Nvidia 40nm update.

[SickBeast]

Originally posted by: waffleironhead

Originally posted by: SickBeast

~snip~
BTW, why is the thread being locked?

Most likely to end debate on the issue. Keys forwarded his info, and now before the masses can hate on NVidia for their suspected grass roots media ploys they are going lock it up. Further pushing members to head to other forums.

This thread has pretty much devolved into what is typical around here. The usual cast of characters come into just about any thread and push their not-so-hidden agendas. It is pretty easy to see the leanings of members and wonder where their enthusiasm comes from. Is it truly just rampant brand loyalty or is there more at stake for the poster.

I don't think they'll lock it. There's not a need IMO.

The video forum is doing much better than it has been previously. We'll see what happens.

 

[bryanW1995]

I think that 40nm ramp is extremely important for both camps. AMD has seen first-hand what having a process advantage can do for you since intel has been clobbering them with it for years. Now they get the chance to deal some of that medicine to nvidia. IMHO the process advantage is the only thing that kept them alive and got amd to 48x0. Now that we're moving on to the next gen, they still have the same process advantage and will be cranking out 40nm chips for the retail channel for, let's say, 3-6 mos before nvidia gets around to it. This will serve them in good stead when they launch the dx11 mid-high end chips later this year. Maybe that's the real reason that the rumors have been flying about amd launching a few months before nvidia this time around.

 

[IlllI]

nobody answered my question

 

[taltamir]

it seemed rhetorical, and even if not we are not psychic, so we can't tell.

 

[Idontcare]

Originally posted by: IlllI
seems to me like they are going to start with maybe 1 card them ramp up and eventually replace all of them with 40nm. maybe by the end of the year they'll all be 40nm products? or could it take longer?

That would obviously be the ideal situation, but it depends on the health of TSMC's 40nm as well as the volume available. To ramp volume requires TSMC to spend money on capex, something they are reluctant to do in this economy unless they have high confidence the customer orders are going to "stick" and not evaporate if the economy declines further. It's a risk that TSMC might be approaching slowly and cautiously, which would then push-out or delay the transition for their customers.

Originally posted by: bryanW1995
I think that 40nm ramp is extremely important for both camps. AMD has seen first-hand what having a process advantage can do for you since intel has been clobbering them with it for years. Now they get the chance to deal some of that medicine to nvidia. IMHO the process advantage is the only thing that kept them alive and got amd to 48x0. Now that we're moving on to the next gen, they still have the same process advantage and will be cranking out 40nm chips for the retail channel for, let's say, 3-6 mos before nvidia gets around to it. This will serve them in good stead when they launch the dx11 mid-high end chips later this year. Maybe that's the real reason that the rumors have been flying about amd launching a few months before nvidia this time around.

Agreed, it will be interesting to see if/when AMD migrates their GPU account to GlobalFoundries. Or if NV does the same. There's a world of difference between TSMC's 40nm tech and GlobalFoundries 45nm SOI tech...I expect those differences to be maintained as GlobalFoundries proceeds to release their 32nm Bulk Si versus TSMC's 28nm.

 

[Zstream]

Originally posted by: waffleironhead

Originally posted by: SickBeast

~snip~
BTW, why is the thread being locked?

Most likely to end debate on the issue. Keys forwarded his info, and now before the masses can hate on NVidia for their suspected grass roots media ploys they are going lock it up. Further pushing members to head to other forums.

This thread has pretty much devolved into what is typical around here. The usual cast of characters come into just about any thread and push their not-so-hidden agendas. It is pretty easy to see the leanings of members and wonder where their enthusiasm comes from. Is it truly just rampant brand loyalty or is there more at stake for the poster.

The issue is that some people generally make a wild speculation based off a single review's to support their point. People rush in to put out the argument and then "bam" it is crazy.

People need to not make these generalizations. Just a side-note, if a Nvidia focus group member, marketer, whatever it may be outside of an engineer... you will be flamed. Expecting anything else is just stirring the pot.

 

[Zstream]

Originally posted by: Idontcare

Originally posted by: IlllI
seems to me like they are going to start with maybe 1 card them ramp up and eventually replace all of them with 40nm. maybe by the end of the year they'll all be 40nm products? or could it take longer?

That would obviously be the ideal situation, but it depends on the health of TSMC's 40nm as well as the volume available. To ramp volume requires TSMC to spend money on capex, something they are reluctant to do in this economy unless they have high confidence the customer orders are going to "stick" and not evaporate if the economy declines further. It's a risk that TSMC might be approaching slowly and cautiously, which would then push-out or delay the transition for their customers.

Originally posted by: bryanW1995
I think that 40nm ramp is extremely important for both camps. AMD has seen first-hand what having a process advantage can do for you since intel has been clobbering them with it for years. Now they get the chance to deal some of that medicine to nvidia. IMHO the process advantage is the only thing that kept them alive and got amd to 48x0. Now that we're moving on to the next gen, they still have the same process advantage and will be cranking out 40nm chips for the retail channel for, let's say, 3-6 mos before nvidia gets around to it. This will serve them in good stead when they launch the dx11 mid-high end chips later this year. Maybe that's the real reason that the rumors have been flying about amd launching a few months before nvidia this time around.

Agreed, it will be interesting to see if/when AMD migrates their GPU account to GlobalFoundries. Or if NV does the same. There's a world of difference between TSMC's 40nm tech and GlobalFoundries 45nm SOI tech...I expect those differences to be maintained as GlobalFoundries proceeds to release their 32nm Bulk Si versus TSMC's 28nm.

Why not use GF for the CPU side and TSMC for the GPU side?

 

[SlowSpyder]

Originally posted by: bryanW1995

Originally posted by: Idontcare

Originally posted by: BenSkywalker

So basically it's a PR response and they have nothing right now.

It isn't PR, it is more of a we have been nagging Keys to find out if nV has 40nm rolling yet and he got the answer to the question. nV knows they don't sell a build process, if anyone was going to release a PR statement about nV being at 40nm it would be TSMC, not nVidia.

Agreed. I'm puzzled by the seeming reactive response this otherwise passive information is receiving. Its an FYI, meant to be nothing more and nothing less.

Keys, please don't let the take-home message for yourself or your NV contacts be that of the negative sentiments voiced here. Some of us do like this update info, it beats nothing (or worse, speculation) which is all we would otherwise have were it not for your going to the effort to create this info thread. :thumbsup:

some of our AMD fans seem to think that keys is evil because he's in the focus group. Thanks, rollo...

Well, if you read some of Rollo's posts, or a certain other poster who's posted in this thread you can see why so many people just roll their eyes and do things like I did and jump the gun and post something negitive when it isn't called for regarding Nvidia.

Though I still think what Key's posted more or less says that they have no 40nm part yet, but are trying to put the proper marketing spin on it.

 

[Idontcare]

Originally posted by: Zstream
Why not use GF for the CPU side and TSMC for the GPU side?

For which node? 45/40nm or you talking 32/28nm?

For 45/40nm thats exactly what AMD is doing.

 

[IlllI]

Originally posted by: taltamir
it seemed rhetorical, and even if not we are not psychic, so we can't tell.

well you seemed wrong. maybe Keysplayr has more info

 

[shangshang]

Originally posted by: Idontcare

Originally posted by: Creig

Originally posted by: Idontcare

Originally posted by: BenSkywalker

So basically it's a PR response and they have nothing right now.

It isn't PR, it is more of a we have been nagging Keys to find out if nV has 40nm rolling yet and he got the answer to the question. nV knows they don't sell a build process, if anyone was going to release a PR statement about nV being at 40nm it would be TSMC, not nVidia.

Agreed. I'm puzzled by the seeming reactive response this otherwise passive information is receiving. Its an FYI, meant to be nothing more and nothing less.

Keys, please don't let the take-home message for yourself or your NV contacts be that of the negative sentiments voiced here. Some of us do like this update info, it beats nothing (or worse, speculation) which is all we would otherwise have were it not for your going to the effort to create this info thread. :thumbsup:

Well, you have to admit that Nvidia does like to try and steal the spotlight away from AMD any chance they get.

-On the 40nm 4770 launch, Nvidia announces they have their own transition to 40nm, but currently have no available 40nm product and won't have any for some time to come.

-On the 4890 launch, Nvidia also announces the release of their GTX 275. But it is in limited quantities until they can ramp up production a few weeks later. They could have easily chosen to stockpile cards until they had enough cards for a hard launch but instead chose a soft launch in order to time its release with AMD's 4890 launch date.

-On the eve of the 4850/4870 launch, Nvidia contacts AnandTech and tries to convince the reviewers to put more emphasis on PhysX and CUDA in an attempt to make AMD cards look like a poor choice compared to their Nvidia counterparts. The resulting information contained in the HD 4890 review regarding this attempt causes the entire Nvidia PR department to pop a collective aneurysm.

It's nice to know that Nvidia is hard at work on their own 40nm cards, but given what we've seen in the past, the timing of this announcement doesn't seem like much of a coincidence given that the NDA on the 40nm 4770 was officially lifted this morning.

I am not here to defend Nvidia's business practices, just trying to explain them in the rarte instances where it appears to me that folks are misunderstanding the why's and the how's that go along with the when's and the what's.

As for the business practices you highlight, I am not an NVDA shareholder but if I were a shareholder and NV were NOT doing something to actively stifle/muffle/counter any advantage their competition was generating then I'd be pissed and irritated by what I'd consider an executive management team neglecting their fiduciary responsibilities to me, the shareholder.

I'd say the same of AMD if their management let NV generate an advantage for themselves that went unchallenged or uncontested or unmuffled by AMD.

These two companies aren't dating or courting each other for marital purposes; they are locked in corporate warfare seeking to steal each other's pending sales.

There's nothing noble or honorable about pulling punches in their world unless they become non-profit businesses and grow a pair of .org website domains. Any perceptions of honor or nobility that you have for AMD over NV is either to their credit for effectively duping you into believing so, or to their discredit for not capitalizing on opportunities to their shareholders benefit in the way that their competition does for their shareholders.

It's just business for them.

What a well written post, grammatically, thoughtfully, and knowledgeable, without the fanboy mudsling. I'll bet you've been in the semi business long enough.

 

[Idontcare]

Originally posted by: shangshang
What a well written post, grammatically, thoughtfully, and knowledgeable, without the fanboy mudsling. I'll bet you've been in the semi business long enough.

Thanks for the kind words Not sure if it counts as "long enough" but I cut my teeth on the 0.5um node, to sheepishly date myself, some 15 yrs ago. It's been an abusive two-way relationship ever since :laugh:

 

[Zap]

Originally posted by: Creig
Well, you have to admit that Nvidia does like to try and steal the spotlight away from AMD any chance they get.

Of course. AMD also does the same thing to NVIDIA every chance it gets.

Originally posted by: Creig
-On the 4890 launch, Nvidia also announces the release of their GTX 275. But it is in limited quantities until they can ramp up production a few weeks later. They could have easily chosen to stockpile cards until they had enough cards for a hard launch but instead chose a soft launch in order to time its release with AMD's 4890 launch date.

I'll quote the whole thing since it is so short (link for reference).

Radeon 4770 not really widely available (Fudzilla)

Since Radeon HD 4770 is ATI?s first 40nm card, and you can imagine that the process or yields are far from optimal. Therefore, you can expect some 40nm HD 4770 cards on the market, but it will be a while until you see a better availability.

The first batch was delivered before the launch and the second batch is expected in first week of May, or should we simply say next week.

Even the next batch won?t satisfy the needs of customers and we believe it will take at least a month if not even more before ATI can pump enough 40nm RV740's into the channel. This was the case with the RV770 launch last year, as this 55nm card was a great performer, but very hard to get in the first month following the launch.

So, if NVIDIA launches with limited quantities, they are evil. What does that make ATI when they launch with limited quantities?

With the GTX 275, the launch was only moved up a week. Yes, initial quantities were low. By the following week (basically the original launch date) cards were shipping in mass quanitity. Was it done to mess with ATI's launch? All evidence says "probably."

Then again, when ATI originally released their 4870, they could have priced the cards a lot higher and still have been the price/performance leader. However, they choose to price the cards really, really cheap. Was it done to screw with NVIDIA profit margins on the GTX 260/280? All evidence says "probably."

These companies are in competition with each other for your business. Not much different than supermarkets saying they will accept competitor's store coupons, or office superstores saying they will price match. End result? Consumers win.

 

[Zstream]

Originally posted by: Zap

Originally posted by: Creig
Well, you have to admit that Nvidia does like to try and steal the spotlight away from AMD any chance they get.

Of course. AMD also does the same thing to NVIDIA every chance it gets.

Originally posted by: Creig
-On the 4890 launch, Nvidia also announces the release of their GTX 275. But it is in limited quantities until they can ramp up production a few weeks later. They could have easily chosen to stockpile cards until they had enough cards for a hard launch but instead chose a soft launch in order to time its release with AMD's 4890 launch date.

I'll quote the whole thing since it is so short (link for reference).

Radeon 4770 not really widely available (Fudzilla)

Since Radeon HD 4770 is ATI?s first 40nm card, and you can imagine that the process or yields are far from optimal. Therefore, you can expect some 40nm HD 4770 cards on the market, but it will be a while until you see a better availability.

The first batch was delivered before the launch and the second batch is expected in first week of May, or should we simply say next week.

Even the next batch won?t satisfy the needs of customers and we believe it will take at least a month if not even more before ATI can pump enough 40nm RV740's into the channel. This was the case with the RV770 launch last year, as this 55nm card was a great performer, but very hard to get in the first month following the launch.

So, if NVIDIA launches with limited quantities, they are evil. What does that make ATI when they launch with limited quantities?

With the GTX 275, the launch was only moved up a week. Yes, initial quantities were low. By the following week (basically the original launch date) cards were shipping in mass quanitity. Was it done to mess with ATI's launch? All evidence says "probably."

Then again, when ATI originally released their 4870, they could have priced the cards a lot higher and still have been the price/performance leader. However, they choose to price the cards really, really cheap. Was it done to screw with NVIDIA profit margins on the GTX 260/280? All evidence says "probably."

These companies are in competition with each other for your business. Not much different than supermarkets saying they will accept competitor's store coupons, or office superstores saying they will price match. End result? Consumers win.

That is a BS article with no proof to back its claims. It says "you can imagine"

 

[Idontcare]

Originally posted by: Zstream
That is a BS article with no proof to back its claims. It says "you can imagine"

Zstream there are those of us here, plenty of us actually, who have access to the data regarding TSMC's 40nm health of the line and while we cannot put the exact numbers into the public domain we can link to every rumor site that van assist us in our efforts to give folks who want a clue, a clue.

For folks who simply want to refuse to accept or understand, well we aren't going to email you the pdf documents and other HOL materials we have just so we can prove to you what others are willing to say "you can imagine". If you want to believe TSMC's 40nm yields are in the high nineties and NV's delays are solely due to NV's own internal tape-out timelines then go for it.

In the meantime there will be plenty of us around here still trying to quietly, read between the lines type info, give some insight to the rest of the crowd who are interested in these things. And sometimes that means we link to FUDzilla or TheINQ when we actually know the "rumor" they are writing about is true.

You can cry BS every time if you like, the fact we aren't in a legal position to disprove your BS claims doesn't actually validate your position on the topic.

 

[Creig]

Originally posted by: Zap

Originally posted by: Creig
Well, you have to admit that Nvidia does like to try and steal the spotlight away from AMD any chance they get.

Of course. AMD also does the same thing to NVIDIA every chance it gets.

Which AMD products do you feel have had their release dates shifted or announced for the sole purpose of competing with an Nvidia launch or product announcement? I can't think of any off the top of my head. I'm not saying they don't exist, but if they happened I somehow missed it and would appreciate knowing. I try to keep an open mind and evaluate all the factors.

Originally posted by: Zap

Originally posted by: Creig
-On the 4890 launch, Nvidia also announces the release of their GTX 275. But it is in limited quantities until they can ramp up production a few weeks later. They could have easily chosen to stockpile cards until they had enough cards for a hard launch but instead chose a soft launch in order to time its release with AMD's 4890 launch date.

I'll quote the whole thing since it is so short (link for reference).

Radeon 4770 not really widely available (Fudzilla)

Since Radeon HD 4770 is ATI?s first 40nm card, and you can imagine that the process or yields are far from optimal. Therefore, you can expect some 40nm HD 4770 cards on the market, but it will be a while until you see a better availability.

The first batch was delivered before the launch and the second batch is expected in first week of May, or should we simply say next week.

Even the next batch won?t satisfy the needs of customers and we believe it will take at least a month if not even more before ATI can pump enough 40nm RV740's into the channel. This was the case with the RV770 launch last year, as this 55nm card was a great performer, but very hard to get in the first month following the launch.

So, if NVIDIA launches with limited quantities, they are evil. What does that make ATI when they launch with limited quantities?

First of all, of course Nvidia is not evil. I'm not a deranged fanatic like some who cough up and post every single little pro-Nvidia/anti-AMD headline they can find. I simply don't like their business practices.

The difference between the 4770 soft launch and the GTX 275 soft launch is that the 4770 launch date was not shifted around simply to coincide with an Nvidia launch.

Originally posted by: Zap
With the GTX 275, the launch was only moved up a week. Yes, initial quantities were low. By the following week (basically the original launch date) cards were shipping in mass quanitity. Was it done to mess with ATI's launch? All evidence says "probably."

And that was my point. Nvidia's decision to launch early was made in their PR department. AMD's decision to launch their 4890 early was made simply because they had enough cards on hand to handle a hard launch.

Wasn't Nvidia proud (and rightfully so) of their 8800GT hard launch? They made a lot of noise about how it well it came off and that there was enough supply to handle the demand. Now we have Nvidia moving the launch date of a card already in limited quantities, to an even earlier date simply to "mess with ATI's launch".

NVIDIA moves to scupper AMD graphics launch

AMD had planned to launch the ATI Radeon HD 4890 GPU on 9 April. Then it apparently decided things were sufficiently advanced to bring the launch date forward by a week, notifying HEXUS of this change by Wednesday of last week.

Meanwhile NVIDIA had originally planned to launch its GeForce GTX 275, which is targeted at pretty much the identical segment of the market as the HD 4890, on 13 April. Then, last Friday, it unexpectedly moved the date to today too. Spooky coincidence eh?

If you have a product that is already predicted to be in short supply, what possible logical reason would you have to launch it even earlier than originally planned other than as a purely PR move?

Originally posted by: Zap
Then again, when ATI originally released their 4870, they could have priced the cards a lot higher and still have been the price/performance leader. However, they choose to price the cards really, really cheap. Was it done to screw with NVIDIA profit margins on the GTX 260/280? All evidence says "probably."

The difference is that the GTX 260 and GTX 280 were artificially priced high, simply due to lack of competition. Also, the 4870/4850 were much cheaper to produce than the GTX 260/280. Still are, in fact. The profits realized from a higher volume of lower priced cards can actually be greater than if the card had been priced higher but sold in fewer quantity.

Can you honestly say that the price of the 4870/4850 was set "solely" to mess with Nvidia as they did with moving their launch date forward? Or were there other factors involved that contributed to its lower level?

Other than for PR decisions, what possible reason was there for Nvidia to move their GTX 275 soft launch to an even earlier date? Or to leak their 40nm GPU plans on the exact day of the AMD 40nm launch? Or to keep pressuring AnandTech reviewers to include the lack of PhysX and CUDA capability when reviewing AMD video cards?

Nvidia's PR department has been getting overaggressive and some people like myself (and obviously Derek and Anand) simply don't like their tactics.

Originally posted by: Zap
These companies are in competition with each other for your business. Not much different than supermarkets saying they will accept competitor's store coupons, or office superstores saying they will price match. End result? Consumers win.

Agreed. Ironically, even though we're in the middle of a worldwide recession, these have been some of the best times for hardware enthusiasts to upgrade.

 

[taltamir]

are you sure the 4870 is cheaper to produce? sure it has a smaller chip, but had to deal with more expensive GDDR5. Since then GDDR5 tech cost has gone down somewhat, but nvidia also shrunk their process tech half a node.

 

[Creig]

Originally posted by: taltamir
are you sure the 4870 is cheaper to produce? sure it has a smaller chip, but had to deal with more expensive GDDR5. Since then GDDR5 tech cost has gone down somewhat, but nvidia also shrunk their process tech half a node.

No, I don't have definitive proof. As Idontcare said above, all we can do is read as much information as is available and come to a conclusion. If new evidence turns up, include it and adjust your viewpoint accordingly.

From everything I've read, from die size to yields to PCB complexity, the 4800 series of card is cheaper to produce than the GTX200 series, at least when it was still a 65nm product. I haven't heard what their yields are like on the 55nm node, but the GT200 is still a 1.4 billion transistor GPU while the 4870 is only 956 million. Even if Nvidia somehow managed to get identical yields on the GT200 as the 4870 now that the GT200 is also being produced at 55nm, the 4870 would still be cheaper simply due being able to fit more cores on a single wafer.

edit - I edited my post for additional clarity while Idontcare was posting his.

 

[Idontcare]

Originally posted by: Creig

Originally posted by: taltamir
are you sure the 4870 is cheaper to produce? sure it has a smaller chip, but had to deal with more expensive GDDR5. Since then GDDR5 tech cost has gone down somewhat, but nvidia also shrunk their process tech half a node.

No, I don't have definitive proof. As Idontcare said above, all we can do is read as much information as is available and come to a conclusion. If new evidence turns up, include it and adjust your viewpoint accordingly.

From everything I've read, from die size to yields to memory cost to PCB complexity, the 4800 series of card is cheaper to produce than the GTX200 series.

What tools do folks have at their disposal for determining/guestimating the cost-envelope of GDDR5 vs. GDDR3 (including PCB build-out)?

For commodity Dram, Flash, SSD's, and memory cards we have DRAMeXchange to give us some indication of the underlying minimum cost-structure for products based on these commodity chips...but I have yet to find a public source documenting GDDR prices.

Professional services exist with this info on a paid-for-use basis, but for forum exchanges it helps to have freely viewable links to substantiate some claims regarding cost-delta's. Is there one?

 

[Creig]

Originally posted by: Idontcare
What tools do folks have at their disposal for determining/guestimating the cost-envelope of GDDR5 vs. GDDR3 (including PCB build-out)?

For commodity Dram, Flash, SSD's, and memory cards we have DRAMeXchange to give us some indication of the underlying minimum cost-structure for products based on these commodity chips...but I have yet to find a public source documenting GDDR prices.

Professional services exist with this info on a paid-for-use basis, but for forum exchanges it helps to have freely viewable links to substantiate some claims regarding cost-delta's. Is there one?

I've looked, but was unable to find a specific instance mentioning pricing models of GDDR3 vs GDDR5. But I would be highly surprised if the price difference between the two was enough to overcome the GPU disparity in size and complexity between the GT200 and 4870. AFAIK, the GPUs cost in engineering, development and production makes it, by far, the most expensive component in a modern day video card.

 

[SlowSpyder]

Originally posted by: taltamir
are you sure the 4870 is cheaper to produce? sure it has a smaller chip, but had to deal with more expensive GDDR5. Since then GDDR5 tech cost has gone down somewhat, but nvidia also shrunk their process tech half a node.

We obviously don't have hard numbers. But I think that because AMD chose to create the 4770 with a 128 bit memory configuration with GDDR5 to get their bandwidth goal vs. going with a 256 bit configuration with GDDR3 is somewhat telling. They have a lot of cards that use GDDR3, and have used it often in the past. I think it says something about the costs that they went with the GDDR5/128 bit configuration over the GDDR3/256 bit configuration.

Originally posted by: Idontcare

<snip> - Reply to Zstream.

IDC, you know I completely respect your opinion and know you have info on 'the industry' that isn't readily available to lowly hobbiests like myself. But the 4770 is available to buy right now. I'm not saying there isn't tight quanities, and TSMC's 40nm is doing great with yields (it's brand new tech, I think it's expected to have some growing pains in the beginning, isn't it?) but it is available to buy right now if someone wanted one. There is a big difference between this launch and say the 9800GTX+ launch. :beer:

 

[Idontcare]

Originally posted by: Creig
AFAIK, the GPUs cost in engineering, development and production makes it, by far, the most expensive component in a modern day video card.

This is, without a doubt and by a large margin too boot, the truth.

Originally posted by: SlowSpyder
IDC, you know I completely respect your opinion and know you have info on 'the industry' that isn't readily available to lowly hobbiests like myself. But the 4770 is available to buy right now. I'm not saying there isn't tight quanities, and TSMC's 40nm is doing great with yields (it's brand new tech, I think it's expected to have some growing pains in the beginning, isn't it?) but it is available to buy right now if someone wanted one. There is a big difference between this launch and say the 9800GTX+ launch. :beer:

The respect is mutual Steve, very much so, and if you are reading my response to Zstream and read Zstream's post and from your perspective as a third-party it appears to you that I misinterpreted Zstream's post (and as such my post was effectively misplaced/needless) then I take that feedback very seriously and I appreciate it.

I don't disagree that the 4770 is available now and ready for purchase, but at the same time it is true that 40nm wafer volume is low at this time and yields are not stellar which means there is a reasonably limited supply of RV740 chips. To call that BS is simply to be ignorant of the mechanics of the business on that side of the business, something I was attempting to clarify in my response to Zstream's post.

Now provided demand doesn't outstrip the supply, the yields and wafer-volume (capacity in industry terms) situation at TSMC is irrelevant to the consumer. But from what I know of the situation, and when I read the info in that FUD, there is a viable supply constraint issue looming in the background at the moment should demand for 4770's pickup in any appreciable manner. The 4830's will be there to soak up some of the demand should that happen I imagine.

 

[imported_Scoop]

What Nvidia is going 40nm?! No shit?! I thought 55nm was going to last forever.

Yeah, that's how useful this info was. No dates, no products, nothing. Just Captain Obvious.

 

[Kuzi]

Originally posted by: Wreckage
The GTS250 generally outperforms a 4850 and is closer to a 4870-512.

Actually in the xBit Labs review out of 15 games tested, the the GTS250 was faster than the 4770 in only 4 or 5 games. The 4770 beats or equals it in the other games while also being cheaper and cooler.

The 4850 was on average 7% faster than the 4770. So I don't see how it can be slower than the GTS250.

 

[Idontcare]

Originally posted by: Scoop
What Nvidia is going 40nm?! No shit?! I thought 55nm was going to last forever.

Yeah, that's how useful this info was. No dates, no products, nothing. Just Captain Obvious.

The information has value in that the response was not "40nm? Uhm, yeah I guess, maybe 2H 2010 timeframe? Or possibly later, mmm, definitely probably most certainly not earlier. Absolutely. Maybe."

Although such a response would tell us a great deal as well, had that been the response.

Leading edge information is like putting together a puzzle. If you aren't willing to put the jigsaw pieces together, over some time, and stand back and look at what the pieces are communicating in toto then you are going to pass up a lot of useful information.

Although to be honest if you are that type of person then having a pre-assembled jigsaw puzzle handed to you is probably of limited value as well because you aren't likely in a position in life to do much with the information to begin with. (that's not a knock on you or anyone else like that, just saying not everyone is impacted by information regarding NV's 40nm release timeline, so what do they care if they are told the precise hour and date of the release? even if they were told it would hold no value to them)