I comprehend that, yes. But I don't run any servers at home and servers are a real small percentage of built computers (even smaller than gamers), don't you agree?
You're kind of missing my point. PCI-E 3.0 isn't a must have for gamers, it isn't a must have for the average PC/mobile buyer either. It is however, a must have for OEM's - as a selling point for new systems ( telling customers that their next product release is just the same as the old stuff isn't a great selling point) and both a selling point AND must have in the server/ workstation/ datacentre/HPC markets. It's not a happy coincidence that the largest PC suppliers also happen to be the worlds largest suppliers of professional/enterprise systems (HP, Dell, Lenovo - of which, both HP and Dell have significantly higher revenue than Intel)
I've already mentioned the fact that AMD's inventory ballooned by $250 million this quarter - that isn't from DIY'ers and guys not ordering from Newegg, that's OEM's looking at AMD's product stack and saying "no thanks, we can't sell this because there's no differentiation between this stuff and the old stuff" - Trinity doesn't offer a compelling feature set over Llano, nor Zambesi over Phenom II. If you think that is a false assumption just look at OEM's product lines. Do you think it likely that if Intel keep hitting their bullet points that AMD's $830m worth of warehoused hardware is going to gain in marketability ? and if it doesn't and AMD have to cut and run, what does that do for their balance sheet and their R&D budget?
As far as servers, HPC and datacentres, PCI-E 3.0 is kind of a big deal since the performance bottleneck is latency, and while these aren't massive markets now...
they will be if/when cloud computing becomes more mainstream*..which brings us back to AMD making a huge deal out of HSA...but not following through with the hardware. Why? because Bulldozer and Piledriver are a done deal and need to use an existing PCI express controller and PCI-E 3.0 will need to wait until it can be incorporated into the CPU design. All this brings us full circle - note that AMD's 800 series chipset was a 2010 release. The 900 series chipset is virtually identical to the 800's and was released last year....do you see any murmerings about a 1090X/FX, 1070 range of chipsets for this year -No. No change in feature set = No new chipset = No new selling point or marketing bullet point....You think Asus, Gigabyte, Sapphire, MSI etc will be happy (and making money) pushing existing 990/970 chipset boards into 2013 ? More to the point, how much mileage can Dell, HP, Lenovo, Acer et al get out of the same three-year-old hardware
In the greater scheme of things, PCI-E 3.0 is just an example of where AMD's marketing/sales, architects and management aren't anywhere close to being on the same page. It's a solitary example I picked to highlight a point. I could just as easily noted AMD hitching its wagon to the IBM gate first foundry process, or AMD placing an over reliance upon an ever changing spec for a server CPU to pull them out of the
sh compost (that would be Bulldozer), or...well, you get the point.
Gauging a companys position by the product stack they have at the present time is a false economy, requirements
change both for vendors and end users* Decisions made today won't see the light of day for around two years. This has been AMD's disadvantage for some time- the management have been reactive to market change, not a shaper of the market. And yes, that is partly due to Intel's dominance. It is also more than partly due to AMD's myopic and unenthusiastic board of directors.