Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×

Comment Re:Folks, have your license and registration ready (Score 5, Insightful) 293

Is that a reasonable assumption? For the first couple of years I lived in the US I don't think it was obvious to me that I should even stay in the car when pulled over. When if my wallet was in my bag in the back of the car? No reason to take it into the front just in case I'm pulled over, after all, especially when you come from a part of the world that has no requirement to carry identification in the car in the first place. Certainly back in the UK I wouldn't assume I couldn't leave the car. This peculiar interaction where the driver has to follow careful rules that are only practically spread through word of mouth and watching TV shows, just in case the cop gets nervous, isn't really optimal, and I don't think assuming that "everyone knows" is a reasonable view of the world..

Comment Re:This is BS (Score 1) 440

That ignores the fact that some traffic controls work specifically to keep traffic moving. Look at variable speed limits for example - the entire goal of the control, and the benefit of having people obey it, is that traffic keeps moving by lowering the speed and removing the stop-start behaviour or racing towards other cars.

In this case the OP wasn't even talking about traffic controls, rather that truck trailers should be fitted with guards that would stop cars slipping underneath. They absolutely should have those fitted.

Comment Re:Of course ... (Score 1) 315

I don't know if that is a perfect comparison. I have been wondering for a long time why liability cover on US car insurance policies is so low compared with UK liability cover. Default cover at a pathetic 100,000USD instead of the 8,000,000GBP that my last UK policy had.

I recently spent some time reading web sites suggesting how to assess how much cover you need and they suggest it based on the value of your assets. So it strikes me that while UK auto insurance is designed to have a high enough liability cover amount to make sure that those you harm don't lose out (within reason, clearly a person who loses a leg loses out, but the value should at least cover their medical bills and given them some compensation given the loss of income), US insurance cover is intended to cover you as a driver against the risk of losing your assets to legal action on the part of those you harm.

So although clearly the payout on auto insurance does go to others, to then cover their costs while health insurance directly covers your costs, the actual goal of the insurance strikes me as protecting yourself in both cases - at least in the US.

Comment Re:So what if the world sees it? (Score 1) 174

The problem there surely is the "subset of the content" aspect. Currently the only TV we watch in the US is iplayer - and most of that is documentaries that the BBC still does much better than the American channels do. If the subset of the content is just major new series, I wouldn't pay for it. I would be very happy to pay a subscription to all content - and even better if, like CBS all access and other options, it then includes a back catalogue so you don't have to catch a programme within 30 days of broadcast.

Comment Re:Those Anti-Science Liberals. (Score 1) 414

Isn't that a statement of the obvious? Most drugs are not beneficial for most conditions - clearly not, they all tend to be good for specific conditions. So not doing anything, for any given condition, is better once side effects are taken into account than almost all drugs. For any self-limiting condition for which drugs offer no real benefit, a common cold, most back pain, imaginary diseases and so on, which are common reasons to visit a doctor, then not doing anything is the best treatment.

Once we realise that, placebos fall out trivially depending on the answer to one very important question: is it better to lie to someone and say they are getting a treatment, knowing it does nothing, or better to tell them their condition is self-limiting and it is best to give them no treatment?

Comment Re: Title condradicts summary (Score 2) 144

No it really doesn't. Just because NVIDIA's marketing department calls something a core does not make it comparable to a CPU core - notice that even AMD's marketing is being more up front about this by talking about an APU having some number of "Compute units" which is the sum of the CPU cores and the GPU CUs.

Even if the GPU did have thousands of cores, though, that does not affect the numbers I discussed. Instead of having 32 cores with 64 ops per cycle you just get 32*64 cores with 1 op per cycle. You still only have a factor of 10 on ALU throughput.

Comment Re: Title condradicts summary (Score 1, Informative) 144

I honestly thought that we'd got away from this 500x nonsense a few years ago. I would suggest that AMD is one source for the information that 2-3 is more reasonable. AMD, Qualcomm, Khronos, any of the members of the OpenCL committee you talk to as well as NVIDIA insiders if you catch them at a conference. I gave multiple public talks countering any factors over about 10 when I worked at AMD, which were approved by management.

Just think the raw numbers through. The GPU has, say, 32 cores. The CPU ALSO has multiple cores. Don't count them, then you're cheating. So let's say we have 8 CPU cores there. Each CPU core has two SSE units or one AVX unit, to be conservative. That core is doing 8 ALU ops per cycle per core. So you have 64 ops per cycle. The clock rate is 3x the GPU so let's call it 196 ops per GPU cycle. The GPU had 32 cores. Each GPU core can do 64 ops/cycle (fair number for GCN). So you have 2048 ops/cycle on the GPU. 2048/196 is roughly 10. That's your peak - now you add in divergence costs on the wide GPU SIMD units (which statistically will hit you much earlier than with the CPU's narrow SIMD units), count the tiny GPU caches leading to more cache misses than the CPU and you can see why that factor of 10 invariably drops to 2 or 3x.

More honestly you're looking at a factor of 10 or so for ALU throughput, and 10 or so for memory throughput - and those are not multiplicative. In real use cases 2-3 is about right when comparing against well-optimised CPU code.

If there is a 500x speedup appearing with Libreoffice here, and the likelihood is that that is somewhat cherrypicked anyway, then what we are seeing is the difference between someone optimizing code and someone else not doing so. There is every reason to think the original code was only lightly optimized, not parallel, not vectorized or some set of the above.

Comment Re:Poor man's limo service (Score 1) 193

Try the same thing at Heathrow airport. You can take a metered taxi from the airport, but you can also request a private hire vehicle from the airport or (as I just did) book a car in advance to meet you in arrivals. In the US you might call that a limo service, but that usually implies a certain higher standard of car. In London we'd still call them taxis or cabs. We're talking about services that are significantly cheaper than private hire and cheaper than Uber as well based on the numbers I looked at.

Uber is little different from such a taxi dispatcher, the only real differences are that Uber is metered, and that Uber uses an app as opposed to grabbing a dispatcher phone number from the hotel lobby. If anything that metering and the roaming around waiting for the system to find you a job make Uber much more like a black taxi than like private hire in London.

Comment Re:Plain ol' C might a better option (Score 1) 296

std::vector a {1, 2, 3,4, 5};
for(int &v : a) {...}

Uses a reference, short and clean. Before C++11 the STL was a little copy constructor heavy, but only if you use it that way. If you use large objects, make your vector a vector of unique_ptr. It cleans up the data automatically when you delete the vector. On top of that, move construction/assignment should remove the majority of the cost of copying objects around in STL containers.

An STL implementation presumably could fairly easily optimise a vector> to a single type erased implementation, as long as the code that called delete had the type available. In the end you're trading code growth against pointer indirection and most evidence I've seen suggests that code growth is the smaller issue in the majority of code bases. I'm sure there are exceptions though.

Comment Re:Plain ol' C might a better option (Score 1) 296

It's more the association of destructors with objects that makes C++ different. Ensuring that a particular function is called when an object is destroyed needs compiler support. RAII is a huge benefit of C++ over C. Although admittedly it is most important in the context of exceptions, which are themselves a C++ feature.

Slashdot Top Deals

Economics is extremely useful as a form of employment for economists. -- John Kenneth Galbraith

Working...