Charlie D Claim Watch

[MrK6]

This is quite clearly black and white. Correct or not. Charlie was wrong, on pretty much everything(he was wrong on absolutely everything at least most of the time).

No, it's not, that's what you're missing. It's a rumor site, that publishes... wait for it... RUMORS. Pretty neat, huh? He was spot on characterizing Fermi... too hot, too loud, too late (or whatever that quotation was). If you want to get caught up in the details, it just shows how much you're grasping at straws (for some obtuse reason, it's some dude on the internet, go outside).

There's not much more I can say here: I was obviously able to glean the correct information out of the rumor sites, as here I am sitting on a pair of 5850's at what I was able to calculate to be 30-40% faster performance at less noise and heat than a stock GTX480. Evidently some of you lack this fundamental critical thinking capability, as I really can't break it down any further.

 

[Daedalus685]

This is quite clearly black and white. Correct or not. Charlie was wrong, on pretty much everything(he was wrong on absolutely everything at least most of the time). /snip

Oh come on.. that is partisan bull shit... No one with half a brain should realistically think that the only options are right or wrong.

Was he perfect? Certainly not...

Did he do OK? I'd say so. You are welcome to disagree (as you obviously do).

As for the comment about the 5870 being slow and not worth launching.. you are putting words in his mouth... The last thing we need is more BS about what he does and does not say. He said the gtx280 was unreleasable based on yields... he was talking out of his ass... but not once did he mention the 5870 in that spiel. Perhaps that is a good indication he is a biased writer, but lets not be spewing hate about things he never said.

I don't suppose science is your thing eh? You'd be a mighty depressed chap the first time you got results back from an experiment and found out you were 100% wrong about everything... In fact you'd be hard pressed to find a single scientist anywhere that is not 100% wrong all of the time given your absolute definitions of the concept.

He sells rumours... Rumours that may or may not have been correct at the time. If you can't weigh a rumour together with current facts, based on how things might be at the time then there is no point in reading any rumours at all. They are subect to change without notice and because a product changes to prevent the exact doom fore told by a rumour does not make the rumour wrong...

He is not right or wrong.. he gives information that this time seems to indicate he has some knowledge... but checking your logic and reason at the door is not advised.

/baffle at all the ludicrous focus on black and white... we don't live in a polar world.. there are more than two options for every situation.

 

[Rezist]

He got alot right, that's why were posting here.

 

[BenSkywalker]

Regardless him being a jackass regarding NVIDIA (and is mutual) the fact is that either he was extremely lucky or he did in fact have some reasonable reliable sources.

Wrong on clockspeeds, check
Wrong on functional units, check
Wrong on power useage, check

What was he correct about? That it was going to be a big chip and it was going to use more power then a small chip? I could have told you that before nVidia started designing it as could anyone who knew much at all about the company. Everything else he made up as he went. On factual counts that he claimed, he was wrong. Perhaps people consider him to be right by the fact that he figured out a chip with 50% more transistors was going to run hotter then its' smaller counterpart? Honestly if that is what people consider having good insider information, well, I would reccomend reading up a bit more on microprocessor design so you can see how much of a leap that is.

I don't suppose science is your thing eh?

Actually, it is.

You'd be a mighty depressed chap the first time you got results back from an experiment and found out you were 100% wrong about everything

I'd be an ignorant jackass if I thought I could claim accuracy on an experiment before it happened. Perhaps my education is old fashioned, you run the experiment to find out the results.

In fact you'd be hard pressed to find a single scientist anywhere that is not 100% wrong all of the time given your absolute definitions of the concept.

Really? Atomic weight of Hydrogen change lately?

 

[GaiaHunter]

Wrong on clockspeeds, check
Wrong on functional units, check
Wrong on power useage, check

Well even NVIDIA was WRONG ON CLOCKSPEEDS, FUNCTIONAL UNITS AND POWER USAGE! AND LAUNCH DATE!

Or the Fermi whitepaper was a fiction?

 

[BenSkywalker]

Well even NVIDIA was WRONG ON CLOCKSPEEDS, FUNCTIONAL UNITS AND POWER USAGE! AND LAUNCH DATE!

Anyone on these forums claiming nVidia is some beacon of virtue and accuracy? Didn't think so.

 

[Daedalus685]

Wrong on clockspeeds, check
Wrong on functional units, check
Wrong on power useage, check

He never quoted clock speeds or power consumption of a 480 core product. How can he be wrong?

Would you rather have a weather report that states "It is likely going to rain tomorrow" or "It will not rain tomorrow" if the outcome is 100% precipitation, but it might dip below freezing?

Certainly "It will snow tomorrow" would be desired.. just like Charlie claiming that "The fermi gtx480 will be 480 cores at 700mhz...." woudl be desired. However, there is a large gamut of possibilities in any situation, specifically rumours like this which are likely even more challenging than the weather. There is such a thing as being usefully 'wrong'. I'd much rather plan for a rainy day and get snow than plan for a sunny day and get snow...

For a rumour you can't simply grade it on how it performs exactly compared to the after the fact reality.. it doesn't work that way as everything in life tends to be fluid. Would it help if each article started with "This is how it would be if the part came out today!" Truth like this does expire you know.

 

[BenSkywalker]

He never quoted clock speeds or power consumption of a 480 core product. How can he be wrong?

I'm assuming that is tongue in cheek as really the fact that he never claimed the product would ship how it actually did speaks volumes about his accuracy

Would you rather have a weather report that states "It is likely going to rain tomorrow" or "It will not rain tomorrow" if the outcome is 100% precipitation, but it might dip below freezing?

100% chance of precipitation with possible wintry mix. Not complicated.

 

[bryanW1995]

Your argument is flawed. You plan on having some noise when buying a high end card.
Same goes for power requirements, you plan on it.

Again, you don't buy a $500+ card caring much for sound.

not true, my gtx 260 core 216 OC is ~ a gtx 275 and is typically quiet as a mouse. A big reason I picked it over a 4870 1gb was noise in fact.

 

[waffleironhead]

Wrong on clockspeeds, check
Wrong on functional units, check
Wrong on power useage, check

There was a reason nV had to respin as many times as they did. Which spin were you quoting those numbers from?

 

[BenSkywalker]

There was a reason nV had to respin as many times as they did. Which spin were you quoting those numbers from?

The only one that matters, shipping product. Each individual chip is going to have differing limits on clockspeeds, voltage requirements and how many functional units will operate under each of the former for ideal performance, even in the final shipping products. The same could be said about the 5870 with some people clearing a GHZ easily, others aren't as lucky. When reporting on what a video card will ship at, normally it is a good idea to report the numbers it will ship at. He claimed he had insider information and never reported on what actually shipped in terms of configuration despite his shotgun efforts to try and cover every possible scenario so at least one of them would be correct.

 

[GaiaHunter]

Anyone on these forums claiming nVidia is some beacon of virtue and accuracy? Didn't think so.

So you say Charlie is wrong because NVIDIA didn't launch any 512 shaders GTX480 running at 600-625 MHz?

So basically Charlie been saying all along that NVIDIA was having problems to get the 512 shaders part out at competitive performance and because NVIDIA didn't get one out he is wrong?

I guess that 512 shaders is really unmanufacturable, so much that NVIDIA had to change strategy and go with 480 part which is a 512 part with disabled shader cores.

 

[Lonyo]

Wrong on clockspeeds, check
Wrong on functional units, check
Wrong on power useage, check

What was he correct about? That it was going to be a big chip and it was going to use more power then a small chip? I could have told you that before nVidia started designing it as could anyone who knew much at all about the company. Everything else he made up as he went. On factual counts that he claimed, he was wrong. Perhaps people consider him to be right by the fact that he figured out a chip with 50% more transistors was going to run hotter then its' smaller counterpart? Honestly if that is what people consider having good insider information, well, I would reccomend reading up a bit more on microprocessor design so you can see how much of a leap that is.

Actually he was right about power consumption. ~295w
Or at least, based on every other card it would be rated at ~295w. NV just decided to fudge the numbers and then make a wild claim for 250w.

 

[Janooo]

Anyone on these forums claiming nVidia is some beacon of virtue and accuracy? Didn't think so.

No matter what you say Charlie was pretty accurate.
The fact is that people in the industry read/listen to him because they know that he has some insiders with a good info. He is getting some respect.

Your opinion is not important in this regard and it won't change a thing.

 

[BenSkywalker]

So you say Charlie is wrong because NVIDIA didn't launch any 512 shaders GTX480 running at 600-625 MHz?

Or do you want to use the 600MHZ 448 core 480 version of Charlie's story? The biggest contradiction to Charlie has been Charlie himself for a long time. He's made so many claims it is shocking that he managed to get them all wrong, he really should have made up one more story so he could have at least gotten the shader count right.

The fact is that people in the industry read/listen to him because they know that he has some insiders with a good info. He is getting some respect.

Like nV exiting the high end GPU market quarters ago, and then they had higher gross profits then ATi had total revenue? That kind of important insider information? Nothing Charlie has stated is going to be taken seriously by anyone inside the industry. It is fodder for die hard loyalists on forums and that's it. The man doesn't even know what an initial tape out stepping is- to say that he doesn't know what he's talking about is obvious is being kind. He guessed that a new nVidia architecture was going to be big, hot, and power hungry. When has that not been true since the launch of the GeForce?

 

[Daedalus685]

I'm assuming that is tongue in cheek as really the fact that he never claimed the product would ship how it actually did speaks volumes about his accuracy



100% chance of precipitation with possible wintry mix. Not complicated.

/facepalm

No, I was being quite serious... You don't seem to understand that all things are not absolutes..

It is not good enough to evaluate a rumour simply on how it relates to the facts months after they were said. There is such a thing as a not entirely correct (wrong) rumour that is still useful. Do you know what a rumour is?

As for the weather thing... yes, thank you... I am well aware how it 'should' be reported or how we want it to be reported (hence why I bloody said that in the post you quoted...). It was an example of how ridiculous it is to evaluate information as only right or wrong. The chap that claims it will rain when it snows is still wrong.. but he is certainly much more useful than the chap that claims it will not rain (or that it will be sunny).

To stick to weather.. Most of the folks that report it are absolutely wrong almost all of the time... Hence why they deal in relatives, chances, probabilities etc. If you hear it will be 28C tomorrow will you freak out if it gets to 29? At what point does the usefulness of the information trump the absolute fact that it was wrong by an overly pedantic definition?

You are entirely correct to say that he was 'wrong' about pretty much everything.. but not in a negative way.. He gave very useful information, whether the numbers ended up perfect in the end or not.

 

[GaiaHunter]

The only one that matters, shipping product. Each individual chip is going to have differing limits on clockspeeds, voltage requirements and how many functional units will operate under each of the former for ideal performance, even in the final shipping products.

What about time frame?

NVIDIA was all this time sitting idle?

Was a static scenario?

 

[Daedalus685]

Or do you want to use the 600MHZ 448 core 480 version of Charlie's story? The biggest contradiction to Charlie has been Charlie himself for a long time. He's made so many claims it is shocking that he managed to get them all wrong, he really should have made up one more story so he could have at least gotten the shader count right.

He never claimed that....

He claimed that the highest volume bin would be 448 cores... At the time he thought there would be no "480" but for a few 512 core parts. This second tier bin ended up what we call the 470, just happens that there are more 480's than he expected because they use a lower bin..

 

[waffleironhead]

Or do you want to use the 600MHZ 448 core 480 version of Charlie's story? The biggest contradiction to Charlie has been Charlie himself for a long time. He's made so many claims it is shocking that he managed to get them all wrong, he really should have made up one more story so he could have at least gotten the shader count right.



Like nV exiting the high end GPU market quarters ago, and then they had higher gross profits then ATi had total revenue? That kind of important insider information? Nothing Charlie has stated is going to be taken seriously by anyone inside the industry. It is fodder for die hard loyalists on forums and that's it. The man doesn't even know what an initial tape out stepping is- to say that he doesn't know what he's talking about is obvious is being kind. He guessed that a new nVidia architecture was going to be big, hot, and power hungry. When has that not been true since the launch of the GeForce?

Could you link to the said story?

 

[blckgrffn]

heh, arguing with Ben? Impossible, give up, ignore!

His argument style is quite impressive to observe, however... pick one line (or a collection of single statements individually) and dispute it, at which time you'll write a lengthy rebuttal giving him something at least disputable. Repeat.

Pretty sure it is some sort of fallacy to dismiss an argument based on one thing, but as we can see here Ben loves his absolute statements

*not really trying to stir the pot here, just laying down my observations, I typically respect Bens educated viewpoint and opinions*



I have to agree that I think Charlie was more right than wrong and I'll be interested in what he has to say about future video card releases, but will not take his info as absolute.

Which gets him page hits and money, so I think he served himself well on this round.

For my part, I am bummed nvidia didn't come out with something better. The 5850 was supposed to follow in the steps of the 4850 and get really affordable, but now I might have to settle for a 5770, and even those won't be price pressured with this release.

 

[BenSkywalker]

For those not interested in the 'why' side of things, the short answer is that the top bin as it stands now is about 600MHz for the half hot clock, and 1200MHz for the hot clock, and the initial top part will have 448 shaders. On top of that, the fab wafer yields are still in single digit percentages.

He tells so many lies, I understand why you all may get confused.

http://www.semiaccurate.com/2010/02/17/nvidias-fermigtx480-broken-and-unfixable/

 

[bryanW1995]

Anyone on these forums claiming nVidia is some beacon of virtue and accuracy? Didn't think so.

I have two words for you:



Wreck
Age



Overall it seems that charlie did very well this time around. It is extremely annoying to weed through his intense anti-nvidia rhetoric, however. Hopefully going forward he will tone that down a bit, he might even get hits from nvidia fans once in a while if he does because his intel this time (and last time for that matter) seems to have been pretty good.

I don't see a 700 mhz gtx 480 with 480 cores and a significantly higher volume gtx 470 with 448 cores as being inconsistent with reasonable expectations that you could have made from reading charlie's articles. In fact, he's so anti-nvidia that most of his readers probably figured that he wouldn't be nearly as close as he was. The long delays, many respins, and ever-evolving release dates dictated changes on the fly for nvidia. Also, I would be very surprised if Charlie wasn't deliberately vague and/or misleading in a few instances in order to protect his sources. Remember the AT article about eyefinity? JHH is a genius, he's certainly smart enough to slip slightly different info to many people to find out who is talking.

The one thing I don't get is why the hell would one or more of nvidia's highly paid engineers stab their employer in the back for a guy who obviously hates them?

 

[1h4x4s3x]

Nvidia castrates Fermi to 448SPs 21 December 2009

Tesla != GTX 480
When will people learn <rhetoricalquestionmark>

Nvidia GF100 pulls 280W and is unmanufacturable 17 January 2010

http://tpucdn.com/reviews/NVIDIA/GeForce_GTX_480_Fermi/images/power_peak.gif
Pretty accurate.
Unmanufacturable; has anyone seen a 512 shader chip?

Nvidia's Fermi GTX480 is broken and unfixable 17 February 2010

see above.

SemiAccurate wrong about Nvidia 480GTX power use 12 March 2010

see above.

Keep it up, it's quite entertaining!

 

[Daedalus685]

He tells so many lies, I understand why you all may get confused.

http://www.semiaccurate.com/2010/02/17/nvidias-fermigtx480-broken-and-unfixable/

Yes, that is exactly what I just said...

He did not claim the 480 would be 448... he thought the first part would be. In fact... if the 480 has zero availability for a couple weeks longer than the 470 doesn't that make him right in an absolute sense?


This is such a waste of time... sigh.

 

[Unkle_Tar]

The force is weak with this one.