Version diversity isn't the only kind. Implementation and hardware diversity matters too - for instance, I've run into a crash bug when attempting to start a new Activity from within a TabHost that only occurs on Galaxy S devices. That sort of thing is really incredibly frustrating, and makes QA far more of a pain in the ass than it should be.

Apparently the Libertarian Party had ballot access in almost every state in 2008, and I assume the same will be true this year.

http://www.lp.org/ballot-access

Take a look at the benchmarks. The FX-8150 really doesn't come out looking good against the 2500k, much less against the i7-980.

The sensor is 41mp. They're subsampling, not supersampling; generating a (supposedly) superior 5/8mp image from an original, potentially fairly noisy but high-detail, 41mp image.

And they still sell Power7 with 8 cores and issuing 6 instructions per cycle at 4GHz+. They're obscenely fast, but they're also not cheap unless you're comparing them to Itanium, SPARC, or Intel's -EX series Xeons.

Bulldozer - their current architecture - was really bad. Slow, mediocre price/performance ratio, and power-hungry. It remains to be seen if Piledriver can make it all better.

A company that fronts as an intelligence publisher... but is secretly an intelligence publisher? Oh, Associated Press, why you make no sense?

CTSS had email before UNIX did - 1964, if I recall.

No, you. The 96km is a mistake in the summary. The number in the article is 96,000km.

FMA is pretty much irrelevant for any of the above workloads.

Real-world benchmarks that aren't linpack. Video encoding. Games. OLTP and OLAP workloads, as tested with MySQL, PostgreSQL, and Microsoft SQL. Even the TPC-C results are pretty unimpressive.

You seem way too personally involved in this.

For specific workloads, Itanium is great. It can sustain 2 FP loads, 2 FP stores, and 2 FMA's in a cycle, which means for certain types of DSP-ish workloads, it has more performance per-cycle than just about any other mainstream CPU. It also has a very high-performance cache hierarchy, with massive blobs of SRAM and a low-latency L1 (one cycle to access.) The problem is that it's expensive, clocked low, and not really ideal for where it's marketed (the mission-critical enterprise server business.) It still has significant advantages over Xeon for some workloads, namely things that are highly cache-sensitive or that scale high enough where directory-based coherence is good to have.

A 3GHz Sandy Bridge core completely annihilates a 3.6GHz K10 core, and Bulldozer's per-cycle performance is significantly worse than K10.

You should come out of 2002 sometime.

I live in a medium-sized Kansas city, and 3G is almost unobtainable. No T-mobile presence (as far as I can tell), no AT&T 3G, poor Verizon and Sprint. The only carrier with solid infrastructure here is US Cellular. (I'm on Sprint, and it's frequently an exercise in patience.)

They're working on it; they have an indigenous light-fighter project, and are co-developing PAK FA with the Russians.