I disagree with your assessment and I will outline why I think your doom-and-gloom scenario is not based on market reality.

Nvidia already has > 300 design wins in the notebook sector with Kepler (more than it had with Fermi). OTOH, AMD's new 28nm GPU line thus far focused entirely on the minuscule >$450 desktop discrete GPU market segment. Based on how HD7950 is barely 20% faster than HD6970 but costs $450, it's hard to imagine anything below HD7890 to be anything impressive. I can't even see $299 HD7870 beating the former HD6970. I am guessing AMD will discontinue HD6950 2GB on purpose because it doesn't look like HD7870 will be any better despite an $80 price hike. In light of HD6950 1GB, HD7850 at $249 will be an impossible sell. HD7770 looks to be crippled by 128-bit memory bus. Yawn.

Not to mention, from a consumer demand perspective, NV is targeting Kepler for Ivy Bridge launch and focusing on notebooks (specifically Kepler is trying to win a lot of designs in the Ultrabook segment). Probably around that time a lot more people will be building new rigs vs. Q1 (a slow quarter). Finally, it's not like NV doesn't have any good cards right now < $300. Why hasn't NV lowered prices on GTX550Ti, GTX560, GTX560Ti, GTX570? Could it be that those cards are still selling well?

Consider this: Last time AMD was 6 months out of the gate earlier than NV. Last time AMD priced HD5850 at only $259 and HD5870 at $369. Last time, Fermi was hot and loud and it took NV almost 8 months to deliver GTX460 after HD5850. Last time, NV had their entire HD5000 series line-up out on both the desktop AND the notebook market before a single Fermi was launched. The head start AMD had was prior to the Holiday Xmas of 2009 too (Q4 has the most demand for PC upgrades in the year!!)

And even after all of that, NV still commanded a 59% desktop discrete market share on the desktop, with about 50% overall, and delivered 3 great financial quarters while AMD's graphics division struggled to turn a profit.

So it doesn't appear that launching 6 months earlier automatically means the competitor is in trouble, unless of course there is very strong demand on the part of consumers. But why would there be a strong demand for new graphics cards? Most of the titles coming in 2012 are console ports. The most significant PC titles of 1st half of the year are Diablo 3, SC2: HofS, Mass Effect 3, but none of those titles needs a next generation card. There is also Max Payne 3 and Syndicate but since those are console ports, it's unlikely they will require anything faster than a GTX280/HD4890 to run. Where is the demand for users with anything faster than GTX280/4890 to upgrade? Eyefinity and 30 inch gamers are a tiny fraction of the overall market.

Actually, chances are that given current market trends of notebooks/laptops exceeding the popularity of desktops, a lot of gamers who will want to play Dota 2, SC2: HofS and Diablo 3 will also be playing on laptops. But where is AMD's new 28nm notebook GPU line? Nowhere, except re-badged HD6000 parts. Desktops are still plagued by hard drive shortage/high hard drive price issues and at this point a lot of people are going to be holding off building new systems due to IVB.

Moving on to Q3. We'll finally see the first demanding PC game in Arma III and the launch of Windows 8. Back to school season and Windows 8 is the first legitimate period of spiking PC sales growth. Q3 and Q4 2012 is where the real battle for market share and profits will take place.

I am not trying to discredit what AMD has accomplished, simply trying to put things in perspective for people who think NV is "in trouble". What NV has done is instead focused on markets that are growing the fastest, such as Tegra 3 in smartphones. NV is also making a lot of $ in the lucrative professional markets with Quadro and Tesla lines. Nvidia understands, the desktop market for high-end graphics cards is not growing by a lot at this time.

What the PC gaming industry needs are 2-3 next generation games like Crysis or BF3 or Crysis 2 in 2012, etc. Until then, only the most hardcore gamers will care to upgrade at this point or people building a new system from scratch. The delay of next generation PS4 and Xbox 720 is really stagnating the PC gaming industry in terms of graphical advancements. As such, I only see that hurting both AMD and NV. We'll just have to wait and see what happens in 2012 in terms of discrete desktop GPU growth. My $$$ is on the laptop discrete GPU market becoming far more important because most people today don't want to build desktops. And us current desktop users aren't going to be spending $450+ on new GPUs to play console ports, not when $200 HD6950 and GTX560Ti is more than capable.

Sour grapes

No, actually being objective about the sad state of affairs in the PC gaming industry and the latest generation of cards. I have been building PCs since Radeon 7000 time and this is the worst time to upgrade for several reasons:

1) Without seeing what the competition brings, it's difficult to be impressed by 20% higher performance > 1 year later;

2) Without next generation games, the only benefits of a new card are Eyefinity (bezels suck) or applying high levels of AA (meh).

3) Given how HD7950 is barely faster than HD6970 (http://www.hardwarecanuck...ahiti-pro-arrives-24.html), that means the entire series < HD7950 is not going to bring much more performance at similar price levels -- which is the whole point of progress in GPUs.

4) Most games today can be easily played on GTX560Ti/HD6950 and those cards are only $200.

5) The future-proofing argument doesn't work either since in 12-15 months when next generation of games come out, we'll have even faster cards.

6) Notebook discrete GPU is huge market. AMD has shown nothing in that regard. On the desktop, discrete GPUs priced above >$199 comprise only 14% of the entire GPU market:

http://www.xbitlabs.com/a...idia-geforce-gtx-460.html

The bottom line is that HD7900 series still cannot provide good playability in games like Metro 2033, Witcher 2 or Dragon Age 2, etc. And at the same time, it's too fast for console ports. It needed to be either MUCH faster or MUCH cheaper. It delivered neither.

I have no problems if you addressed the points I brought up. Instead, your response lacks any sort of counter-argument?

Plenty of people have cards that are now 2 or more generations old. They may want to upgrade now.

1) >20% performance (remember driver optimizations are forthcoming) with lower power consumption is impressive. Also, the few situations where NV's GK104 card will beat Tahiti is in contrived benchmarks and a few TWIMTBP games.

2) Some people like those things. the 79x0 cards are high-price, low-volume products. There's nothing wrong with that.

3) Your link shows the 7950 is about 20% faster than the 6970, which is already pretty fast. I question your understanding of the word "barely." If cards at each level bring 20% better performance at a similar price, what's the problem?

4) People willing to pay >$400 for a graphics card are doing things that those cards will suck at i.e. 2560x1600, eyefinity etc.

5) This is always the case. If it doesn't work for you, come up with another rationalization, or never buy another graphics card because a year from now there will be a better one.

6) So, because AMD hasn't released a product to compete with a product NV hasn't released yet, their 7900 cards aren't good? xybit said NV would be envious, not be beaten in all market segments by huge margins.

Your response to a total of twelve words from xybit seems a bit defensive. You are entitled to your opinion. If you don't want one, don't buy it.

1) Driver optimizations will apply to new NV cards too; so an irrelevant point. 20% is not enough to provide more playability over GTX580 in demanding games. Go ahead, check any review with modern games at 2560x1600. HD7970 is still not fast enough. Yet, it's way too fast at 1080P, wasting its potential at resolution where HD6950/GTX570 is already fast enough.

2) Ya, but that entire market is very very small. The majority of users do not care for these things. They want faster cards at <$300 and so far AMD has not delivered at all in that regard over HD6950/6970/GTX560Ti/570.

3) What are you on about? HD7950 is 20% faster than HD6970 but HD6970 cost $369 not $450. Also, HD6950 was unlockable essentially giving us HD6970 level of performance for $250 or less. So what you are saying is let's ask $450 ($200 or 80% higher price) for just 20% more performance. Right....that's not at all like getting 20% better performance at "similar price". The only card that was similarly priced to HD7950 was GTX580. HD7950 is only 5% faster for $50 less. Alternatively, overclocked GTX570s were $370 with ~ GTX580 level of performance 12 months ago.

4) Doesn't make any sense. HD7950 and HD7970 are not fast enough for 2560x1600 gaming in the demanding modern games where HD6970 and GTX580 struggled in the first place (Metro 2033, DA2, Witcher 2, etc.). A single HD7970 is also not fast enough for 3x 1080 Eyefinity gaming (http://www.bit-tech.net/h...deon-hd-7950-3gb-review/4). Therefore, you'd need 2 of those, which means $900+ on GPUs, at minimum. Who spends $900 on GPUs without seeing what the competitor brings at the very beginning of the new 28nm generation?

5) It's not always the case. Last 40nm generation lasted from September of 2009 when HD5870 came out to January 2012 when HD7900 28nm came out. That means for the next 1.5 years or so, we are entering the 28nm-generation. AMD just showed their hand and it only delivered 20% more performance than NV's best card on 40nm. That means there is a huge possibility for a price war or for NV to deliver a much faster card at 28nm since AMD was very conservative with factory clocks. Normally, a generation jump (i.e., 9700Pro --> X800XT --> X1800XT, HD3870 --> HD4870 --> HD5870 brings a massive performance increase). In this case, it didn't happen (not without 30% overclocks on the HD7970).

6) Why would NV be "green with envy" when they had 300 notebook wins with Kepler, more than they ever had with Fermi? That sounds like they are doing much better already. Since AMD has not brought anything to the table in regard to 28nm HD7000 GPUs on the notebook segment, why would NV be "green with envy"?

We'll see how 2012 plays out. I remember when NV launched GTX280 for $650. Shortly, AMD responded with HD4870 for $299. I can't afford to drop $600 on a GPU and not care if the competitor brings a $299-349 card that's only 20% slower OR if the competitor's $550 GPU is 50% faster than GTX580, which by all rumors is actually very very likely.

By all means if you can upgrade every 6 months, go ahead and grab the HD7970.

http://lenzfire.com/2012/...price-release-date-43823/

Nvidia Kepler GTX680

850MHz Core clock
2 GB 5.5GHz GDDR5 Memory
1024 Stream Processors
512bit Bus Width
Priced at $649
45% faster than HD 7970

Nvidia has a clear winner in its hands and that’s why they have commented that they had expected much more from AMD. But still we already know that AMD is preparing its revised GCN HD 8000 series graphic cards, which may be release in the latter half of this year.

Nvidia Kepler GTX670

850MHz Core clock
1.75 GB 5GHz GDDR5 Memory
896 Stream Processors
448bit Bus Width
Priced at $499
20% faster than HD 7970

Some people like having the fastest, newest things. They will not be discouraged by your troll posts on XBit articles. Just sayin'

Ya, if you can afford to buy $450-550 graphics card every 6 months, then it doesn't matter at all to wait for NV's offering. But then people who can easily afford to buy $500 cards every 6 months don't really care to read Xbitlabs do they? They could just as easily walk in into a computer store and ask for what the fastest card is, get the answer HD7970, grab it and leave.

GTX280 was >60% faster over 8800GTX.
GTX480 was >50% faster over GTX280.
http://www.computerbase.d...schnitt_leistung_mit_aaaf

That means there is a strong chance that GTX680/780 will be >50% faster than GTX580, leaving HD7970 about 25% slower in the process. If you are impressed by a mere 20% bump in 14 months over GTX580, then what's another 6 months wait for likely a 25% increase over HD7970?

This time for NV, it will be a major gate shrink (40nm to 28nm) plus a new architecture. Last time that happened was 8800GTX vs. 7900GTX (>2x performance increase).

I'd expect a much faster card from NV and from AMD to retaliate once 28nm process is matured. Right now AMD is just trying to get yields higher by 'underclocking' HD7900 at the factory. I fully expect AMD to have HD7980 @ 1150+mhz by the time high-end Kepler is ready. I think it's worth waiting for since there isn't a single next generation game coming out on the PC in the next 6 months.

Exactly! Perfect series for people that run HD4890 or slower. HD6950/6970/GTX570/580 users are wasting $ imho.

HD7900 series doesn't do much better in console ports but costs $200-300 more:
http://techreport.com/articles.x/22384/4

At the same time, it chokes in very same games where HD6970 and GTX580 aren't sufficient:

http://techreport.com/articles.x/22384/6

I think you are making the best decision to wait for IVB. You'll get a new rig and hopefully with more competition prices will drop. If you waited this long, might as well wait until April 8-17th (IVB launch week).

No, I am just giving good advice for people to wait because there is a huge chance either NV delivers much faster performance at similar price levels OR it brings GTX580 level of performance for much lower price than HD7950 did.

Considering Q1 2012 is all console ports for the PC, that's actually very sound advice for people who don't make $600 a day and can afford to upgrade to the next best thing.

Come back in April 2012 and tell me I was wrong IF Kepler fails to provide way more performance or brings an excellent price/performance card in the form of a GTX660Ti and I'll fully admit that I was wrong. Otherwise, only an AMD fanboy or a person with high disposable income would blindly buy an HD7970 for $550 today.

It actually would be very unwise to advise that anyone spends $450-550 on a new GPU without seeing what the competition brings, unless of course you only buy AMD cards....but even in that case, we might see much faster HD7970 from AIBs or price drops. This isn't like HD5870 when AMD priced their cards very attractively and Fermi was ways away. Kepler is almost here and rumors point to it completely blowing HD7970 out of the water. This is not at all like Fermi, which was nowhere near in sight, and rumors pointed to it being barely faster (~15%).