NVIDIA GT300 Yields Are Under 2%?

These companies all have their ups and down. Back when Nvidia was pushing out the 7000's and then came out with the 8000's ATI has somewhat similar issues getting out a new GPU and then when it finally arrived it it wasn't very spectacular compared to Nvidia's offering. Eventually the 3000's came out and ATI/AMD decided to do the right thing and sell them at a great price and then kicked out the 4000 line while.

Similar thing happened with CPUs in 2003/2004 when AMD launched their athlon64 chips and made intel look like fools. Intel wasn't able to beat AMD completely till the C2D.

I take this type announcement like a grain of salt. More important type of news that came out a while back when Nvidia's mobile chipsets were flawed and failing left in consumer's laptops.
 
quimby999 said:
Charlie is not the most reliable person to listen to on things Nvidia he has a strong anti Nvidia bias and is not above making up facts about Nvidia.So take thibgs he says with a grain of salt and I'm not sure why any website would link to an article by him when hes been proved wrong so many times.
Click to expand...

This has been my point in other sites like Tech-Report, that post links to this idiot's site on the front news page...

Obvious bias and constant bad information about a certain company, makes "him" worthy of front news page ? What the hell is wrong here ?

Also, the amount of AMD banners on "Charlie's" site is quite revealing :)

I've often theorized that he was being payed by AMD, to talk trash about NVIDIA and to promote AMD. Those banners kind of prove that :)
 
Big D. said:
Anyone have the numbers for AMD chips? I haven't heard any terrible news about them.
Click to expand...

That's because no one is a raving AMD fanboy like Charlie (though there are followers...), that ONLY writes bad things about NVIDIA.

There's no NVIDIA fanboy that makes up things about AMD left and right and posts it on their website...If there was, you would've known by now.
 
panhead said:
We will never know what the actual yeilds are for Nvidia's, AMD's or any other manufacture. What we do know is that Nvidia has just received engineering samples and AMD is launching their DX11 part next week.
Click to expand...

We do ? According to several "other" sources, GT300 has taped out months ago...
 
ZerazaX said:
Unless you can read Japanese, the translation is going to differ across different translators. Seriously, plugged into 10 different translators, 10 results come out...

Yeah, it's likely he just translated it wrong, but here's some other post:

http://forum.beyond3d.com/showpost.php?p=1334603&postcount=2123

He's been spot on this entire generation (called RV870 specs long before anyone else did) and he even wrote the 1.67 a day before Charlie did, though he actually meant that sarcastically and he's saying it was close to 10%. Charlie probably copied his number actually...

I'm more inclined to believe that they are indeed at 10%, especially if his narrative of what happened to the GT21x series is indeed true:

http://forum.beyond3d.com/showpost.php?p=1334958&postcount=2151
Click to expand...

LOL, you are hilarious.

Anyone that reads beyond3d knows that neliz is as blatant of an AMD fanboy as Charlie is...Actually, he posts quite a lot of what Charlie writes about too, which isn't surprising.
 
jaydeejohn said:
Itll be interesting to see which way this turns out
Click to expand...

No matter how it turns out I'll wait and see how Nvidia's new card compares to AMD's new card and I'll decide what new videocard to get from that info. If Nvidia however lags behind and has no DX11 hardware in their new GT300 well then, I'll look at both numbers and take that into consideration as well.

I honestly don't know why HardOCP posts anything Charlie writes about as it drives most Nvdia users mad as we know who Charlie backs.
 
They drive anyone with GTX295 280SLI 260 SLI, anything higher than GTX260 mad.
Ati users are just... ugh...
If they got a nice videocard that is.
 
AtkinS said:
They drive anyone with GTX295 280SLI 260 SLI, anything higher than GTX260 mad.
Ati users are just... ugh...
If they got a nice videocard that is.
Click to expand...

I meant about.
ati's performance, that they got their videocard out in less than 10 days, and if you try to say physx isnt a big deal at all, they get all mad and say i havnt tried it, and then i say i have a 285 and then they complain about the next thing.

I just complain about how much my system have decreased in value over the past 12 months, probaly more than my car.
Which sucks, but i never learn :) new videocard soon :D
 
The anti-charlie bias is nearly as bad as charlies anti-nvidia bias :p

I find it hard to believe, not because of charlies alleged track-record of always being wrong (pretty sure he isn't btw), but because that type of yield is unthinkably bad, act-of-god bad.

I don't understand how it could be that bad unless all the engineers working at the fab were on crack, or someone drove a forklift into the etching machine thing half way through, or they accidentally used a pizza base instead of a silicon wafer or something.
 
MartinX, it is not that hard. GT300 is much larger chip, which makes errors more common. Another problem is that ATI have much bigger knowledge of what traps TMSC 40nm process have (they learned it with HD4770), so they were able to avoid them at designing.
Simply, if you have a chip with half of size of the other guy, then with same damaged area of waffer you still got double of their chips :p.
 
faugusztin said:
MartinX, it is not that hard. GT300 is much larger chip, which makes errors more common. Another problem is that ATI have much bigger knowledge of what traps TMSC 40nm process have (they learned it with HD4770), so they were able to avoid them at designing.
Simply, if you have a chip with half of size of the other guy, then with same damaged area of waffer you still got double of their chips :p.
Click to expand...

Ah, so it's half the size now ? All rumors concerning GT300 say it's a <=500 mm2 chip. So unless 330-340mm2 is half of 500 mm2, which math says it isn't, you are grasping at straws...As is the case with Charlie btw.

As for the HD 4770 bit, NVIDIA has been working on 40nm for as long (if not longer) than ATI, so they have pretty much the same knowledge. ATI just thought that it made sense to release a card with poor yields. NVIDIA didn't.
 
MartinX said:
The anti-charlie bias is nearly as bad as charlies anti-nvidia bias :p
Click to expand...

Wasn't the anti-charlie bias created BECAUSE of Charlie's anti-NVIDIA bias ?

So it's obvious that one bias will grow as the other one does too.
 
MartinX said:
I find it hard to believe, not because of charlies alleged track-record of always being wrong (pretty sure he isn't btw), but because that type of yield is unthinkably bad, act-of-god bad.

I don't understand how it could be that bad unless all the engineers working at the fab were on crack, or someone drove a forklift into the etching machine thing half way through, or they accidentally used a pizza base instead of a silicon wafer or something.
Click to expand...
I think this is the best point of the thread.
 
As for the HD 4770 bit, NVIDIA has been working on 40nm for as long (if not longer) than ATI, so they have pretty much the same knowledge. ATI just thought that it made sense to release a card with poor yields. NVIDIA didn't.

I wonder why nvidia is complaining their ass off while ati smiles?
 
Ati 5xxx series are looking good, I don't believe the gt300 will be another fx5800 saga but we will see.

Even if they are having fab problems im sure they will have a decent product when it's finally released, i guess the questions are how much market will they loose and will the gt300 be faster than ati? (hope its similar performance to drive the prices down :))

OT: Whats with people quoting themselves?, i've read 2 threads tonight with people doing that - There is an EDIT button you know.
 
Silus said:
Ah, so it's half the size now ? All rumors concerning GT300 say it's a <=500 mm2 chip. So unless 330-340mm2 is half of 500 mm2, which math says it isn't, you are grasping at straws...As is the case with Charlie btw.
Click to expand...

That was not related as a direct confrontation between RV870 and GT300. It was a example why can nvidia have much lower yields than ATI. Smaller chip means less possibilities for error.
 
faugusztin said:
That was not related as a direct confrontation between RV870 and GT300. It was a example why can nvidia have much lower yields than ATI. Smaller chip means less possibilities for error.
Click to expand...

So it was related to what exactly ? Plus, we don't even know the actual size of GT300. How can you even compare without proper information ?
 
Can you say "Amd" made nvidia gpu's next year? I could believe that this is some what accurate as nvidia highend gpu's are more complex to build than ati which means the more complex the more that can go wrong. If they were to simplify there gpu's like ati they would probably see better yields.
 
Silus said:
So it was related to what exactly ? Plus, we don't even know the actual size of GT300. How can you even compare without proper information ?
Click to expand...

Will be NVIDIA bigger ? Yes, it will be, without question. And that means, bigger chance for NVIDIA to have problems with yields than ATI with it's 330mm2 size. Even if GT300's size will be the last rumored 452mm2, that is still over 100mm2 bigger chip than the one from ATI.
 
Looks like yields really are bad as Nvidia is strangely silent in the weeks running up the AMD 5000 series launch. I think at best Nvidia is only 1 or 2 Q behind in their roadmap. If you want to buy Nvidia right now, wait till after the ATI launch because the Nvidia will have to respond with pricecuts to stay competitive. I love competition.
 
JUSTJOEL73 said:
Looks like yields really are bad as Nvidia is strangely silent in the weeks running up the AMD 5000 series launch. I think at best Nvidia is only 1 or 2 Q behind in their roadmap. If you want to buy Nvidia right now, wait till after the ATI launch because the Nvidia will have to respond with pricecuts to stay competitive. I love competition.
Click to expand...

Not saying they aren't bad, but some of the best thunder stealing PR is letting your opponent put out his good stuff and then coming out with your "beta" stuff and "beating the crap out of it" in benchmarks.

Note that "beating" your opponent by PR stunts will often involved rigged tests to exaggerate the difference. SOP in the tech industry, unfortunately, regardless if you AMD/ATI, Intel, Seagate, Nvidia, et al.
 
faugusztin said:
Will be NVIDIA bigger ? Yes, it will be, without question. And that means, bigger chance for NVIDIA to have problems with yields than ATI with it's 330mm2 size. Even if GT300's size will be the last rumored 452mm2, that is still over 100mm2 bigger chip than the one from ATI.
Click to expand...

"without question" ? Being influenced by Charlie I see...

And what if it is bigger ? If it ends up faster, what problem does that cause ? Oh right, you must also be assuming that it won't be faster...

It's no surprise to see AMD fans jump on any horrible speculation in regards to NVIDIA and any tremendously good speculation in regards to AMD, but damn...there must be a limit to this madness...and it's not Sparta...
 
JUSTJOEL73 said:
Looks like yields really are bad as Nvidia is strangely silent in the weeks running up the AMD 5000 series launch. I think at best Nvidia is only 1 or 2 Q behind in their roadmap. If you want to buy Nvidia right now, wait till after the ATI launch because the Nvidia will have to respond with pricecuts to stay competitive. I love competition.
Click to expand...

Great conclusion there...If they are silent, that means yields are bad :rolleyes:

The last time there was this much lack of info on NVIDIA's new chip...well, I think you know what happened. So history does not approve of your conclusion, even more so when it's based on a Charlie link...
 
Its charlie, hes a fanboy idiot for no apparent reason. I cannot believe that people get paid to make shit up entirely
 
Mr. K6 said:
We're sorry someone posted depressing information about your favorite company. Would you like a tissue? *group hug*
Click to expand...

Someone ?! It's the same guy...always the same guy...

And it's no shocker that you take this info as "depressing" i.e. bad for NVIDIA, because you believe in it, even though, as usual, there's nothing to back it up...

Given all the AMD banners in your Charlie's website, maybe you get some bonus for promoting his links ?
 
Don't get why you guys are causing such a fuss over Charlie. He posted some rumors, Kyle has heard the same rumors from other sources, so just wait for the cards to be benchmarked upon release before you buy.

Bunch of pissy girls in this thread, damn.
 
Silus said:
Someone ?! It's the same guy...always the same guy...

And it's no shocker that you take this info as "depressing" i.e. bad for NVIDIA, because you believe in it, even though, as usual, there's nothing to back it up...

Given all the AMD banners in your Charlie's website, maybe you get some bonus for promoting his links ?
Click to expand...
Why do you think it's just Charlie? Kyle came in the thread and said he heard the same information form a couple of sources.
 
Mr. K6 said:
Why do you think it's just Charlie? Kyle came in the thread and said he heard the same information form a couple of sources.
Click to expand...

And why do you think they aren't the same sources for both Charlie and Kyle ?

Or the fact that Kyle "heard" it makes it true ?
 
Silus said:
"without question" ? Being influenced by Charlie I see..
Click to expand...

Huh. What does Charlie have with simple facts, i even didn't read the linked article.

512 bit interface have it's requirements for size, chip of same or higher complexity than GT200 can't be much smaller - 65nm version pf GT200 was 576mm2, 55nm GT200b was around 490mm2. By simple math 40nm version of GT200 would be 350mm2, but you can't go much lower than GT200b size due 512-bit memory interface. So yes, GT300 will be a bigger than RV870, and not by a small margin. And if you have more cores (330 vs 450mm2) on one waffer, then you have a lot bigger chance to have more working cores. It's not a rocket science, it's a simple math.

But why do i argue with a blind nvidia fanboy :rolleyes:.
 
faugusztin said:
Huh. What does Charlie have with simple facts, i even didn't read the linked article.

512 bit interface have it's requirements for size, chip of same or higher complexity than GT200 can't be much smaller - 65nm version pf GT200 was 576mm2, 55nm GT200b was around 490mm2. By simple math 40nm version of GT200 would be 350mm2, but you can't go much lower than GT200b size due 512-bit memory interface. So yes, GT300 will be a bigger than RV870, and not by a small margin. And if you have more cores (330 vs 450mm2) on one waffer, then you have a lot bigger chance to have more working cores. It's not a rocket science, it's a simple math.

But why do i argue with a blind nvidia fanboy :rolleyes:.
Click to expand...

You're right about size requirements for a 512 bit memory interface, however you are way off in regards to how much "lower" it can get for a 512 bit memory interface. R600 had one and it measured around 420mm2 @ 80 nm. Also your calculations for GT200 @ 40 nm are bit on the high side, since there's much to consider like transistor density, etc. That said, GT200 @ 40nm wouldn't go much over 300mm2, if at all. And this is assuming a direct shrink, without modifications, which isn't the case of GT300, since it's an entirely new architecture.

And I'm not "arguing" with you, though you seemed like a Charlie cheerleader, with your "without question", without any kind of info to back you up. Although not entirely correct, at least you tried to prove your point, with something other than "Charlie's links". Kudos for that.

As for your "fanboy" remark, sure I'm one. A fanboy of actual trustworthy news sources, with actual proof or other sources, to back them up. You know, that's called "journalism", though Charlie flunked that class, if he ever attended one...
 
And given the lack of an edit button :) I forgot to add that we don't even know for sure that GT300 will have a 512 bit memory interface. With the use of GDDR5, NVIDIA can use 256 or even 384 bit memory interfaces and get the targeted bandwidth.
 
You must log in or register to reply here.
Back
Top