The official GT300/Fermi Thread

[***Deimos***]

someone wants to make a picture story of the fermi is comming in december??

why december?

Wait a couple weeks and make one for "09" with 3 wise men and Fermi in the middle.

Any news/rumours/roadmap of HD57xx equivalent from nVidia?

[LordEC911]

Well... it looks like nVidia and ATi are both having the same problem with TSMC. It seems ALL of their 40nm high performance products are getting a crappy 50% yield.

http://www.fudzilla.com/content/view/16421/1/

Nope... AMD/ATi isn't having any problems with yields on their current chips, Cypress yields are slightly lower than expected but still fine, Juniper's are better than expected.

A yield of 50%, esp when on a radically new chip architecture with a considerable die shrinkage, is actually pretty good. It's when yields on first batch runs dip below 20% that you start getting serious concerns about profit margins. Some have rumored Fermis yields to be sub 2%, so 50% would be a huge success story by comparison.

The problem is too much talk is being generated from mere rumors. ATI may be increasing their yield, but fact remains no one has 5870s in stock, and if their yields initially were so much better than Nvidia's as implied, that would not be the case. Some seem quick to defend ATI's ability to increase yield, but not Nvidia. Product development/improvement works both ways gentlemen.

Nvidia is NOT getting 50% yields...
I don't think that you guys are understanding what is going on, yields |= capacity. Just because yields are improving doesn't mean they can get more wafers from TSMC. The reason for the shortage is the fact that TSMC was suppose to increase AMD/ATi's capacity from thousands of units per week to 10k units per week(units |= wafers). This was suppose to happen before Q4 and was delayed and delayed for some unknown reason and then the reason that TSMC gave didn't add up.
http://www.driverheaven.net/news.php?newsid=322

ummm.. no.

Yield is chips without defects that make unusable. Different circuits have different defect susceptibility. Bigger chips are more likely to have defects.

If your quote was true, and AMD is getting 50% yield on HD58xx, and nVidia is getting 50% on their tiny GT220 DX10.1 40nm chips, then nVidia might as well give up. A 10x bigger complex chip would have dismal pathetic yields like under 1%.

Delay wise, AMD had A2 taped out like June, 4 months before launch. nVidia just taped out Fermi A2, and rumours A3 needed to fix bugs. So we're hoping for March.

AMD had A2 silicon? Where?

why december?

Wait a couple weeks and make one for "09" with 3 wise men and Fermi in the middle.

Any news/rumours/roadmap of HD57xx equivalent from nVidia?

Nothing until late Q2/Q3 '10.
No G300 derivatives have been tapedout yet.

[Alien Grey]

someone wants to make a picture story of the fermi is comming in december??

Done.

[justin.kerr]

just as well change it to Jan, Feb... just to have it ready.

[LedHed]

find it hilarious how many ATI fan boys are trolling this thread, there won't be much to troll when the GT300 does release and sets the new bar of GPU performance.

also it's pretty sad it took ATI over 13 months and even 2 generations to beat out the old top dog (295), even with GDDR5/DX11 they couldn't beat it until the 5970 which is in very slim quantities.

[RealTelstar]

So the new ETA is jan/feb '10?

[LedHed]

So the new ETA is jan/feb '10?

at best, probably later

[justin.kerr]

that GTX 295 you have been around for 13 months? lol
since the gtx 295 came out in JAN 09 and the 4870x2 was AUG of 08, then two generations after that would be the 6XXX series... so WTF? lol
I run mostly Nvidia, and am eagerly waiting for Fermi's release, but cut the crap. lol

[LordEC911]

New hints are dropping of Q2... which seems to imply A3 isn't going to be final silicon. I have no idea if this is true or not.

that GTX 295 you have been around for 13 months? lol
since the gtx 295 came out in JAN 09 and the 4870x2 was AUG of 08, then two generations after that would be the 6XXX series... so WTF? lol
I run mostly Nvidia, and am eagerly waiting for Fermi's release, but cut the crap. lol

Psssttt... don't pester the troll. He enjoys trolling Nvidia threads.

[LedHed]

that GTX 295 you have been around for 13 months? lol
since the gtx 295 came out in JAN 09 and the 4870x2 was AUG of 08, then two generations after that would be the 6XXX series... so WTF? lol
I run mostly Nvidia, and am eagerly waiting for Fermi's release, but cut the crap. lol

2 Generations = 4800 and 5800, pretty obvious.

Even with the 5800 series they still couldn't surpass the 295. I'm sorry I said 13 months instead of 12 months, big deal...

and yes I'm the troll, not the people posting this:



give me a break

[justin.kerr]

so the 295 came out after the 4800... so add 2 generations, that would be 6XXX.. Ok done feeding.
Focker out.

[LedHed]

so the 295 came out after the 4800... so add 2 generations, that would be 6XXX.. Ok done feeding.
Focker out.

4890 came out after the 295, April 2 09, 4 months after the 295 launch.

[InCredible]

295 was fastest for 10 months..before that ati was fastest for 5 months..and going to be fastest for atleast 5 months after..and the 4890 isnt a generation..its a refresh....3 series is a generation..4 series..then 5 series..so it took ONE generation to beat the 295..because the 4 series isnt getting beat by ANYTHING other then the 295...however ati had the better price/performance by far not that it matters..

[LedHed]

295 was fastest for 10 months..before that ati was fastest for 5 months..and going to be fastest for atleast 5 months after..and the 4890 isnt a generation..its a refresh....3 series is a generation..4 series..then 5 series..so it took ONE generation to beat the 295..because the 4 series isnt getting beat by ANYTHING other then the 295...however ati had the better price/performance by far not that it matters..

I know the 4890 is not a generation...I was pointing out that the 4 Series was still being released after the 295 which is opposite of what he was saying. 2 Generations would be the one going on during the 295 launch (4800 series) and the following 5800 series. That is unless you guys don't consider them to be 2 generations. They did just double all the performance numbers (ROPs/TU/Shaders/etc) and stuck to the same architecture, which is why they came out so much earlier than a new design like the GT300.

[InCredible]

^ agreed the only reason i dont say 2 generations is because the 4 series and gt200 series competed together...gt200 won overall performance..1 generation later(5 series) ati took it back..and ur probably right on the reason they came out much faster..the only thing im not happy with nvidia about is 3 yrs of the same arch and late to the party with fermi..hopefully it is a strong competitor..ati has use the same arch however are saying something new will come in the end of 2010...

now we can expect the launch dates to be delayed..its happend more then once..ati was late with r600 ..nvidia late with..whatever u want to call it..fermi.gt300/gt100 idk....now we'll see if ati's new arch(6 series) will be on time or not..


i just wish something new would show up with proof for fermi..im not impressed as of yet since we have no actual graphics situations to go by...gpgpu doesnt do it for me

[Alien Grey]

2 Generations = 4800 and 5800, pretty obvious.

Even with the 5800 series they still couldn't surpass the 295. I'm sorry I said 13 months instead of 12 months, big deal...

and yes I'm the troll, not the people posting this:



give me a break

Someone asked for a picture story of Fermi coming in December and that's exactly what I gave him.

Don't take it to serious I'm not an ATI fan boy and also not a Nvidia fan boy. I buy whatever I think is the right choice at the moment. My first choice is a Nvidia graphics card because it has a little better support in Linux but that could change once Fermi is released and the price, performance and power consumption isn't as good as the ATI graphics cards.

So I'm going to wait and see what Nvidia can give me and than I decide what I'm going to buy.

[kiwik]

lol @ the rock pics

[justin.kerr]

I know the 4890 is not a generation...I was pointing out that the 4 Series was still being released after the 295 which is opposite of what he was saying. 2 Generations would be the one going on during the 295 launch (4800 series) and the following 5800 series. That is unless you guys don't consider them to be 2 generations. They did just double all the performance numbers (ROPs/TU/Shaders/etc) and stuck to the same architecture, which is why they came out so much earlier than a new design like the GT300.

You funny in the head boy?.. Jan to April is 3 months not 4..you tell me that the 4890 is a next generation, then say in your next post it is not...your level one noob skills need to go.
Nvidia and ATI will always have new models coming out within a series(generation) Nvidia released a whole crap load of cards after ATI released the 4890, so again you make no sense at all. The 4890 was not designed to compete with the GTX 295, so again you hold no water with anything you say.
You said two generations after the 295 had been released, your words, that would make it ATI's 68XX series, plain and simple.

[LedHed]

You funny in the head boy?.. Jan to April is 3 months not 4..you tell me that the 4890 is a next generation, then say in your next post it is not...your level one noob skills need to go.
Nvidia and ATI will always have new models coming out within a series(generation) Nvidia released a whole crap load of cards after ATI released the 4890, so again you make no sense at all. The 4890 was not designed to compete with the GTX 295, so again you hold no water with anything you say.
You said two generations after the 295 had been released, your words, that would make it ATI's 68XX series, plain and simple.

I guess you consider the other 23 days in January not to exist after the 295 launch on January 8. Would you rather me say 3 months and 23 days? Does that satisfy your retarded needs? I mean who the hell cares about 8 days in a month being rounded?

Also I never said the 4890 was designed to compete with the 295, I simply said it was released after the 295. This would in turn mean the current generation to compete with the 295 at release was the 4800 series (since nothing else was available, unless you know of some secret series). You add two generations from the 295 launch and you get the 4800 series and the 5800 series.

You guys can argue all day what I said doesn't mean that, but that is what it means.

[kiwik]

GT 200 is the same generation as R700. Be it GTX 295 or GTX 280, it's the same chip, it only had a revision with the die shrink (GT 200b), but then so did ATI with the 4890 (RV790).

GT 300 will be the same generation as R800.

Nice try there.

[weston]

just ignore ledhed, he got banned from another forum for this same stuff. He just loves his GTX295 so much...

[phinix_mike]

Guys, wanted to ask you here, cause I forgot where I read about it:
So first will come Fermi single chip card, right? Q1 2010. When was the dual chip card scheduled? Was it 3-4 months after single? I read it somewhere and cannot find it now... Was it said somewhere?

[kiwik]

just ignore ledhed, he got banned from another forum for this same stuff. He just loves his GTX295 so much...

[***Deimos***]

FYI - I dont think anybody is trolling here. And certainly not ATI fanbois here.
I loved the $100 specials on 8800GT, and "supported" nVidia with my wallet over years
But, looking at FACTS, the situation looks grim.

chip date #months_difference
GF1 Oct99
GF2 Apr00 6
GF3 Feb01 10
GF4 Feb02 12
FX Mar03 14
68 Apr04 13
78 Jun05 15
79 Mar06 9
GTX Nov06 8
G92 Nov07 12
280 Jun08 7
Fermi Dec09? - 19months and counting since G200.

No matter how you put it, nVidia has slowed down tremendously. Took 12 months for simple die shrink of 8800GTX!
Its already 19 months since the G200 line and counting, and likely by launch it will be close to 2 years! Past "record" was 6800->7800 15 months. Its like after the summer of 08 everybody at nVidia took a 1 year holiday. They missed and skipped a cycle.

source
http://www.neeyik.info/timeline/

[LordEC911]

FYI - I dont think anybody is trolling here. And certainly not ATI fanbois here.
I loved the $100 specials on 8800GT, and "supported" nVidia with my wallet over years
But, looking at FACTS, the situation looks grim.

chip date #months_difference
GF1 Oct99
GF2 Apr00 6
GF3 Feb01 10
GF4 Feb02 12
FX Mar03 14
68 Apr04 13
78 Jun05 15
79 Mar06 9
GTX Nov06 8
G92 Nov07 12
280 Jun08 7
Fermi Dec09? - 19months and counting since G200.

No matter how you put it, nVidia has slowed down tremendously. Took 12 months for simple die shrink of 8800GTX!
Its already 19 months since the G200 line and counting, and likely by launch it will be close to 2 years! Past "record" was 6800->7800 15 months. Its like after the summer of 08 everybody at nVidia took a 1 year holiday. They missed and skipped a cycle.

source
http://www.neeyik.info/timeline/

Think you need to put G200b in there if you are counting G92 and G71.

Guys, wanted to ask you here, cause I forgot where I read about it:
So first will come Fermi single chip card, right? Q1 2010. When was the dual chip card scheduled? Was it 3-4 months after single? I read it somewhere and cannot find it now... Was it said somewhere?

The original rumor/speculation/PR BS from Fud/Theo was that a dual card was going to be launched at the same time which is very highly unlikely. It is also unlikely for Nvidia to put a full GF100 on the dual card, they will either have to use a cutdown version like in the past or wait for a die shrink or maybe both.