IF Intel were to ever buy nVidia... it wouldn't be at anything close to their current market worth... intel can be quite brutal when it comes to aquisition, they'd insidiously cut nVidia to its knees and buy them up cheap (or just attract their main staff away and basically gut the company that way).
RE: Drunkenmaster - I heard the same things said about nVidia with the FX series launch... and they are still very much here and now... so I don't think they are going anywhere quickly. Tho the failures with tegra will have hurt them badly - but they've always come back from being on the backfoot in the past so I wouldn't be so quick to count them out.
Oh and ARM is just awesome
EDIT: Oh and yeah intel will never succeed in the performance graphics market however much money they throw at it... despite being capable of very good software and hardware development... unless theres a huge shift in mentality - their whole approach to products while working very well in other markets doesn't suit performance graphics well.
See you said this, for a long time, every time Larabee was brought up, however their Sandybridge GPU shows a MASSIVE, monumental leap forwards in low end gpu power, also most importantly, Anand who used to mock Intel IGP's for giving awful IQ and dodgey drivers suggest their new GPU is massively massively better and giving equal IQ aswell as performance to a 5450.
Lets be honest, theres VERY little difference between a small 12core Sandybridge gpu and sticking one with 10 units stuck together on a discrete card.
They've clearly addressed, significantly, IQ, power, ability to deliver it and considering the supposed power of Sandybridge chips, their 12core gpu seems set to use very little power, to the degree that you could stick 20 of these units together and have a workable product.
I mean a Fermi high end and low end, a Nvidia IGP or AMD IGP is little more than just one or two units of a high end card, the architecture is all but the same.
Also no, no one said anything similar at the FX launch, no one predicted Nvidia's downfall or their unwillingness to change their ways, it was one bad product, with a good one before and after it. the 280gtx through to the current gen ALL show the same signs of sticking to certain design philosophy that has made them progressively worse over 3 years.
The situation around the FX wasn't the same, the FX was memory bus limited, it was on a new process, it had a new type of memory that wasn't working well, the fx5900 came out not much later at all, with double the memory bus, back to the original ddr memory, with almost certainly some idea's of fixes for the new process, etc, etc. THe problem was, when Nvidia had that trouble, it was just too early for the process, it wasn't part of an ongoing problem, the volume part of the company wasn't about to fold, and they were selling gf4mx-ti in the rest of the range without issue that made profit and were good cards for the prices.
Same with AMD and the 2900xt, while that and the FX5800 were late and had problems, the ENTIRE rest of the range were selling like hotcakes, making the companies oodles of money and neither company had ANY problem, it took Nvidia 3 months to go from a uncompetitive 5800(with the first dual slot cooler) to a 5900 which was hugely better and had no problems.
These days, Nvidia is 6 months late, have EOL'd the previous chips, have no mid or high end cards selling in quantity and the low end is based on the previous gen, isn't that competitive. LIkewise around the 5800 Nvidia had huge brand recognition with partners who loved them, this time around they've gone through 2 years of publicity disasters, hugely higher RMA numbers than normal due to faulty parts, huge problems for many of their partners. Nvidia can't count on the console market, probably at all, in the next gen, they can't count on the low end, they can't count on Tegra for a long while as it quite obiviously needs to be redone from the ground up. Considering everyone else can get working ARM cores done right, on spec, on time, the only difference is Nvidia using its own gpu's on board, so take 2 guess's which part of Tegra is over power and late and off spec? Again pointing to the fact that its their own gpu's that have fundamental issues, those being, they are completely unsuitable for mass production on process's this small, they have old thinking and its killing them.
Remember when the 5900 came out, it was a high profit part, and the low and mid end were making money and they had the massive share of the market.
This time when Fermi finally launched, it made a loss on every sale, and continues to be utterly slapped around at the current price the 470gtx sells at, the "5900" of the Fermi, the GF104, isn't fast enough to compete while the fx5900 was on par with a 9700, the gf104 is massively slower than cores that are smaller and sold at a higher cost, the GF104 is barely scraping a profit.
The situations aren't remotely comparable, for it to be the same, for AMD, while the 2900xt was screwed the x1950pro would need to be making a loss or just breaking even, or EOL'd, and the low end would have had to be up against Nvidia's 8xxx low end, it wasn't, Nvidia's mid and low end weren't out before AMD's, the x1950 was competing against the 7900, not the 8600 it should have been, same goes for the low end.
Nvidia is screwed across their entire product range, in the 2900xt or the fx5800 days it was ONE product, with the rest selling with a healthy profit with huge volume. None of that is true of Nvidia now.
