have just pulled off the biggest upset of the graphic card industry this century

PS. I do think it strange that an ATI insider would say that it would bring the performance up to the level of the NV30, which would mean that currently, the NV30 IS faster than the R300, and I don't think that ATI would want to say that their refresh part will only bring it up to the performance of the NV30 and not surpase it.

[Philibob]

How would ATI know that the NV30 is faster than the R300? :confused:
Do they have one?

[StealthHawk]

Originally posted by Philibob
How would ATI know that the NV30 is faster than the R300? :confused:
Do they have one?

when even ATI expects/believes that the competition will be superior to their own product, i think we've got a few problems ;)

[budd_wm]

Don't be naive. ATI probably knows a good deal about the NV30. It's their job to know. And I'm sure nVidia is aware of ATI's future plans. Business is all about getting into other people's business.

[SnakeEyes]

From Anandtech - Inside ATI & NVIDIA: How they make frames fly
Once a manufacturing process is decided upon, it is extremely difficult to, months later, go back and attempt to revise the design for a different process.

I'm sorry, but I have a really hard time believing that the 9700 will be out on a smaller process. The R350, I believe, but not the 9700. Especially not as early as January - March of next year. (Though March is getting close to the earliest I'd expect they actually could get a .13 revision of the R300 off the drawing boards and onto the market).

Just my $.02 US

[Megatron]

Originally posted by StealthHawk


when even ATI expects/believes that the competition will be superior to their own product, i think we've got a few problems ;)

I dont see any problem at all. I think if any competitor (Nvidia/ATI) has upwards of 5 months or more to pick over the oppositions latest offering...it is a no brainer that they will do all in their power to make sure their part comes out faster.

If this rumor is true , ATI is simply giving credit to the other major player in this industry to be able to take the extra 1/3 of the year and put it to good use in their next product.

Now if Nv30 by any means does not deliver on this anticipated chance...then I think we`ve got a few problems.

Lets face it Nvidia is late here, they missed the boat, not ATI...their part HAS to improve on what the R300 is already delivering. If it does..oh well..It SHOULD...if it doesnt...then its a waste of the extra time.
Its a win/win for ATI at any rate.

[Bigus Dickus]

I agree completely.

[StealthHawk]

that's assuming nvidia refined and improved NV30 over what it originally was going to be. i can suppose that if anything, the memory used would clock higher. but we don't know what, if anything, that they did.

suppose again that NV30 really does use a 128bit bus. i find it hard to believe it would be faster than R9700 in that case. but yes, regardless of what happens, consumers win :D

[kingsna]

just think about this for a second,

would ATI order 12" wafers to make new .13u r300 parts from WITHOUT actually having the designs ready to be sent over to the fab for tapeout?

i don't think so.

ATI had atleast 3-4 months after original r300 tapeout to be able to shrink the die from .15u to .13u so it's reasonable to assume that they're ready to go smaller for more chips per wafer since they don't need more speed for r300, the yields should be good when using .13u and thus more profit for ATI!

in the meantime they can keep better performing chips out of those made using .13u to use for faster 9700 products in the future!

good strategy!

[Crisler]

The rumors I have been hearing are that a version of the R300 has been worked on at .13 for a while now, evidently as long if not longer than nVidia. The reason the larger die was used in the release was that it was ready to go and would not delay the production schedule. The .13 version there by could very well be close to complete and ready to go. By going to the .13 they could ramp up the speed still futher. Add to this the fact that the R300 supports the newer systems for DDR this means even HIGHER memory bandwidth.

As for the speed question. Any assumption of speed is based on the paper specs for the NV30. So what they would be saying is the new version of the R300 would be as fast as the NV30 claims to be on paper.

The real rub here for nVidia is they lost the leadership role. Even as ATI prepares at .13 R300 they are already working on their next generation chip. Based on the average production cycles, just as the NV30 begins to get solid the new Radeon next gen chip will be coming out.

nVidia is in a serious crunch right now. They stand on the brink of keeping the race tight or dropping out, depending on how they do with this release. However delays much longer than what has already occured could make the NV30 a still born child.

Edward

[StealthHawk]

even if ATI could bring out an improved R300, why would they do so within 3 months of the launch of R300? no, that is crazy. and it is crystal clear from the naming scheme of R9700Pro that the current card will be the top of the line from ATI for quite some time. we'll be seeing slower, cheaper chips from ATI, not more expensive ones.

[jbirney]

Why would it be crazy? Its all about staying on the top. Thats what is important to them and good for business. There would be nothing wrong with ATI relasing a 400 Mhz clocked part using DDR2. It would be at the same price point at the NV30 and its perfromance would be at or above the NV30. Why is that crazy?

Would I be pissed off? No. I know that every few months someting faster is out. I paid 338 for mine and its working great. For almost a month now I have enjoyed high res, max aniso and AF so yea I am getting my moneys worth. Besides how often do CPUs increase in speed? Same thing with them.......

[kingsna]

let me explain what this die shrink means to ATI.

as you know manufacturing these 110m transistor chips costs ATI more than making the same chips using .13u process.

now that they've made the chips using .15u and have been able to get 325Mhz from it reliably, they can target the same speed with .13u process BUT get more chips per wafer and less cost of manufaturing per chip since they don't need to improve performance for 9700 pro.

now TSMC has many formulas for making the same chips:

one for maximizing performance but with less yields.

another one for maximum yields but with less performance of what can be achieved with using other methods.

and some other options.

so as you can see ATI can order the chips to be made using the Max yield formula instead of Max performance and save a bundle on overall costs of manufacturing,packaging, power, etc across the board and that's really smart if you ask me and if ATI can keep on doing things like this, nvidia can kiss they're crown forever since they're already behind in performance and everything they make will still be surpassed in 3-4 months of release.

so what i think nvidia should do is to quit fooling around with other projects such as XBOX and start focusing on PC's if they really want to be the king.

peace!