You can run 16-bit applications natively on 32-bit, but not on 64-bit.

No, I booted it from USB. Disabling the ExpressCache had to be done from Windows, no BIOS option. There is some option with the Samsung software to do it.

I tried to install Ubuntu 12.10 a few months ago, using the UEFI boot instead of the regular BIOS boot loading options on a Samsung laptop. The installer started, and all I got was a black screen. When I tried to turn it on again, all I got was a black screen. I assumed it was a hardware problem, and managed to get a replacement laptop. I then tried to do the same procedure again, and I also managed to brick the second laptop. Since the internal SSD is not serviceable, I was not able to resolve the issue, and Samsung was unable to help me in any way. I returned the second laptop, and then I disabled the ExpressCache from Windows before I wiped the system and installed Ubuntu Linux without using UEFI.

Indeed, do you have a source for the 30-50 ms faster figure? My source says 80 ms faster than visual stimuli. Source: "We ran a t-test based on the tabulated average individual reaction times to each stimuli, and established that the mean average individual reaction time to light (0.28005 sec.) was statistically significantly different than the mean average individual reaction time to sound (0.20407 sec.; P-Value= 1.79E-07 .05, the statistical level of significance)." ( http://www.colorado.edu/eeb/courses/1230jbasey/abstracts%202010/37.htm )

NP is "Nondeterministic polynomial", not "Not polynomial."

And to think I used to wonder why some people looked at VB experience as essentially worse than no experience at all...

Indeed, I had a project a while ago in which we used two separate cameras mounted on a robot. We barely had to touch the camera rig to get the cameras out of alignment. Once it was mounted it was pretty good though. There are companies such as Point Grey Research that makes Stereo Vision Camera setups that stay aligned properly even if you move the rig around, but such setups aren't particulary cheap.

The kinect doesn't have stereo cameras. It has one color camera which isn't really used for much, a IR projector (that projects IR dots all over the scene) and a IR camera. The IR camera uses the pixel distance between the dots to find the distance. The depth image you then get is used as input to the algorithm that detects the body parts and their orientation etc.

The last university I was at provided shell access to 64-bit Linux and Solaris to all students, but also AIX if you had supercomputing classes. It'd be interesting to know if the 64-bit Linux system was exploitable, but I'd rather not give it a try.

It's the same in Norway, during the first year we have courses with failure rates in the 40%-50%s, and I've even seen 4th year courses with 75% failure rates. Some courses have one A every 3rd year or something. No one has ever graduated from my faculty with straight A's. When I had Physics I, if you got a D you were in the top 10% (we use the ECTS system, E is lowest passing grade).

Switzerland is almost like that, they have 5 CHF coins (3.5 EUR). I wouldn't mind it if they did something similar in Norway (turning 50 NOK bills into 50 NOK coins (6.3 EUR). But, we already have 20 NOK coins (2.5 EUR), so it's not that bad. As a side note, our least valuable coin is 50 øre, which is 6.3 Euro cents. We're talking about removing it in favor of just using 1 NOK as the smallest coin (12.6 Euro cents). I haven't seen anything costing less than 1 NOK since the early 90ties, so I don't think it'd have a huge impact (some sweets etc that are 2.5 NOK now, might go up to 3 NOK, but that's about it).

It's not the cube of the distance. It is governed by the inverse square law (I = I/d^2). The strength is essentially energy/(surface area of a cube) = Energy/(4*pi*r^2). This means that the strength will decrease by the square of the distance (and not the cube of the distance). This is provided the antenna is omni-directional, if it is directional, the signal will be even stronger.

indeed, we did the same thing, except we fit the matrix into L1 tiles, L2 tiles and CPU-cache tiles (two cpus with 4 cores each), did vectorizing of inner loops, and also unrolled the inner loops (with #pragma unroll). In addition, since we were working with symmetrical matrices, and only needed to calculate half the matrix (along the diagonal), I had to fiddle some with the scheduling for the OpenMP pragma to get the best possible performance out of it.

You got the fastest time simply by playing with the compiler flags? We had a similar problem where we had to do a matrix multiplication on symmetric matrices for C = AB^T+BA^T (rank2k update with alpha=1.0, beta=0.0) and there was nothing the compiler could do for us to get even remotely near good scores. Doing the simplest implementation we got about 5 FLOPS/cycle on an 8 core system, optimizing just with SSE etc, I got it up to about 13 FLOPS/cycle, and by splitting up the matrix in tiny parts to avoid cache trashing etc I was able to get it up to 47 FLOPS/cycle. For comparison Intel's MKL library managed about 85 FLOPS/cycle on the same hardware. I believe the best in my class was about 50 FLOPS/cycle, and it took an insane amount of fiddling for any of us to get above 25-30 FLOPS/cycle or so. That said, most things done on a computer is rarely that limited by memory access, and then the compiler does an awesome job :)