Comment Re:Duh (Score 1) 454
Then the employer would only consider hiring H1B's from other companies and not do the petitioning themselves.
Then the employer would only consider hiring H1B's from other companies and not do the petitioning themselves.
t I was trying to explain that there was a vast number of applications using classic supercomputer type technology, ranging from academic research down to multiplayer games. A modern game console now uses multiple cores, vector processors, vector chaining, kernels (if you consider vertex, fragment, geometry shaders as kernels), client-server communication to update players moves. Even geometry data is streamed across the network as some game MMORG worlds are so vast, all the data couldn't be stored on one disk drive. Then there are techniques of "sharding" games world so that the same geometry can be used, but with different groups of players.
Those numeric and simulation libraries were optimized in conjunnction with the Fortran compiler to take advantage of the hardware. The most obvious example; having fixed sized multi-dimensional arrays as global variables. For regular grids, the compiler can then decide which way to slice that data block up so that every processing node gets assigned a chunk of data. Since each function is not more than a few loop counters reading previous and current state for each grid cell, those get optimized into parallelised for-each calls. Once the scientists and engineers started writing and sharing numeric and simulation libraries and building on top of them, there isn't any need to change.
That's why many simulations are still written in Fortran - the compilers were optimized to handle multi-dimensional grid arrays, which is what fluid dynamics and other solvers use.
The number of floating point operations (FLOPS) performed by a next-generation game console outranks early days supercomputers like the Cray.
Cray-2 = 1.9 GFLOPS
http://www.dcemu.co.uk/vbullet...
Dreamcast | CPU: 1.4 GFLOPS | GPU: 0.1 GFLOPS | Combined: 1.5 GFLOPS
PS2 | CPU: 6 GFLOPS | GPU: 0 GFLOPS | Combined: 6 GFLOPS
Xbox | CPU: 1.5 GFLOPS | GPU: 5.8 GFLOPS | Combined: 7.3 GFLOPS
Wii | CPU: 60 GFLOPS | GPU: 1 GFLOPS | Combined: 61 GFLOPS
Xbox360 | CPU: 115 GFLOPS | GPU: 240 GFLOPS | Combined: 355 GFLOPS
PS3 | CPU: 218 GFLOPS | GPU: 1800 GFLOPS | Combined: 2018 GFLOPS
PS4 | CPU: 102.4 GFLOPS | GPU: 1843 GFLOPS | Combined: 1965 GFLOPS
Console games are getting to the point of having 2 TeraFLOPS of 32 floating-point performance
Supercomputers are designed to be unlimited in scalability (super-scalar). Everything is duplicated from the cores on a single chip die to the boards, racks, rack-frames, aisles of rack-frames and interconnect fabric. The only limits to the size of a super-computer are financial; component cost, office space lease and electricity bills. Usually, it's the last one that's the problem. The slowest proocessing nodes can be pulled out and replaced with more powerful ones as time goes by.
Fourier analysis was first developed in the 1800's. It took 80 years for the first programmable mechanical hardware to appear in the form of weaving looms in the 1880's. Then the development of mechanical analysis systems like this happened another 20 or 30 years later. Another 70 years, and we can play music on our home PC's and see funky animated digitial audio equalizers.
There was the concept of the "Perceptron". You have your camera that takes live video. This feeds into the perceptron. At the lower levels, edges, arcs, corners and dots are detected. Then at a higher level, shapes like circles, squares and triangles are detected. Higher still, objects like faces, cats, and balls are detected.
The brain seems to generate a set of hypotheses about what something could be then pick the closest match.
Human retinas have a resolution of 100 million neurons each. But there are several layers to the retina that detect spots, edges, color opposition (blue vs. yellow, red vs. green, white vs. black). All of this information gets compressed down to around 1000 chunks of data which then go through 10 million neurons in each optic nerve.
You have neurons, which are arranged into "cortical units". These in turn are arranged into wide striate layers (for increased resolution) and pyramids (for higher levels of cognition). With human vision, the neural pathways follow the topology of the retina.
http://en.wikipedia.org/wiki/T...
With human audio, the neural pathways follow the frequency of sound (http://en.wikipedia.org/wiki/Tonotopy)
This research paper covers the evolution of the human brain when compared to reptiles and other mammals:
Many of the cities in the Bay Area were originally agricultural. Retirees moved there for the sun, peace and quiet and cheap rents. Then the tech industry started to grow. For every 100,000 square foot office block built, that's 1000 employees who want 4000 square foot lots for their homes. All the land got used up rapidly for roads, homes, offices, schools, hospitals and clinics. And those came at a cost. Retirees suddenly saw their property taxes go up and up to pay for all these services that they didn't use. The cities then get round this by granting permission for a company to build a new campus on the edge of their city, leaving the housing, schools and transportation access to their neighbors. The same retirees opposed high-rise apartment blocks because they lost their sunlight, and MVA (market-value assessment) meant their home was assessed the same value as the six unit triplex block next door. So they brought in a tax Proposition to grandfather in property taxes and block the construction of high rise concrete apartment blocks (also due to earthquake risks).
Thousands of cores means a web service that can scale up infinitely and is invulnerable to a DOS attack or whole group of slashdot readers arriving like a horde of barbarians on a horseback raid.
A 2005 GPU based laptop with 2.8GHz dual-core CPU can still edit 500,000 triangle meshes.
It's called a "reality tunnel". Working for a startup involves writing new software. And you have to believe that the software is going to be useful to other people in order to have motivation to keep working on it. So there is that kind of belief.
Maybe there are radioactive elements underground as well.
For me, the best the sleeps I had were in a hotel room which had air filtering, blackout curtains and was on the end of the top floor, well away from all the other guests banging and clattering their suitcases through the corridors.
But move to the same kind of room right next to the main hallway, and it was impossible to get a deep sleep, because there was always someone every hour who figured the best way to open a door that opened inwards was to hit it with a large suitcase. The same thing happens if the hotel room has emergency lights that come on whenever the main lights are switched off.
Air flow could be another problem. Even on Earth, sleeping under a lie-in (sloping part of a roof) always gives me a sore head due to the lack of air flow. The CO2 seems to build up. The only way I could stop that, was to sleep directly underneath the skylight window and keep it open. Maybe the shape of the sleeping pods leads to CO2 build up.
You are doing bug tracking by hand??? We used to that back in the 1990's. I was given a three day task to sort, reorder, prioirtize about 200 open tickets in a single text file. Wrote a script in six hours to do this automatically, and had the report completed in minutes. Today, we would just use bug-tracking software like Jira
Math is like love -- a simple idea but it can get complicated. -- R. Drabek
