Poor Nvidia

Im thinking its more that the sandwich design is eol'd permanately. Theyll be using the newer GDDR5 in their next gen, so it wouldnt really make sense to use 2pcbs anymore, besides, theyre cutting costs every way they can, and a 2 pcb solution is more costly

I love how they extrapolate all kinds of FUD (which makes sense considering the source) from the fact Nvidia is fine tuning a product.

How is this poor Nvidia? They are (finally) coming out with a single PCB GTX295. This is a good thing for all involved. Should use less power and be cheaper to make.

The only part that sucks is they aren't introducing anything new/faster. If the TMSC 40nm process has the issues that it has, I don't see them making a 40nm chip any time soon, if at all. I really get the feeling that GTX295 is it, until they come out with their DX11 part. After all, why make a faster DX10 part, if it will be passed by the DX11 part only a few short months later?

My other question is if they NEVER developed a DX10.1 card, will their DX11 card be worth while? MS came up with the DX10 original spec, then dumbed it down after Intel and Nvidia whined. Neither company has built the things required for DX10.1 (AKA "real" DX10) while AMD has been toying with it since they built the Xenos so many years ago. I almost feel that Nvidia is going to fall far behind once we move to DX11, unless MS dumbs it down for them again. Time will tell.

For the sake of balance, I don't see AMD coming out with anything either. They will release the 4890, but that is possibly all we will see from them for DX10/10.1. I do believe we will see DX11 cards from them first, its a smaller step from 10.1 then from 10.

By the looks of it, we should be seeing ATIs first 40nm chip this year .

What sort of alteration? The X2 is already a single pcb design. I think what you will see is a 4770X2 that will only use less than 160w but be as fast as the 4870X2.

zzf is right, the 770 is right around the corner, and it makes sense as well, going to 40nm with R700 and getting 75% power usage and possibly a slight dip in perf on the cheap. The gate widths were too small if I recall, going to 40nm, http://www.altera.com/literature/w [...] evices.pdf check fig. 2. Its what were facing in all things silicon, up the core speeds, heighten the leakage. Take measures to prevent the leakage, it slows the core speeds.
Anyways, 40nm will be here soon, and well see how high those clocks go. ATI releasing a mid level chip shows there is problems with the 40nm IMHO, but with LRB around the corner, neither nVidia nor ATI is going to be sitting on their hands

Sorta. Theyre the "same". Ive heard the 4770 is not only downsized, but has fewer trannys as well. Plus the narrower bus. Maybe what youre thinking about is a R700 refresh across the whole line, which has been a rumored release? Q3 I think it is, while R800 isnt due til Q4, but to me that makes no sense, tho theres been rumors.
IMHO, ATI needs to get out their mobile, and the 4770 is the result of that

They are clearly leading Nvidia in the power area, thats for sure. I still cant believe we havnt seen an 8800GT without a plug.

Isnt there already a mobile 4870?

Yea, some of those laptops are pretty stupid. If your laptop needs to be pluged in to game for more than 10 minutes than it completely eliminates the whole purpose of owning a laptop.

Or they could do a hybrid, making it portable for work and basic use by using integrated, and plugged in for heavy gaming. I mean you could game in power saving mode, but it would need to be older games Since the VIdeo card down clocks.

If im not mistaken, the 4770 will be replacing the 4850, and the 4750 will be replacing the 4830......is that right?

The 4870X2 was on 1 PCB from the beginning



They have come out with the 4860 (its 40nm ), so maybe that will help all of you laptop gamers.

Ive seen the 4860 pop up on a few charts, but ive yet to actually see it in stock anywhere. ATI is getting just as bad as Nvidia when it comes to these card names. I mean is there really a need to wedge a card between the 4850 and 4870? No, there isnt.

ATI has to why?

they been fine with the 4850 and 4870 so far, why are you defending their wrong move?

Its been a plus for them that they haven't followed Nvidia, every1 agreed that its what made them look innovative in the public's eye. Now they do it, and they are still angels.

You can't close your eyes, and be blind to the fact that ATi is doing a wrong move as well with that card.

It really sounds stupid when you blame a company for what the OTHER company is doing.

Well Intels doing it, so that means I have 2 do it.

It works out worse for the 2nd company that did it, when Nvidia did it, every1 was pissed and I agreed 2, but seeing ATI do makes me think they know it pisses us off, they just don't care.

So stop blaming the green cuz u like red, and vice versa.

Who makes more money on gpu's? Nvidia or ATI?

Who makes the best gpu's on p/p?

How do you beat your opponent when they try to confuse the market by releasing so many cards, half of which are rebadged?

ATI needs to beat Nvidia at their own game. Nvidia will pull out every dirty trick they know in order to stay in the game, and ATI are out to finish them in the same way that Nvidia have been trying to finish ATI for the past 3-4 years.

ATI have the bottom-mid range completely sown up, just like Nvidia did 1 year ago.

The difference is, ATI won't blow $millions on stuff that is an attempt to put their competitor out of business. That's what Nvidia did - that's what physx and cuda is all about. ATI just need the money to continue releasing great cards, and by doing that they will put Nvidia out of business anyway because Nvidia are self destructing and only need a competitor that doesn't do the same.

What you say there is completely true btw. ATI will have the bottom-mid range sown up entirely for the next year because Nvidia only have the huge 200 die and there is no way that will ever scale downward like the g92 did. That's why they are going for the same approach as ATI with the 40nm chips. The day of the monolothic chip is over and Nvidia are basically starting from scratch. For their sake, it better be a small miracle.

nVidia is taking a calculated risk, and maybe their only alternative by having physx and CUDA, as its a growing market, I believe, and LRB and Intel will bite into margins/marketshare. Chipsets will be hinderered, gpu sales as well, by things such as SoCs. As Ape would put it, is CUDA and physx a deal breaker? Not any more than DX10.1, but its also nice to have. People arent buying nVidia simply for CUDA or physx, and wont, not the vast majority.

G200b is almost 500mm² While the R700 is under 275 mm² , so it doesnt rocket surgeon to figure out whos bleeding profits here.

I (slightly) disagree with this. Remember that when you come out with a new high end card that if you also move to a new manufacturing process that introduces two new things. If there is something wrong, you don't know if its because of the process or the new card. Its easier/smarter to put the new highend card on the older process, that way you know its only the card if you encounter problems. (midrange cards aren't as complex, so they are safer to test on a new process.) This also gives you the added benefit of offering a refreash on the highend once you move it to the newer process and can offer faster speeds. I've heard the rumors about TMSCs 40nm and in no way am I claiming that it doesn't have any problems, I'm just not convinced that AMDs decision to put a midrange card on 40nm is a bad sign. Test the new process with a smaller/simpler card, then bring it to your card family.

Mostly correct. They could disable more SP and bring it down, with the rumors about their yields I wonder why they haven't done that. I suspect it has everything to do with how wonderful the g92 is, and the "poor" performance of the original GTX260. Going from what I remember, the GTX260 wasn't much faster then the 9800GTX. (20%, sometimes even less.) If you disable another group or two, it might be only as fast or even slower then the 9800GTX. And because the die is smaller, its cheaper to make the 9800GTX then the "GTX240". Nvidia blew it when they came out with the GTX series, lets hope they learned a bit when their DX11 parts arrive.

I agree its a test chip at 40nm as well as an early intro chip for their mobile solutions as well. It doesnt make alot of sense to move to 40nm for a few reasons. The economy, its more costly, riskier and with DX11 and a whole new arch coming, its too much all at once for just a simple shrink, unless you are getting a jump on things like how your former process scales, and also moblie.
Each shrink we see from now on, at least until theres a bonafide HKMG alternative, or HKMG itself, is even more riskier than in the past. Its not like cpus, where they spend over twice as long before a new arch comes about. And yes, even tho the true arch hasnt changed from the G80 to the G280, or the R600 to the R700, it does have to be alot more flexable than a cpu

"BTW poor OP" wasn't an insult from rangers. I don't think that's what he intended.

But that's my opinion.

Your half right, well your a quarter right. Nvidia isn't in trouble, as a company its richer than AMD/ATI (well atleast the ATI division), Don't forget that AMD needs to compete against Intel and Nvidia, and it was losing to both for a while. AMD's money problem was MUCH greater than Nvidia's now.

Don't forget this isn't a dick measuring contest, its more of p/p contest, and AMD is winning, but don't count out either company. Don't forget that AMD's stock plummeted about a month ago.

As for your Nvidia chipset being uninformed, thats a stupid way of saying it, seeing as not all of us can put down cash for the new i7s CPUs, which forces us to buy DDR3 ram, new board etc, and with the old CPUs we need to be doing sli so we need the Nvidia chipsets. Yes they suck, every1 knows it, but don't think that we don't need them ATM for older CPUs. So its not about being uninformed, its about not having the means to use the Intel Chipsets.

But you were right about 1 thing, the 4860 is a Mobility card:
http://ati.amd.com/products/mobili [...] index.html which is replacing the 4830 mobility if I'm not mistaken.

280M is a die shrink, overclocked and more SPs. I believe...
Don't forget that ATI owns almost nothing of the Laptop industry, it might one day, but not right now.

as for your last comment, yes 1 company is playing it safe for the recession and the other is throwing risks. Which would u prefer?

Fanboyism gets you know where, just clouds your judgment, this isn't about whos the better company, its just who brings out the card for you.

Mark my words, ATi shuts down, Nvidia boys will be crying, and if ATI shuts down well its the same thing.

I'll be in here laughing at the ppl that worshiped either company to the end. Frankly I'm about to let go of both, and stick to console gaming. Can't really tell the difference from a far, its cheaper, With the ps3 having a second launch now, the dev kit that puts more power in it, and stealing 360s games, it should be a good year.

Ati and Nvidia suck, its just that ppl need to pick a side, only smart ones are down the middle, who can see both faults.

Liquid no offence meant but actually claiming to be 'down the middle' when every post you make reeks of Nvidia fanboism is just laughable.

At least I admit I cannot stand Nvidia because of their abhorrent business practice. Why anyone would want to defend this company is beyond me.

I wouldnt defend the 280M debacle. Thats even bad renaming.