And there is no better way of screwing yourself over than by delaying things on the PC front.
If you want to make the PC the best is can be and show just how poor everything else is, you want to storm ahead like a powerhouse.
Not slow down and confirm that it's not advancing.
I totally agree with you but Intel will easily bankrupt themselves if they don't make money off the advances they're making.
Not to mention, the market has been declining for the past 4 years, this isn't news to anybody. Getting Broadwell out right now doesn't change anything, it's not going to get the market out of the decline.
Trust me, Apple, GF, TMSC, Samsung, or other fabs can't do this better than Intel. These fabs are starting from a weaker position and when they get to Intel's levels, they'll run into the same thing.
Right now, Intel probably would not do this again if they would open up their fabs for other companies to use.
Apple buying access to Intel's fab to build their A series of CPUs would be awesome and would help fund Intel's future work on fabs. Not to mention if Apple starts building their own GPUs later on, that's more work for Intel.
How about an option to have 32GB of RAM?
Why 32GB?
- Run multiple VMs concurrently to simulate a cluster
- Run databases and be able to cache lots of data
- Scientific computing
- Run the whole Hadoop stack natively and be able to give the different components a lot of RAM
It is a limitation in the design they're using. They can't stack 4 sticks of RAMs in there, takes up too much space and may require more cooling that they don't want to compromise.
You'll see 32GB coming with DDR4, which can be denser and more power-efficent. Intel is not going to support DDR4 until Skylake in later 2015 or early 2016. However, seeing the delays they went through the last 2-3 years, it is likely late 2016 and early 2017 at this point.