PDA

View Full Version : Nvidia may be in trouble according to this article


Pages : [1] 2 3 4 5 6 7 8

Madpistol
02-18-10, 06:50 PM
I have no clue how accurate this is, but....

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


Discuss.

LydianKnight
02-18-10, 07:25 PM
Sounds like kind of a counter-measure against this: Geforce GTX 480 to be as hot as GTX 285 (http://www.fudzilla.com/content/view/17723/1/)

Maybe... maybe not. While I don't appreciate Charlie Demerjian's writing style, I have to admit he has been somewhat accurate about what he told about the need of an A3 revision and some other things... so I don't know what to say...

For me, it's the same as I said... I just expect Fermi to be a refresh of my current graphics architecture, I would consider myself more than satisfied with 'just' 512 cores even if they end being lowered in clocks.

Heck... my current GTX260 is clocked at 576MHz so in naive terms, I have 216 cores @ 576, as long as I can get my hands on the 512 core version, I wouldn't mind even having them at the same clocks (even 550 would do ok for me, I don't pretend to have an ATI-crushing machine, competition does nothing for me, I just want a home media+gaming machine).

What keeps me intrigued is the TDP... That would be the only real showstopper for me... more than 250W and... no, thanks.

Other than that... we'll have to wait and see :P

Just my 2 cents.

Julio

Madpistol
02-18-10, 07:34 PM
It's pretty obvious that this guys is an ATI fanboy. I just don't know how accurate it is.

If it's even remotely accurate, though, Fermi will not be the showstopper it's been touted as since September of last year. The 5870 and 5970 will still be the reigning champs of the Video Card industry.

Johnny C
02-18-10, 07:54 PM
It's Charlie....so if we take out all the cynicism and hatred of nV.....still looks like it might be a flop.

Worst thing is....that Charlie has been right about alot of key Fermi info thus far....

Johnny C
02-18-10, 07:56 PM
It's pretty obvious that this guys is an ATI fanboy. I just don't know how accurate it is.

If it's even remotely accurate, though, Fermi will not be the showstopper it's been touted as since September of last year. The 5870 and 5970 will still be the reigning champs of the Video Card industry.

He isn't a fanboy of ATi....he just really really hates nV....which makes it seem like he's an AMDTi fanboy....

XxDeadlyxX
02-18-10, 08:12 PM
Well there's a lot of in-depth detail there, so a lot of it could be true unfortunately :(

JasonPC
02-18-10, 08:19 PM
All the nvidia PR guys are responding by saying Charlie is going to be eating his words come March. But we'll see. Are they right or are they PR guys being PR guys.

shadow001
02-18-10, 08:52 PM
All the nvidia PR guys are responding by saying Charlie is going to be eating his words come March. But we'll see. Are they right or are they PR guys being PR guys.


And you'd expect them to say anything else?,since they are from Nvidia's marketing dept afterall and besides,even Nvidia's own CEO was caught in a lie when he stated Fermi might be available in late november 2009,most likely to keep potential customers from buying ATI cards this round.


He knew there was no chance in hell of fermi getting released that early in any sort of volume,yet he still said it anyhow,and had to personally retract on that statement later on,and that's the example set from the leader on Nvidia itself,so what do you expect the PR guys from the same company are going to do?....;)


Fermi is so horribly late at this point that refresh cards based on the GF100 GPU should be out not much after this first release anyhow....Kinda like what happened between the NV30,also being horribly late like Fermi,and the NV35 which was released about 3 months later.

Johnny C
02-18-10, 09:36 PM
Well there's a lot of in-depth detail there, so a lot of it could be true unfortunately :(

He's also been right about alot of Fermi stuff so far.....

Sazar
02-18-10, 09:59 PM
saying the GPU will be out by X date is a lie? you have proof he knew it was going to be delayed when he said that? Proof? When I wrote software sometimes I would release it when I said I would and sometimes I got it out after my estimated release date. It happens. Does not mean I lied.

Darth, in a manner of speaking, yes.

Much like the NV30 setup, the whole thing of "wait till this date, do NOT buy the other product, ours will be out by x date and will be faster and better".

It's not marketing, that's just lieing to prevent someone from buying a superior product that is out on the market now. Whether or not Fermi is faster whenever it is released is moot because their competitor has had a faster product out now for 6 months or so and the only response has been to lie to potential buyers to prevent them from going out and getting something else.

In this case, there is no indication anywhere that Nvidia even tried to get their card released by November of last year so, unfortunately, that would classify what he said as being a lie.

shadow001
02-18-10, 09:59 PM
saying the GPU will be out by X date is a lie? you have proof he knew it was going to be delayed when he said that? Proof? When I wrote software sometimes I would release it when I said I would and sometimes I got it out after my estimated release date. It happens. Does not mean I lied.


Given that Charlie said several months ago that for the november date to be even possible at all,they'd have to get all the features,clocks and yeilds for Fermi working the way they wanted to from the very first revision of the Chip,which is something that pretty much never happends on the first revision of anything,never mind one with 3 billion transistors on a brand new 40nm fabrication process.


Then there was the little fact that when he made that statement in late september,as some of the GP-GPU related features of fermi were officially revealed one week after ATI unveiled the HD5870 cards,they didn't even have the first revision of Fermi back from TSMC yet,so how can one make availability predictions on something they don't even have working silicon of as of late september?.


Remember the hastily made card that Nvidia's CEO was proudly holding and showing to the crowd,stating that it was a fermi based,when it clearly wasn't by the sawed off PCB and wood screws holding the heatsink in place ....Nvidia's CEO lied about that plain and simple too....He's basically a top flight bull****ter of the worst degree to be a willing part of a stunt like that.

Johnny C
02-18-10, 10:17 PM
<object width="560" height="340"><param name="movie" value="http://www.youtube.com/v/mJOv3VlkEjQ&hl=en_US&fs=1&"></param><param name="allowFullScreen" value="true"></param><param name="allowscriptaccess" value="always"></param><embed src="http://www.youtube.com/v/mJOv3VlkEjQ&hl=en_US&fs=1&" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" width="560" height="340"></embed></object>

@ 2:31...."This puppy...here.....is Fermi"

hehe....it was nothing....

Sazar
02-18-10, 10:25 PM
AMD and Nvidia will always go back and forth. I have been using both companies GPUS since Nvidia game out with its Geforce 256 and before then even with my
3dfx setup. your mistake is your myopic perspective. stand back and see the trees. stating and missing a release date does not equate to a lie automatically

I have been using products from both companies for a long time as well.

But what he did was lie.

Missing deadlines due to various delays is perfectly acceptable.

Lying to prevent people from buying a competitors product is a different story.

I completely understand how shipping schedules, parts availability and what not work in the industry darth :) I also understand how reality distortion fields work.

Sazar
02-18-10, 10:40 PM
again, how can you prove you KNOW that he said something he KNEW was not true? You cannot. the proof is on you since you make the charge. I am not saying he did not lie as I have no way to prove that either but then I am not making charges against him.

Well a couple of things.

First: there is historical precedent. He made the same kind of statements around the time of the NV30 launch, which was only delayed for like 6 or 7 months or so. It also was not ready for launch and yet we were fed information about how it would be out by a certain date and how it would be so amazing and so fast. And yet, we find out afterwards, that was a concerted effort to feed b.s. while preventing people from buying other cards. I ended up buying a 9700pro and was more than satisfied for a long, long time.

Second: there is nothing to indicate that Nvidia has had any kind of shipping schedule or parts availability that would have prompted a November launch or even talk of one. Since the date has already passed, I can say that we never had any test units in our labs, like we typically do and I didn't see any information regarding sku's, or anything really, either from our contacts or from the grapevine online (which is usually VERY accurate) about this kind of thing. Outside of the people at semi-accurate who seem more sensationalist than anything else.

I am in a position where I affect sales. If I go out and tell people I am working directly with that I have a shipping date while knowing full well that I have absolutely no way to actually facilitate delivery of the product, that is lying. In the case of Jen here, I think you can't go wrong calling it a lie.

shadow001
02-18-10, 11:17 PM
they can make predictions based on their previous experience I would guess. Both AMD and Nvidia have been having production issues (granted at different stages). I am sorry but you have to be able to prove intent to say this person or that person is lying. Intel said Larrabee would be out by last October . . . were they lying? No. That was their best guess when they made the statement.


And that's the first mistake right there,since you never base availability on past experience with previous products,since previous products where using no where near the same transistor budget or fabrication process that fermi is,as a lot of things can go wrong between theory and actual product availability,you either have it or you don't,it's really that simple.


What you don't do is anounce availability at a given timeframe for a product that doesn't exist yet in actual silicon form,and having any sort of volume that would allow for a launch in november,means producing thousands of final Fermi chips beforehand,supplying those chips to Nvidia's usual partners to build the actual cards,getting drivers finalised for launch and all the marketing campaign that goes with it,so it's not a process that's done overnight,but takes a couple of months at least.


Basically when ATI had their launch on the 23rd of september,the final version of fermi would already need to be in high volume production by then to even remotely allow a launch in late november with decent availability of cards to actually sell,not just a paper launch,and that was not even close to the case and Nvidia's CEO knew it.




as for Charlie: I don't give a rats arse about him. I can make a call and get right to the source. Does my source know EVERYTHING? no, of course not. But he does know more than some pinhead with an axe to grind (my bud used to be a hardware reviewer before working for Nvidia).


If you can,please do as people have had enough with the delays,performance speculations,power consumption and the actual launch date,which has still yet to be anounced officially BTW.

shadow001
02-18-10, 11:18 PM
From what I have been told, I was non-Nvidia person to actually handle the Fermi GPU (when I visited Nvidia in November of last year). a little over a month after that video.


Was it actually in a PC and running a game or benchmark?

shadow001
02-18-10, 11:27 PM
Well,it looks like the launch of fermi in march or april will largely be a paper launch afterall,since availability will be limited,with larger volumes only available on the second quarter of this year,and this is confirmed by an Nvidia executive,not just another rumor floating around.


http://vr-zone.com/forums/561413/nvidia-quot-fermi-quot-gpus-delayed-yet-again.html


Real availability of GTX 470 and 480 cards only starts in may at the earliest it seems,and they even state that more budget oriented DX11 class GPU's aren't important since users buying cards in those prices ranges don't demand the latest functionality and that Nvidia's current DX10,DX10.1 cards are fabulous for that segment,so that leaves ATI with their HD5500/5600/5700 cards with the mainstream/budget DX11 market all to themselves basically.

Sazar
02-18-10, 11:54 PM
you have not proven anything. I have been talking with them about this product and release dates for quite some time and have a totally different impression. You cannot prove why he said what he said and what the basis for it was as you do not have access to that information.

I guess we'll have to agree to disagree :)

In my opinion, he lied to save sales for his company and reduce sales to the competition while knowing there could not be a product delivered to the market by the stated time-frame. They have done this before and it appears they are doing it again.

In your opinion, it was a delay as a result of something else.

ir123
02-19-10, 12:47 AM
Darth, in a manner of speaking, yes.

Much like the NV30 setup, the whole thing of "wait till this date, do NOT buy the other product, ours will be out by x date and will be faster and better".

It's not marketing, that's just lieing to prevent someone from buying a superior product that is out on the market now. Whether or not Fermi is faster whenever it is released is moot because their competitor has had a faster product out now for 6 months or so and the only response has been to lie to potential buyers to prevent them from going out and getting something else.

In this case, there is no indication anywhere that Nvidia even tried to get their card released by November of last year so, unfortunately, that would classify what he said as being a lie.

Worked for MikeRowesoft.

Madpistol
02-19-10, 08:12 AM
Well,it looks like the launch of fermi in march or april will largely be a paper launch afterall,since availability will be limited,with larger volumes only available on the second quarter of this year,and this is confirmed by an Nvidia executive,not just another rumor floating around.


http://vr-zone.com/forums/561413/nvidia-quot-fermi-quot-gpus-delayed-yet-again.html


Real availability of GTX 470 and 480 cards only starts in may at the earliest it seems,and they even state that more budget oriented DX11 class GPU's aren't important since users buying cards in those prices ranges don't demand the latest functionality and that Nvidia's current DX10,DX10.1 cards are fabulous for that segment,so that leaves ATI with their HD5500/5600/5700 cards with the mainstream/budget DX11 market all to themselves basically.


It looks like ATI has won this generation by default.

I bought an 8800 GTX, and now I've bought a 5870. Both are epic WIN! :D

Johnny C
02-19-10, 09:34 AM
It looks like ATI has won this generation by default.

I bought an 8800 GTX, and now I've bought a 5870. Both are epic WIN! :D

Agreed,

8800GTX was probably the best card for performance/longevity I've ever bought.(except the release drivers....I remember waiting for game support for the first couple months)

4870 1gb was a good card...and still is...but doesn't match the GTX for it's increase in performance...

5850 right now is by far the best buy killer performance smallish price....can't go wrong.

Back on topic.

nV is pooched this round....we all know it. While I respect Darth's opinion....it is obvious due to his current affiliation/employment with nV he has access to info that's NDA'd that we do not. I don't think that I'd go out of my way to support a CEO who is blatantly full of ****....the only can of whoop ass he's opening up is on himself....

Everything else aside....I can't wait for this damn thing to get released so I can either go x-fire for cheap....or get a refresh card.

Johnny C
02-19-10, 10:08 AM
uh, but I am not defending his actions I am just saying making extreme accusations such as outright lying have not been proven. I have not said there is not marketing going on ;) You also notice I have not and will not bash AMD's stuff.

That's why I respect your opinion....


But the fact is: "this puppy here....is woodscrews and superglue" and he knew it....just watch the big nervous swallow he takes after saying that....

shadow001
02-19-10, 12:26 PM
Dude give it a rest already... Stop trying to pretend everything is/was perfectly fine with ATI. Your beloved 5000 series was available only in very limited quantities until a few weeks after the official release.


Actually,ATI had already 300 000 cards ready for the official launch between the HD5700/5800 series when it was first launched,but given how well they performed and that rumors regading fermi delays were already floating around the web,they sold out that initial lot in record time,as a lot of people simply decided not to wait for Fermi if it's that late to the market....Patience has limits for everything.



Now this latest article just rubs it in even further with the likelyhood that getting your hands on one of those cards from the initial batch,if it really is under fermi 10 000 cards worldwide,will make getting an HD5800 look easy by comparison,with higher volumes only really being available starting from may onwards.



As hard as it is for some to admit,this first generation of Fermi is over and done with given all the delays,much like what happened with ATI and the original R600 which was also delayed 6+ months compared to the G80 based cards,and where it was replaced very quickly with the RV670 based cards about 3 months later,since at least those were cheaper to build and used less power than the original R600.


Basically,look for a second generation of the GF100 GPU to show up soon after this one,this time taking all the lessons learned with the first version in terms of yeilds,power use and actually hitting the clocks that were intended for this first version.....ATI is coming out with with refreshed versions of their HD5*** lineup and they've also got a new architecture popping up with the northern islands series(HD 6*** line),later this year,so Nvidia have to look beyond the original GF100 at this point.

shadow001
02-19-10, 12:29 PM
Did he say it was not a prototype? Did he say it was a functioning unit? Like I said, I HELD prototype units in my hands a month after that video (at least that is my guess according to the date on youtube) and it did not 'appear' to be superglued and all. I also got to learn about alot of cool things, one being the multi-monitor 3d ;) Honestly I was hoping Intel would have their Larrabee out (I even applied for a marketing job for that project a year or two ago - back when it was a go).

The funny thing about all this is that I do not see why people get so damned worked up. It is just hardware. There are better things to get pissed about. This is how I see it and I am more involved in this industry than most if not all of you. Perspective.

if you want to bitch about a crooked CEO . . .or former . . .
http://www.engadget.com/2009/11/02/ex-amd-ceo-hector-ruiz-steps-down-as-globalfoundries-chairman-am/

http://images.trustedreviews.com/images/article/inline/8194-amd_hector_ruiz.jpg


A prototype is still a working sample no matter what,even if it isn't operating at whatever the final specifications that the retail cards would have been working at....That card wasn't working in any shape or form and didn't even have a Fermi GPU in it to begin with,and Nvidia's CEO never said it was a mockup of what fermi will look like....He simply stated this was fermi right here,which is a big difference.

shadow001
02-19-10, 01:04 PM
how do you know what was inside it? you don't


The pictures are available that showed the PCB being sawed off,as the PCI-e power connectors were cut in half,as well as the woodscrews holding down the cooler,so it wasn't a working PCB in the first place,so if it wasn't working,how could you test the fermi GPU inside of it?....Why would you even waste a supposedly working Fermi GPU,by installing it on a non functional PCB?....logic dictates it wasn't a Fermi GPU in there in the first place.