^ Some of the decisions Microsoft have made in the recent past have been pretty questionable.
1) Windows 8 and that metro UI - Terrible. I don't fricking want a mobile UI on my desktop.
2) Surface RT - What were they thinking , and now they are launching a Tegra 4 version .
3) The small rift they had with Intel in moving towards an ARM tablet - Now this is an odd one , on one hand , I think it was needed to wake Intel up, but on the other hand launching RT has been costly to MS.
Moving Forward
4) What they should do now is kiss and make up with Intel . Work on getting a Windows 8 tablet on an Intel SOC like Baytrail at the same time dump RT.
5) Personally I'd like to see the opposite strategy for tablets, I'd like to see more of a desktop UI on them. That's one of my biggest complaints with the iPad, its basically big enough to have a desktop UI - and I miss not having the ability to have numerous windows open and running on the screen. Maybe a 12" minimal bezel tablet with windows 8 desktop UI would be awesome.
6) For the Windows based phones, Id like to see them go in a direction the Ubuntu Edge took. A phoone you can dock which then runs the desktop UI so you can use it as a desktop. It would be great to plonk the phone into a dock at work , flick a switch and have it show up on the monitor and usable via the keyboard and mouse. At home I could just dock it to continue working on my word docs etc.
I disagree, Windows RT is the only thing out of the whole Windows 8 movement that made sense. Especially given the pricing war in that space means overall hardware has tended to be lacking for pricepoint when paying the Intel premium.
Intel is having serious problems these days. In terms of finances
Intel's profits dropped 29% despite their gross income only dropping 5%. Part of the apparent cause of this would seem to be related to why the largest fab in the world had problems producing their last iteration of Z-series Atom processors for mobile on their then rather mature 32nm process technology.
The aforementioned Z-series made on Intel's 32nm process made for phones basically ranged from
this to
this. Note these are both single core designs and the top bin has a top normal frequency of only 1.5GHz, which would be painful to run a modern Windows Shell on. This points to issues with making viable bins for mobile, and ARM's newer architectures are more powerful and more power efficient without requiring binning out ideal silicon.
It should be noted that Saltwell is still an in order design, and thus doesn't have the additional overhead circuitry of an out of order engine. Silvermont, aka the 22nm iteration, will have an out of order engine. While out of order engines help with the potential performance a processor can reach, it also adds to the intrinsic power overhead. Architecturally these will tend to be worse in the lows and idles then Saltwell was and need to rely on race to execute and sleep states to try to compete.
In terms of process technology at this point Intel and everyone else is using Argon-Fluoride lasers for their lithography.
Those lasers have problems with imaging below the 28nm node. With their 20nm node TSMC is switching to double patterning, Intel has been holding off doing full scale double patterning till 14nm.
Note that holding off full double patterning should tend to mean their defect rate is higher, which there have been indicators of since Ivy Bridge debuted their 22nm process. It should be noted IBM, TSMC, Global Foundries, and Samsung are prepping their 20-22nm processes for 2014 when Silvermont is expected to hit so Intel won't really have a node advantage when Silvermont hits. Indeed given IBM will bring SOI tech to that node and TSMC is making double patterning policy, Intel's Silvermont is actually liable to be on a worse process then what will be available to ARM space which combined with ARM architectural tricks on these newer designs means Intel will be in a worse position then ever relative to ARM.
All of that however is secondary to Balmer's madness related to trying to transform Microsoft into a "platform company" and in more recent interviews a "devices and services" company. Personally Balmer killing off Flight Simulator and Encarta is the axe I have to grind as much as anything else.
