Nvidia works on second generation Fermi

[happy medium]

"Since Fermi is Nvidia’s big transition that got delayed, it looks like Nvidia plans to have two new major generations in a single year."

http://www.fudzilla.com/content/view/17387/1/

 

[GodisanAtheist]

Wait... so now NVIDIA is telling us to pass on Fermi too?

This is certainly an interesting strategy.

 

[gorobei]

probably more like they planned for fermi1 to be out Q4'09 and fermi2 to be Q4'10. i.e. one year between gens. since fermi1 got pushed back, both gen end up being released in '10.

 

[tviceman]

Wait... so now NVIDIA is telling us to pass on Fermi too?

This is certainly an interesting strategy.

No, fud isn't exactly reporting it right. They're just saying that Fermi is going to be getting tweaked/refreshed. It will be comparable to when the gtx200 series got the gtx275 and gtx285 and when then 8800 series got the 8800gt and 8800gts 512.

 

[Qbah]

No, fud isn't exactly reporting it right. They're just saying that Fermi is going to be getting tweaked/refreshed. It will be comparable to when the gtx200 series got the gtx275 and gtx285 and when then 8800 series got the 8800gt and 8800gts 512.

That's what I understood too. Not for people if they already have a Fermi card. A good new buy though (when you're on older generation or getting a new PC in general).

 

[Lonyo]

No, fud isn't exactly reporting it right. They're just saying that Fermi is going to be getting tweaked/refreshed. It will be comparable to when the gtx200 series got the gtx275 and gtx285 and when then 8800 series got the 8800gt and 8800gts 512.

The overall situation is like the X1800 to X1900 ATI series.

 

[WelshBloke]

If all goes well, the second generation Fermi architecture should be coming in late 2010

:\

Deja vu from last year

 

[taltamir]

i don't believe fudzilla on anything

 

[Genx87]

In other news the Sun rose in the East and will set in the West today. Of course Nvidia is working on a refresh of Fermi. And it may even be out on its original timetable which makes it closer than normal to the original launch.

 

[lopri]

The overall situation is like the X1800 to X1900 ATI series.

It's probably more akin to 2900 XT to HD 3870.

 

[bryanW1995]

or maybe 8800 gts 512 to 9800 gtx to 9800 gtx + to 9800gtxwtf +++

 

[ronnn]

Yet another reason not to buy until November!

 

[PingviN]

So... Nvidia wants to sell the Fermis coming up in a couple of months, and then, 6 months later, they plan to release the next generations Fermis? Well, that's got to put a dent in the GF100 release.

 

[lopri]

or maybe 8800 gts 512 to 9800 gtx to 9800 gtx + to GTS 250

Fixed for a factual accuracy. ^_^

 

[waffleironhead]

I'll wait till reviewers have Fermi in hand and performance reviews up before contemplating the refresh.

 

[TheRickRoller]

Fellas, its time to stop working and actually ship something.

:hmm:

And No.

Something new, not another G92 rebadge.

:sneaky:

 

[Lonyo]

or maybe 8800 gts 512 to 9800 gtx to 9800 gtx + to 9800gtxwtf +++

It's nothing like those. They were scheduled refreshes/renames which went along fine and were more to make up for a complete lack of a new mid range product than due to any delays.
Things like X1800 -> X1900 were product refreshes which came quickly (4 months) due to a delay in the initial product, and no delay in the refresh.
Refreshing a product as normal isn't the same.
The 5800 -> 5900 would be another comparable situation.

The main difference is that (we assume) Fermi will perform well, whereas traditionally delayed products that get quickly refreshed don't perform up to snuff.

Nothing particularly unusual about seeing a delayed product get replaced fairly swiftly, both companies have done it before.

 

[Illyan]

Will it be as big a leap as x1800 > x1900 though? Or more 7800 > 7900? I'd assume any Fermi refresh would just be clock increases.

 

[Lonyo]

Will it be as big a leap as x1800 > x1900 though? Or more 7800 > 7900? I'd assume any Fermi refresh would just be clock increases.

Who knows. The first part isn't even out yet, so speculation on how much of an improvement a refresh would give is just crazy.

 

[nitromullet]

Will it be as big a leap as x1800 > x1900 though? Or more 7800 > 7900? I'd assume any Fermi refresh would just be clock increases.

Don't you think 6-9 months sounds like a pretty long time to wait for just a clock increase?

 

[Illyan]

Don't you think 6-9 months sounds like a pretty long time to wait for just a clock increase?

Fermi's die is pretty huge already with 512 cores, what more can Nvidia do than increase clocks? Unless they manage to either greatly increase the performance of each core, or pack more cores into the same space. Anything's possible though at this point I guess.

 

[SlowSpyder]

In other news the Sun rose in the East and will set in the West today. Of course Nvidia is working on a refresh of Fermi. And it may even be out on its original timetable which makes it closer than normal to the original launch.

Yea, kind of a pointless article. Of course Nvidia is working on their next part, it's not like they just work on one of these things at a time.

 

[MarcVenice]

Maybe Nvidia is betting on 32 or 28nm? I'm hearing various people claim different things though. Some say no bulk 32nm, so they move straight to 28nm? But others claim GF will do 32nm, although that's irrelevant to Nvidia.

 

[happy medium]

Maybe Nvidia is betting on 32 or 28nm? I'm hearing various people claim different things though. Some say no bulk 32nm, so they move straight to 28nm? But others claim GF will do 32nm, although that's irrelevant to Nvidia.

Die shrink = clock increases,less heat, and less power consumption= refresh?

All the things that the first Fermi will fail on possibly.

 

[Lonyo]

Yea, kind of a pointless article. Of course Nvidia is working on their next part, it's not like they just work on one of these things at a time.

I think it's more about the refresh being on time rather than there being a refresh.
It's reasonable to assume given the nature of the problems that the refresh wouldn't be delayed, but this is just a confirmation that the negative experiences of Fermi won't impact future products (in much the same way as mainstream launches should be happening shortly after Fermi with no delay due to its delay).

It's almost non-news, but not quite. It's just confirming what would be mostly expected (that there will be a refresh and it's not been affected by Fermi delays).