osmala

Speed itself is killer app. Tenth of a second is the limit to aim for responsiveness. Now we have acceptable speed for surfing but it quite often doesn't look like instant.

Going from ADSL to ethernet reduces latency by order of magnitude for the connection to local internet. That is the big jump. And that latency reduction helps gamers too. This is the big advantage of going for Fiber instead of lets try how we can put as much bandwidth as possible through this copper never designer for it.

As for bandwidth, lets consider the other side of story where you get those bits from, they too have limited bandwidth and have to support maybe thousands of simultaneous users with it.

And once they found the killer app for the 1Gbit internet they are not going to recognize that. As bandwidth is only visible when there seems to be too little of it, not when there is enough of it, and those who have too little of it often just need to wait some additional seconds and never realize that it could be almost instant, and totally miss the point of having additional bandwidth. Only way to recognize the killer app for this is use with Gbit fiber and then go with friends house and use it with their ADSL and realize you cannot stand how slow/limited it is.

Once you start replacing last mile with Fiber there is no point trying to save few bucks in an expensive project and have order of magnitude less bandwidth than what they reasonably can offer.

I'd pass this one. Resolution is nice. However its surface area doesn't do justice for that number of pixels. I'm having 30" 2560x1600 monitor, and any upgrade from that needs have both bigger physical size AND more pixels. Improving one without the other isn't good enough. Yes more pixels give nicer fonts and better image quality, but bigger area is needed to actually use it effectively. When considering down sides of showing it as two separate displays on Linux I'd say "These are not the monitors you are looking for."

Problem with nuclear is that they cannot operate in situation where there is other players which destabilizes the network and are profitable with subsidies even if they sell at negative price. Nuclear is long term base load you just cannot shutdown a nuclear plant in windy situations, and expect it up in next hour and save anything by doing it. Solar and wind is more expensive if you take away all subsidies and force everyone to handle the smoothing their electricity production to grid needs. But force buying in short term spot prices and giving more subsidy than cost of production per kw/hour for intermittent sources is actually terrible way of handling the grid, it increases greenhouse emissions by forcing to use low efficiency gas plants when sun doesn't shine, (not high efficiency since you cannot bring them online fast enough).

Nuclear gets loan guaranties which are not free money, and regulatory fees which are additional tax for nuclear industry.

Each language has its difficulties. Fully as fully fonetic language learning how to read Finnish atloud should be far easier than in some other languages who's speakers actually have spelling competitions for kids.

Hard part of English complex vocabulary compared to Finnish, as we usually derive things from existing vocabulary instead of inventing new except for loan words that normally are similar to English but more Finnish style.

Kirja=book. Kirjain=letter as in alphabet. Kirje=letter as in mailbox. kirjoitus=writing Kirjasto=library Kirjoittaa=write Kirjoitettu=written Kirjata=to register [as write to register] Kirjaamo= registry office. Kirjoituskone="writing machine"= type writer

Actually we have been spending a lot time with Amdahl's law already. Software people assume it only applies to parallerizing software. In "computer organization & design" By Patterson and Hennessy it is stated in the general form. And example is used how much multiply unit should be sped up to get five fold improvement in execution time when 80% of time is spend in multiplication?

Execution time after improved=

(execution time affected by improvement / amount of improved)+Execution time unaffected.

I've seen that working part time is enormous reduction in efficiency of personal projects and commute and task switch takes equal amount of time anyway, and for some its almost equal of not doing the other project at all. So one recommendation is try to get a 1-3 month full time project, with good enough salary to save your living expenses for entire year since humans always underestimate the time it takes to finish their project, and you are a human too.

Its relatively punishing also to be poor and follow your passion. Then the opposing end is also punishing to be I've heard. And then some where in the middle there are people who envy with either end of spectrum and they think they would be happy with the end they see while telling them self that they cannot be happy before getting the benefits of being at the point in the spectrum they envy, while not realizing that they are in a nice point in the spectrum of trade offs. People want what they don't have and ignore the trade offs the others had to pay for it.

The article perpetrates The myth of any working hour is equal. Actually its energy which limits the amount of work person does, and if we add more hours the same amount of work gets done spread on more hours, of course there are exceptions to this rule like where you just sit and wait and do some standard routine when someone comes like selling tickets when number of customers is low enough that your working speed doesn't matter.

The reason we have 40 hour workweek is that industrialists realized it maximizes the productivity of physical labor, and its studied quite through out. In mental labor the studies say that maximizing long term productivity is even lower than that. You get more done for a week if you add additional hours for a week, however within week or two your productivity dips so that you produce LESS than when you were working 40 hour week. If we add commute to 40 hour workweek, and count the stuff people must do to keep their self in working condition and not too smelly, there is less than 40 hours of leisure time. I just did Excel out of what I want to do in a week and what I MUST do (like eat, sleep...) Then end result was 187 hours a week, unfortunately for me there is only 168 hours in a week.

Comment to claim in that large scale democracy automatically becomes more partisan.

In large scale democracy partisanship is not necessity. American system has some specific methods to cause it, compared to Finland where it isn't really a problem. And its really these TWO issues that make the american politics more partisan than rest of the world.

A) Election system is strongly geared towards creating two party. When new party has gained enough support it replaces one of the previous two parties at the control, and there is new TWO parties which compete each other. Compare that to systems in which allows smaller parties to get seats. For examble in Finland there are 3-4 big parties that total 60-70% of seats and rest goes to smaller parties. That means to pass anything you pretty much have to have two big parties agreeing on the matter and either third big agreeing or some small parties agreeing.

B) Executive branch is Winner takes all. In Finland the parliament(legislative branch) selects Executive branch and because no party has half or near half seats there is need to be enough parties agreeing on the matter. So if you are REALLY uncompromising you NEVER get to be part of executive branch EVEN if you become the biggest party of elections, since you didn't get more than half the seats anyway in multiparty system. The biggest party gets first attempt to create the cabinet, but they still need to get enough others onboard. The cabinet pretty much needs majority of legislative branch to support it. They need a program which over half the legislative branch supports and usually give seats in the cabinet to enough other parties that have over 50% of parliament. If biggest party fails to create agreement of the program the responsibility of trying to form the cabinet goes to next largest party. And if all else fails parliament may agree a minority government OR even putting some professional bureaucrats instead of politicians in top positions of executive branch. Both are extremely rare situations in here.

a) Fructose is only metabolized by Liver, and not brain nor muscle. The liver converts it to energy form like bad cholesterol. In small quantities that exist in fruits with the fiber limits the speed of absorption so it doesn't overwhelm the livers primary mechanism for dealing with it. But moderation of sugar is impossible without actually spending time to find how much sugar there is in different products, and cutting those that come with mercilessly until you consume 15 grams of it a day. Its about 1/3rd of smallest coca cola bottle here or single serving of yogurt. I remember some study that majority of food items in market have added sugar in USA so moderating it is extremely hard without spending time to think about it.

Also sugar messes up the system that tells the brain that you are no longer hungry, which means if you eat stuff that has sugar in it, it takes longer for you to feel full.

Also while there is health reason to exercise burning more calories isn't it. Single snack is often more than hour of exercise.

You should watch "sugar bitter truth" lecture by dr. Lustig on Youtube. www.youtube.com/watch?v=dBnniua6-oM

Here's another point, how much clockspeed suffers from adding this feature? What about instruction latencies? Does it increase the pipeline length for bigger branch missprediction penalty? Does it interfere with microarchitecture optimizations we are using or could be using? All these are performance questions that may result from adding this instruction. There is reason why RISC eliminated most of mixing control flow with arithmetic.

As for his numbers on the performance penalty of doing it in software it surely depends on micro-architecture. All intel x86 processors before HASWELL did only ONE branch per cycle, then in haswell they increased that to two. So the cost of doing it in software should of went down recently. And Intel is adding new instructions to speed up arbitrary precision arithmetics in broadwell so they probably have already thought about this and decided its not worth the cost in hardware.

The Mother of All Games.

This kind of paper probably gets lots of money from EU and government sources. It has couple of people from different EU countries and it has people from different fields co-operating.

This is probably something that people think lets do this and see what happens at least we get maximum government funding for being interdisciplinary and co-operating between institutions. Results are probably not bullshit at all, but idea to study this is probably because its easier to get funding is for interdisciplinary co-operation.

This is public service that is subsidized by goverment and its not profitable and its goal is to help elderly get around in the city. It has some additional stops beyond normal bus stops, and it has far more limited service area than normal busses in the Capital. For instance the air port is outside the service area. Its also a nice complementary service for public transport on uncommon trips that would take too much time to change between 3 different busses or more.

6 wheels out of which 4 are driving wheels.

The prefetch will always take same amount of data no matter what the size of python object in question is. That leads two results. 1) You can miss if your cacheline size is less than python object size. 2) You miss if your python object is not alligned at the cacheline and comparison is with the part that is in the next cacheline side.

If these two condition apply then there is limit of 2 on speed up and increasing python object size reduces the effect of prefetch which still helps.

Another about prefetch distance I'd pick 5 based on his testing data. 1) It has no downside on current tests. 2) Due to prefetch constant size any prefetch that hurts because you pick distance of 5 instead of 3 is unlikely to have any meaningfull effect at any prefetch distance. 3) It allows some improvement in comparison speed. Like if there where JIT on python side to specialize the test. 4) It allows a nice factor of improvement on CORE execution speed relative to memory latency. 5) It allows you to handle situations of page misses better.

What a laugh I got when I read their description of what their business is compared to what their brand name is.

http://translate.google.com/#fi/en/ansa