Hey, look at this

I can make a graph layer with this

Cray is looking for a Senior Semantic Web Engineer in Pleasanton, CA.

Qualifications for this position include: “Sound understanding and experience with RDF. Proficient in the SPARQL query language. Knowledge of Ontology Modeling and Description Logics. A minimum B.S. degree, or equivalent, in a technical field (Engineering, Science, Mathematics, Computer Science, etc.); advanced degrees desirable. Proficiency in the following: C, C++, multithreading programming, Java, Linux, shell programming, Semantic databases, web 3.0 applications stack. Analytical problem solving ability. A thorough understanding of vertical informatics applications and databases.” Semantic Web

The reason I quit the industry (aside from being a jerk) is that I find the process of convincing the hiring manager that I am qualified actually is slower than getting the job done via open source. So, what does Cray thinks it knows that is a big secret, anybody guess? I can get more things done with no cost by stealth with this blog and open source than anything Cray has invented. We, have the entire research of convolution style data queries going 30 years back.

You also know right away something is wrong, because you actually have to go to Pleasanton and stand in from of the big red thing. That not semantic networking.

My best advice to Cray, push the BSON folks on a graph convolution, a superfast graph layer embedded into the risk architecture. Once the BSON folks get moving then Cray can take their big box and claim to have the fastest semantic network available, open source, open standards; leverage the demand for this stuff from paying customers.

What if they are targeting me?
I know my ontology stuff comes up on their searches, they are in this huge business already. What can I do for them? Well, send me to the BSON committee with some cash and I can bribe and kick butt. The fact that I am being a jerk on Cray's behalf would certainly indicate to everyone that Cray will have the fastest semantic net available. But, again, this blog, some occasional market and technical research and detective work, I can get Cray what is wants. But at the same time, I get Sea Micro what it wants, but them's the breaks.

Look here is some architecture:

The XK6, announced on Tuesday, is made up of multiple supercomputer blade servers. Each blade includes up to four compute nodes containing AMD Opteron CPUs and Nvidia Tesla-architectureGPUs. It marks Cray's first attempt to blend dedicated GPUs and CPUs in a single high-performance computing (HPC) system.

NVIDIA, the same crew that screwed up my graphics driver on Windows 7. I could look at what Nvidia is doing. Maybe they can build a triplet flow architecture, move a few million of them per second. Parallel processing is what NVIDIA is pushing. Does this work for graph layer? Dunno, yet, depends on how fast we can get in and out of each processor and how much cache each one has.