As the thing was very bad with the theme of its new software and NVIDIA CUDA is going to come to revolutionize the global world but none of the lords of colors so we were well prepared to deal with this style of a talk given our high level but ethyl low level of awareness, we decided that the best thing we could do was follow tocándonos eggs and send to our favorite cat. I do not know is that the big Mixu is going to crack this piece of article.

If I run Crysis CUDA to 120fps at 1920 x 1200 ... If I see the veins CUDA the teats of Lara Croft ... If I CUDA ...

NVIDIA is making its European tour of the presentation of its new technology called CUDA. , For the uninformed in the face of the earth, is the new tool that provides us use very different types of calculations on our GPU and much accelerating application performance adapted. The conference was held in the Spanish Hall of the IFF (Facultat d'Informàtica de Barcelona) and there were walking letter. What are the benefits this technology offers us? Do we help further accelerate the games? Avid readers read on ...

At about 3 pm, after passage through the pipi-room, got to the point where extraction was a lot of people waiting at the door, to which I stayed at the bottom but quickly saw that they were not marujas had registered with time and the last to leave. So I kept the tail of a very targeted and maja lady (of the few that had already known ... fame FIB) after asking for my name gave me a nice "Scholarship Kit, a CD ... we Spam Nvidia for the mundane, and a nice pen of BSC (Barcelona Supercomputing Center, the center that organized the conference).
He then entered the room and I am in a position where they do not place near miss any details. I prepared my bártulos, some known as mepetrés, jafotos chamber, and the inseparable duo famous pen & paper. After a while a man named Nacho (not Vidal, surely there are few users who were being tone) gave a presentation David Kirk, chief scientist of Nvidia since 1997. Once it began accommodated with their PowerPoint (R).

First began to speak in the name of its new technology. CUDA is the name that comes from "Compute Unified Device Architecture", and not other obscenities that you may be thinking. Mr. Kirk perjuró and swore by Gokuh are, for the coverage of its mobile and snoopy that the name is not related to any religion but Pope (jai paaaallooo) propaganda has created the name ...

David explaining why the name chosen, or why not, or why do not ...

Apparently some "polaquito of CUDA here said in Polish significava" miracles. " Which of these clever ... as if NVIDIA did not want to incite people to their products are the result of a miracle! Oh well ... as is the fruit of the hard work they carry.

And then went on to entertain too much comment on the various fields of existing computer and then later to explain what it means an increase in performance. Basically the idea with which you have to quedaros is 2 to 3 times faster is "because I see no difference," from 5 to 10 times is a "good will, whether it is worth the effort, and 100 Sometimes it is a "see it with a new world of multiple orgasms unimaginable." Once we are clear that we temita
discusses the evolution of graphics cards from history and how they have evolved to be programmed to go 100% and in conjunction with the CPU time.

Then took the step to show the benefits of using its technology in various fields of computing applications. In an application of biological computing get a performance boost of about 240 times the original, in a simulation of the neural circuits get a 130x, a research astrophysicist is obtained over a 300x and MATLAB (Propramme that hates all telecommunications and if someone does not preguntádselo ...) was a 17x. Come on, painting is the invention of the century (computerized talking ... this year will be), or maybe a miracle ... CUDA CUDA CUDA!

Once completed to show what its results are beautiful (not going to say it's bullshit, it is clear ...) we went on to explain the programming model of this new technology and the differences of using it against a CPU or GPGPU ( using graphical programming API's, or in other words, do calculations using the GPU using DirectX or OpengGL).

Mr. Kirk explain the differences in different methods of calculation

Mainly of a CPU load current, as your latest Intel Core Duo Octo Quetecagasderapidoqueva, you need a string of sequential instructions to go running and jumping between them to withstand, according to the conditions are provided. On the other hand GPU has a calculation that many units can be used in parallel and are specialized in making a kind of calculations. If the type of calculations that needs to be done need not be sequential, used to use the card to process the whole lot of calculations pull in several groups, and also that the chart is prepared to make these calculations, because we get we have made the same calculations in much less time.

Then use the chart to make these calculations, right? But we also have this other technology called GPGPU, but why not CUDA and GPGPU? Well, the explanation is that you save many overheads CUDA processing using the API's for 3D graphics, is programmable in C and need not have knowledge of 3D graphics programming, the execution threads can share data and synchronize, and can read and write in any position in the memory as much as you like.

Now is when anyone comes and tells me ... but what are you counting? Let's talk me into ejpañó! And I explain it with apples and oranges ... Well ... then we will explain it for those who are not involved in world hardware.

Suppose we are going to buy from a fruit, we take our bags with different kilos of apples, oranges, pears, grapefruit, melons, turnips and cabbages (viciosetesssss;)). So if we are educated people and not going against society in general you're addicted to you andalusia although stock exchanges, set the price, and join me so that the total pay and leave through the door without thy jamonero behind with a knife. Images, ladies and any other lady / ita, which is dependent on a CPU lifetime, he himself should take the first bag, weigh, mark the amount per pound of product type, take the result of what goes the bag, save it ... then take the bag and another, then another bag, and another, and others, and joy (to another dog pilot uproar!) another bag! And in the end would make the sum of all the fruit truck that you take home, which tiraría a long time until you finally home to eat half a kilo of pears come hungry because you've been waiting 80 years to get copper.

Now imagine that on the other hand the clerk tienea a string of chinorris without a contract and underpaid (as is customary in this society ...) who are experts in spite bags of fruit and calculate the amount and add it all juntitos. Now is when you go with all your bags and said the retailer will give its Chinitos that unless Paris Hilton gets off your panties and have to pay the amount and the clerk is waiting to pay. The latter is the case of using CUDA and your group would chinorris GPU installed on your computer.

After all this happened to Mr nvidioso give a few examples of how easy it was to modify and / or create an application using CUDA to benefit from the power of our graphics card. Once concluded the presentation go to the round of questions, which will not comment because most were about scheduling problems and concerns of the implementation architecture of the model underlying CUDA, which I doubt that you are interested and as I have not rolled and ...

Mateo Valero giving the gift to David Kirk.

To end the conference Mr Mateo Valero, director of the BSC, made him a little present for David remembered his tenure in Barcelona. The gift, if not more, as the boy who likes graphics and such, was a book on the views of Barcelona.

And running out ... we have seen new technology and its improvements, which tests have very good looking and much improved performance. In relation to video games is very possible that this will help us to set a fairly widespread use of GPU's to use as well for accelerator physics, AI, graphics, and other calculations, but everything will depend on the programmers behind the game if you plan to take advantage of GPU'so not.

Mixu

PS: If anyone is interested in the audio chat, read it and do magic.