Ok. I read something while I was semi-concious during my recent defeat at the hands of a 72 hour bug (that positively whupped me upside the head, stomach and other parts).
I read Cisco is coming out with blade servers.
Ok. Here is the 2 + 2 = 3 moment. Yeah, assume I am off the mark. Pure speculation. Shai, feel free to tell me here that I am full of it.
My guess is, that if this is true, Cisco will purchase Shai’s ScaleMP, and enable its blades to build HPC systems of very large size on the fly.
Simple. Cisco is not populated with dummies. There are some very smart folks there. There are lots of blade vendors. They all offer variations on a theme. Cisco wants to make their different (assuming the story on them to be true). Cisco will make these blades well suited for virtualization … or … aggregation via Shai’s software.
Take them off the market. Restrict them to Cisco’s hardware.
Wow. This could be a very interesting time in the HPC world. Need 128 cores and 2 TB memory for a few hours? No problem. Will assemble for you on demand.
That’s why this is so interesting.
Now couple this with stuff like the Pervasive Software DataRush app, or similar apps that benefit from huge numbers of cores with shared memory. Programming shared memory is generally easier using tools like OpenMP and related. Cilk just released, and several other tools are working on making multi core and shared memory programming even easier for developers. Though OpenMP is really quite easy, it isn’t the be-all/end-all.
My gosh … its like the SGI origin really did win … in terms of concept. But using commodity parts.
Or it could be a flight of fancy on my part.
No, I/we own no stock in Cisco, nor any part of ScaleMP.