NV: Everything under control. 512-Fermi may appear someday. Yields aren't under 20%

[CitanUzuki]

Hey BenSkywalker are you affiliated with nvidia in any way?

 

[Keysplayr]

Weird thinking. All cars including fast ones brag about gas consumption if it compare well with the competition. Even fix or repair daily has learned that lesson.

Indeed. They also brag about options and features. Horsepower. Things that nobody pays any attention to.

 

[Keysplayr]

Hey BenSkywalker are you affiliated with nvidia in any way?

No, he isn't. Even though you didn't ask me.

 

[thilanliyan]

Go into the Garage forum, ask the posters there if they think the CTS-V would be a better car if it had the 300hp V6 instead of its' current engine.

The CTS-V is a "tuned" version of a car that there is a "regular" version of. If you're looking at a CTS-V then you know you will be getting a larger engine with the associated pluses and minuses. For your analogy to make sense, the regular CTS would have to have only one engine choice wouldn't it? If only the V8 was available, customers looking for a "regular" CTS should and probably would complain that they are not looking for huge performance. Like I said, I don't disagree with the points you made about the video cards, but I don't think the car analogy works in this instance because there are so many more markets and price points in the auto industry compared to video cards, even looking above a certain performance threshold.

 

[Keysplayr]

The CTS-V is a "tuned" version of a car that there is a "regular" version of. If you're looking at a CTS-V then you know you will be getting a larger engine with the associated pluses and minuses. For your analogy to make sense, the regular CTS would have to have only one engine choice...the V8, in which case customers looking for a "regular" CTS should and probably would complain that they are not looking for huge performance. Like I said, I don't disagree with the points you made about the video cards, but I don't think the car analogy works in this instance because there are so many more markets and price points in the auto industry compared to video cards, even looking above a certain performance threshold.

Now you're catching on. Perfect analogy as Nvidia is also looking into more markets, not just a video card, but more.

 

[thilanliyan]

Now you're catching on. Perfect analogy as Nvidia is also looking into more markets, not just a video card, but more.

As I said in my post, I agree with what he said about the video cards but I don't think that specific car analogy makes sense. To me it would make more sense if the V8 was dropped into a Chevy Malibu or something (and was the only engine choice)...in which case there would be more complaints than just "fans of the competition" as Ben said. And relating it back to Fermi, I'm pretty sure not only ATI fans would complain about the negative points of Fermi (namely power, heat, noise, and possibly price). Maybe I'm taking the analogy too literally...or I'm looking at it from a different perspective.

 

[SlowSpyder]

People comparing power requirements have no business talking about high end cards. simple as that. Its like buying a fast car and crying cause the engine gets hot and you use lots of gas.

I disagree. While power consumption can certainly be blown out of proportion, I think it does have a place when comparing two similarly performing cards. On average I'd say the GTX480 is around 15% faster than a 5870... It's definitely the faster part. But, in my opinion the much more well rounded 5870 is close enough in performance that it's worth looking at things like power consumption as well. I think when you have two cards that perform in the same ball park, power consumption is a perfectly worthwhile aspect to add to the comparison.

If Fermi was 40% faster on average and pulled an additional 50% power, I don't think anyone would say a word about it. It would justify that level of power consumption. But 15% faster with 30%+ more power consumption, well that might be worth talking about to some people.

Also, you have to remember, higher power consumption will mean more heat. More heat can mean a louder cooling solution. Some of us do care about noise.

 

[Keysplayr]

As I said in my post, I agree with what he said about the video cards but I don't think that specific car analogy makes sense. To me it would make more sense if the V8 was dropped into a Chevy Malibu or something (and was the only engine choice)...in which case there would be more complaints than just "fans of the competition" as Ben said. And relating it back to Fermi, I'm pretty sure not only ATI fans would complain about the negative points of Fermi (namely power, heat, noise, and possibly price). Maybe I'm taking the analogy too literally...or I'm looking at it from a different perspective.

I guess it's a perspective thang. Funny, I actually considered the CTS-V when I was car shopping in October. Anyways, for all the negatives about Fermi, it is still fully loaded. Pros and cons.

 

[ronnn]

I certainly hope gm gets better than a 20% yield, which brings us back to nv pr.

 

[thilanliyan]

Anyways, for all the negatives about Fermi, it is still fully loaded. Pros and cons.

No doubt. I think there's a lot of positives performance wise. If I actually played more games I might actually consider getting a GTX470. As it stands now I won't be looking to get any card for a while...maybe at the end of the year.

 

[Keysplayr]

I certainly hope gm gets better than a 20% yield, which brings us back to nv pr.

What is the ACTUAL yield? If it's not less than 20%, I'm sure it's something like 20.00001%.

 

[blanketyblank]

What is the ACTUAL yield? If it's not less than 20%, I'm sure it's something like 20.00001%.

I agree with that. Considering how much NV's PR will exaggerate things if it's positive if it was at least 25% the statement would be our yields are WAY OVER 20% plus some laughing and jeering at unmentioned rumor sites.
I know how a lot of people say Charlie gets a big head and thinks NV is talking about him, but honestly I think sometimes they actually are.

 

[Lonyo]

I agree with that. Considering how much NV's PR will exaggerate things if it's positive if it was at least 25% the statement would be our yields are WAY OVER 20% plus some laughing and jeering at unmentioned rumor sites.
I know how a lot of people say Charlie gets a big head and thinks NV is talking about him, but honestly I think sometimes they actually are.

Well the interesting thing is that there were two stories.
S|A claimed yields were under 20%.
Needham and Company claim [...] that the yields of Fermi chips is between 20% and 30%.
http://www.xbitlabs.com/news/video/...t_20_30_Further_Delays_Possible_Analysts.html

Now, they denied the <20% S|A rumour, saying it wasn't under 20%, but effectively did not comment on the other rumour (since that specified above 20% as a low point). You could suggest that either the yield is indeed in that 20~30% range (so they couldn't deny it without lying), or that NV feels that they need to answer to S|A more than some financial analysts, or that they were only aware of the S|A yield claim (the middle option being the least likely).
That doesn't mean it is definitely true, but NV haven't said it's untrue, while they have said another claim was untrue, so...

 

[3DVagabond]

For those who are using the performance car analogies to make a point that efficiency isn't important, how can you even argue the point? If anything the automobile business, even where performance cars are considered, was overwhelmingly dominated by the more efficient designs. Even professional racing needs to consider fuel economy.

 

[Keysplayr]

For those who are using the performance car analogies to make a point that efficiency isn't important, how can you even argue the point? If anything the automobile business, even where performance cars are considered, was overwhelmingly dominated by the more efficient designs. Even professional racing needs to consider fuel economy.

I'm sorry but who says efficiency isn't important? That isn't what that car analogy was for. You misunderstood it.
And if you've noticed in the auto industry, most performance cars are subject to a gas guzzler tax. I should know, I just had to pay it in October.
And the only time I could see professional racers considering fuel economy is cross desert racing. One place you don't want to run out of fuel.
And you don't have to pit a Fermi.

 

[Lonyo]

I'm sorry but who says efficiency isn't important? That isn't what that car analogy was for. You misunderstood it.
And if you've noticed in the auto industry, most performance cars are subject to a gas guzzler tax. I should know, I just had to pay it in October.
And the only time I could see professional racers considering fuel economy is cross desert racing. One place you don't want to run out of fuel.
And you don't have to pit a Fermi.

Formula 1, NASCAR, Le Mans, Indycar (or whatever it's called these days).
The only time you don't consider fuel economy in racing is when it's short races.
As soon as you have long races, you start to consider fuel economy both due to weight issues, and the time it takes to pit.

Most people who buy expensive cars to use on the roads don't care about fuel economy, because they are typically rich. That's what NV is arguing for Fermi. If you can afford it and want to buy it, you want performance, and you should be rich enough not to need to worry about ancillary costs.
For $500, you could buy a full "good enough" computer for most typical home users. That's not who Fermi is aimed at.

And car analogies are really terrible. Really really terrible.


Reply to Keys' reply below this post because there's no need for more posts on it:

Wouldn't you say any one of those vehicles in the races you listed (Formula 1, NASCAR, Le Mans, Indy) use a tad more fuel than a lesser performing vehicle? Say a corvette, ferrari, lamborghini, viper?
Yes, the car analogies aren't the greatest.
The main point was that it isn't unheard of for a higher performing GPU to utilize more power, generate more heat, make more noise.

While performing the same task? No. I would expect that a Dodge Viper going at 150mph average with lots of acceleration and deceleration like an F1 car does, or going at say 175mph near constantly like an oval track racing car would be less efficient, due to being heavier and not being as aerodynamic etc.

Nonetheless a Formula One engine is over 20&#37; more efficient at turning fuel into power than most small commuter cars, considering their craftsmanship

In much the same way Top Gear did a test comparing an M3 to a Prius and got better fuel economy from the M3 because of the speeds.

So lets stop talking about comparing the efficiency of different cars, and maybe stop talking about power altogether other than to say that if you can afford a $500 Fermi, you can afford to pay the power bill.

 

[Keysplayr]

Formula 1, NASCAR, Le Mans, Indycar (or whatever it's called these days).
The only time you don't consider fuel economy in racing is when it's short races.
As soon as you have long races, you start to consider fuel economy both due to weight issues, and the time it takes to pit.

Most people who buy expensive cars to use on the roads don't care about fuel economy, because they are typically rich. That's what NV is arguing for Fermi. If you can afford it and want to buy it, you want performance, and you should be rich enough not to need to worry about ancillary costs.
For $500, you could buy a full "good enough" computer for most typical home users. That's not who Fermi is aimed at.

And car analogies are really terrible. Really really terrible.

Wouldn't you say any one of those vehicles in the races you listed (Formula 1, NASCAR, Le Mans, Indy) use a tad more fuel than a lesser performing vehicle? Say a corvette, ferrari, lamborghini, viper?
Yes, the car analogies aren't the greatest.
The main point was that it isn't unheard of for a higher performing GPU to utilize more power, generate more heat, make more noise.
Can you imagine if GTX480 had launched with all 512 cores, 4 additional TMUs, additional polymorph engine and clocked at the 750MHz/1500 everyone said it was "supposed" to be?
40nm was/is a biotch for both sides unfortunately. More so for Nvidia with a far more complex die. A billion more Xtors.

 

[Creig]

Can you imagine if GTX480 had launched with all 512 cores, 4 additional TMUs, additional polymorph engine and clocked at the 750MHz/1500 everyone said it was "supposed" to be?

At a guess, I would say it would have broken the 300w TDP barrier and subsequently not been granted PCI-E approval.

 

[Keysplayr]

At a guess, I would say it would have broken the 300w TDP barrier and subsequently not been granted PCI-E approval.

Yes Creig, and if 40nm went swimmingly, what else would it be?

 

[Lonyo]

Yes Creig, and if 40nm went swimmingly, what else would it be?

Beneficial for consumers?

 

[Creig]

Yes Creig, and if 40nm went swimmingly, what else would it be?

Well, then I'd imagine that Fermi would be faster, run cooler and have better yields than it does right now. And ATi wouldn't have had to double up on vias, so their chips would also be faster, run cooler, have better yields and be smaller than they are right now. But that's all just wishful thinking. TSMC 40nm is what it is.

 

[Pantalaimon]

Yes Creig, and if 40nm went swimmingly, what else would it be?

Could have, would have, should have. The fact remains that they weren't able to manufacture the card in an acceptable way according to what they originally designed. No wishful thinking now can improve the card from what it is in its current form.

 

[imaheadcase]

http://www.supercars.net/cars/3335.html

?? has nothing to do with what i said.

 

[imaheadcase]

Weird thinking. All cars including fast ones brag about gas consumption if it compare well with the competition. Even fix or repair daily has learned that lesson.

I've never seen that. I highly doubt people dropping 100k -1million dollars car about such things as well. If you bought a $100 video card and it sounded like a GTX480 then its something worth complaining about.

 

[imaheadcase]

I disagree. While power consumption can certainly be blown out of proportion, I think it does have a place when comparing two similarly performing cards. On average I'd say the GTX480 is around 15% faster than a 5870... It's definitely the faster part. But, in my opinion the much more well rounded 5870 is close enough in performance that it's worth looking at things like power consumption as well. I think when you have two cards that perform in the same ball park, power consumption is a perfectly worthwhile aspect to add to the comparison.

If Fermi was 40% faster on average and pulled an additional 50% power, I don't think anyone would say a word about it. It would justify that level of power consumption. But 15% faster with 30%+ more power consumption, well that might be worth talking about to some people.

Also, you have to remember, higher power consumption will mean more heat. More heat can mean a louder cooling solution. Some of us do care about noise.

My argument goes the same with noise, not worth crying about with a high end card. Unless you plan on gaming completely without sound. Never mind you can control fan speed or use aftermarket cooling. Heck most people I know use headphones so not a issue.

You are buying a high end card for gaming right people? Not just to cry and run benchmarks 24/7 and compare the box design it game in?