480 GTX: 275 watt TDP?

[OCGuy]

The threads here really need to be merged.....


If anyone is dumb enough to run their high performance PC 24/7, they deserve whatever their power bill is.

If anyone is dumb enough to think TDP = 24/7 power usage, they need to go post at a less educated forum.

 

[3DVagabond]

I think the importance of this is that it shows there's really no room left for Nvidia to improve on performance without a fairly dramatic change. They're right at the edge of the PCI power limits. It really seems that a dual GPU Fermi is out of the question. The 480 will probably be the end all of performance for Nvidia until the next shrink.

 

[BenSkywalker]

So Charlie claims that he is getting his information from nV partners, he claims that TDP is 275 watts and this a 50 watt increase over a few weeks ago when he said it was a 280 watt part....... I'm having a bit of trouble with the math, not gonna lie about that. He also bring up Oak Ridge again-

A little later we got on the phone with someone in Computing and Computational Sciences at Oak Ridge National Laboratory and they stated that the SemiAccurate article was inaccurate

http://www.legitreviews.com/news/6999/

Doesn't he have the sense to not bring back up points where he has been proven a liar?

Mabe they raised the clocks to the target 750 from 625?

The only time 750MHZ has been quoted as a target clock for Fermi was on an ATi slide. Noone associated with nVidia has ever used that as a target clock, in fact based on the performance numbers nVidia has been quoting since they first started talking about Fermi 625MHZ is on the high side of their targets.

 

[3DVagabond]

I think he's claiming that 275 is 50 higher than t he 225 that was originally "supposed" to be what the 480 was to draw.

I think calling him a liar is too harsh. He's a journalist trying to "scoop" everyone. Sometimes he's wrong, I guess. I haven't read a lot of his stuff before Fermi though. That must have been where he got this rep I hear about him not being accurate. Either that or people are just haters. He's been pretty much spot on with Fermi, so far.

 

[Keysplayr]

I think he's claiming that 275 is 50 higher than t he 225 that was originally "supposed" to be what the 480 was to draw.

I think calling him a liar is too harsh. He's a journalist trying to "scoop" everyone. Sometimes he's wrong, I guess. I haven't read a lot of his stuff before Fermi though. That must have been where he got this rep I hear about him not being accurate. Either that or people are just haters. He's been pretty much spot on with Fermi, so far.

What he does isn't journalism. It's blogging.

 

[BenSkywalker]

I think he's claiming that 275 is 50 higher than t he 225 that was originally "supposed" to be what the 480 was to draw.

According to him the 225 Watt figure was for the 470.

The 448 SP version is listed at 225W TDP

That is from a previous article.

He's been pretty much spot on with Fermi, so far.

He has been right about it being late, that's about it.

Then there is the whole problem of 512 SPs promised versus 448 SPs delivered.

Another one of his quotes- wrong again. His repeated quoting of 725MHZ as a target speed and saying it was a nV design goal which came from an ATi press slide(this one seemed to fool a lot of people too). He also claims to be a fab expert but doesn't even know what an A0 stepping is(making everything else he says laughable on the subject honestly). The only thing we have seen Charlie get 'right' is that Fermi would be later then some expected, that's it.

 

[3DVagabond]

Only 1/2 those quotes are mine. I'm not sure where you got the others from. Don't quote me out of context, please.

 

[dug777]

If there's anything more tragic than ferocious bickering on the internet about the past speculations of a tech blog, I can't think of it

Who cares what this bloke said? It's hardly relevant to how Fermi finally turns out, and we'll know for absolutely certain very soon.

 

[BenSkywalker]

Only 1/2 those quotes are mine. I'm not sure where you got the others from. Don't quote me out of context, please.

Half the quotes are Charlie's, I mentioned which are from him in the post and at no point attributed them to you.

Who cares what this bloke said?

A lot of people who are tired of his idiocy being touted as fact. The man doesn't know what an A0 stepping is and he is supposed to be an expert on fabbing chips. If there weren't so many suckers that actually believed what he posted I don't think anyone would care much, that he takes in so many people is why people respond to his BS.

 

[ViRGE]

So Charlie claims that he is getting his information from nV partners, he claims that TDP is 275 watts and this a 50 watt increase over a few weeks ago when he said it was a 280 watt part....... I'm having a bit of trouble with the math, not gonna lie about that.

Perhaps I'm just reading this wrong, but I think the 50W increase is in reference to what NVIDIA is telling their partners as official guidance - which implies that they were previously told 225W and now are being told 275W.

280W in turn has been the number Charlie's moles have been telling him based on the lab results of the card. The lab results wouldn't necessarily match NVIDIA's guidance if NVIDIA expected the shipping cards to be better tuned.

 

[BenSkywalker]

Perhaps I'm just reading this wrong, but I think the 50W increase is in reference to what NVIDIA is telling their partners as official guidance - which implies that they were previously told 225W and now are being told 275W.

So three weeks before launch nV's partners don't already have the parts? To say that isn't likely is a bit on the conservative side. Charlie implies that his numbers come from nV partners, and then says that nV's partners were surprised that they are using even less power then what they thought, but that is 50 watts higher then it was supposed to be?

 

[ViRGE]

So three weeks before launch nV's partners don't already have the parts?

It certainly sounds that way. AMD did the same thing; no one had Cypress parts until right before the 5870 launch, and that was only so that their partners would have enough time to slap a sticker on a card coming from AMD and throw it in to a box. It sounds like NVIDIA is doing the same thing here.

 

[BenSkywalker]

It sounds like NVIDIA is doing the same thing here.

Someone has to be assembling them for nVidia.

 

[dguy6789]

He has been right about it being late, that's about it.

He has maintained for the past 6 months that Fermi was a huge die, expensive, difficult to manufacture, very power hungry, will be late, will be available in extremely limited quantities when it does finally launch. He was the very first to mention all of this. Virtually nobody believed him.

So far he's got a lot more right than wrong. He's 5 watts off on his TDP prediction.

 

[Soleron]

@BenSkywalker

He still says Oak Ridge did drop Nvidia - he says that article is wrong and Oak Ridge are effectively lying to protect Nvidia. So when we see Oak Ridge computers running Fermis, or not, we'll know the truth.

--

@Everyone

Those queries about what card he means when he says 225W etc. were raised in the forums, and Charlie's connection is intermittent so they didn't get a chance to ask him everything, but he did say a few things and if you have more questions/inconsistencies he WILL answer them on the SA forum when he gets back from GDC. He's perfectly reasonable when you ask him questions like that, by the way.

But the 275W figure refers to the 480 and 448SP Teslas, not the 470. The 'raised 50W' applies to Teslas - final Geforce TDPs were not known earlier, but Nvidia did say less than 225W for the Teslas and now that's 275W. Make sense?

So, yes, they did hike a published number by 50W. And that is for a 448SP part.

 

[Keysplayr]

He has maintained for the past 6 months that Fermi was a huge die, expensive, difficult to manufacture, very power hungry, will be late, will be available in extremely limited quantities when it does finally launch. He was the very first to mention all of this. Virtually nobody believed him.

So far he's got a lot more right than wrong. He's 5 watts off on his TDP prediction.

He is? Where are the official power consumption figures? Dude, as of right this moment, Ben is correct in saying the only thing Charlie got right so far, is that GF100 would be late.
As information becomes available very soon, THEN you can either glorify or crucify Charlie.

 

[Keysplayr]

@BenSkywalker

He still says Oak Ridge did drop Nvidia - he says that article is wrong and Oak Ridge are effectively lying to protect Nvidia. So when we see Oak Ridge computers running Fermis, or not, we'll know the truth.

--

@Everyone

Those queries about what card he means when he says 225W etc. were raised in the forums, and Charlie's connection is intermittent so they didn't get a chance to ask him everything, but he did say a few things and if you have more questions/inconsistencies he WILL answer them on the SA forum when he gets back from GDC. He's perfectly reasonable when you ask him questions like that, by the way.

But the 275W figure refers to the 480 and 448SP Teslas, not the 470. The 'raised 50W' applies to Teslas - final Geforce TDPs were not known earlier, but Nvidia did say less than 225W for the Teslas and now that's 275W. Make sense?

So, yes, they did hike a published number by 50W. And that is for a 448SP part.

No, he isn't perfectly reasonable. He can't even post a positive ATI only blog without cutting Nvidia some sludge. Look at the latest "ATI moves to improve dev relations" or whatever it's called. He needs help. Treatment of some sort. Definitely a bit of sickness there IMHO. And apparently, he will absolutely NEVER get over what Nvidia did to him.

 

[MrK6]

No, he isn't perfectly reasonable. He can't even post a positive ATI only blog without cutting Nvidia some sludge. Look at the latest "ATI moves to improve dev relations" or whatever it's called. He needs help. Treatment of some sort. Definitely a bit of sickness there IMHO. And apparently, he will absolutely NEVER get over what Nvidia did to him.

Sounds like you won't either. Who's worse?

 

[MarcVenice]

No, he isn't perfectly reasonable. He can't even post a positive ATI only blog without cutting Nvidia some sludge. Look at the latest "ATI moves to improve dev relations" or whatever it's called. He needs help. Treatment of some sort. Definitely a bit of sickness there IMHO. And apparently, he will absolutely NEVER get over what Nvidia did to him.

Wait, you don't like what he's saying, don't like some people actually give him some credit, so you try to dismiss him by claiming he's mentally ill? You a shrink?

 

[Lonyo]

Wait, you don't like what he's saying, don't like some people actually give him some credit, so you try to dismiss him by claiming he's mentally ill? You a shrink?

Obviously the only reason someone could so vehemently hate NV is if they were mentally ill.
The normal situation is for people to be in love with NV and worship their every move, so since Charlie seems full of hatred, the only logical explanation must be that he is mentally unbalanced.
Seems fairly simple to me.

 

[dug777]

Wait, you don't like what he's saying, don't like some people actually give him some credit, so you try to dismiss him by claiming he's mentally ill? You a shrink?

meh.

Calling someone mentally ill is not very nice, or indeed something that appears to be based on anything other than a dislike of what the gentleman in questions says.

 

[Soleron]

No, he isn't perfectly reasonable. He can't even post a positive ATI only blog without cutting Nvidia some sludge. Look at the latest "ATI moves to improve dev relations" or whatever it's called. He needs help. Treatment of some sort. Definitely a bit of sickness there IMHO. And apparently, he will absolutely NEVER get over what Nvidia did to him.

His articles are sensationalist and biased. Yes, I agree.

On the forums, he is actually reasonable. Look through his post history. He rarely brings up Nvidia unless the conversation is about it, his arguments are well structured and well written (not personal attacks and informal like articles) and he will respond to most questions asked with explanations and evidence.

I think he deliberately puts on the 'controversial' style for the articles, to get views. It's a shame though, because it makes him look like a raving fanboy and weakens his credibility.

He even says that Nvidia has good engineers and it's only the upper management and PR he's got a problem with.

He is? Where are the official power consumption figures? Dude, as of right this moment, Ben is correct in saying the only thing Charlie got right so far, is that GF100 would be late.
As information becomes available very soon, THEN you can either glorify or crucify Charlie.

Yes, agreed. No one can call it at this point.

 

[Maximilian]

Well it looks like ATI has won this round.

 

[hectorsm]

He has maintained for the past 6 months that Fermi was a huge die, expensive, difficult to manufacture, very power hungry, will be late, will be available in extremely limited quantities when it does finally launch. He was the very first to mention all of this. Virtually nobody believed him.

So far he's got a lot more right than wrong. He's 5 watts off on his TDP prediction.

NV chips in the last few generation have been big, expensive, power hungry and difficult to manufacture. Considering NV’s monolithic design philosophy that will continue to be true. Anyone can predict the same thing will happen in the next model released after fermi.

However, reporting that Fermi was going to be late is the only thing I will give him credit for.

 

[Tempered81]

Donamin Haber has a news post saying that the 480 OC model is rated at 305 watts, and that normal models go 270-298 watts. I suspect it's the zotec AMP! that pulls 305w. However, the article also states that although he has numbers for power draw in his documentation, he still cant see stock or OC clock & shader speeds... Sounds unbelievable to me. How could he see internal documents with TDPs but not clockspeed ratings.

All the bs aside, I think the 480 model will come in between 260w and 280w for Nvidia's TDP spec