Spring 2005: CPUs, basic badges and memory
AMD: Athlon and his 64 bits... that are not the most important thing
The name (and the design of the nucleus) Athlon has an altar between the fans (pardon, "believers") of AMD. It has been a commercial, technical success and, perhaps more important, of mark recognition. Has he ever wondered how much costs the mark "Pentium"? Since it remains short, sure...
After the Athlon XP, rapid and very effective but not so different from the not-XP, we are at present in the generation K8, that of the Athlon 64, the prophesied one "Hammer" (hammer) of 64 bits that would squash Intel. The truth is that it has not been for so much, but yes that it turns out to be a refreshing breath of fresh air.
To be sincere, the least important of the Athlon 64 is precisely his support (his extensions) of 64 bits: Microsoft does not stop being late with his Windows XP x64 (to all lights finished, but I am missing of drivers optimized, question of cardinal importance), Linux uses very little (deplorable, but true; or at least, little in PCs of office, in servants it is different to sing), NOBODY needs any more than 4 memory GB (good, I not, and you neither), and neither it improves tantísimo that we say the yield with regard to 32 bits.
But the case is that his yield with 32 bits is excellent, he has this support for 64 bits (Intel finishes of adding it, but it is not not much less widespread), there are mikes that warm up very little (useful for overclocking or, simply, to be able to sleep nearly one burning PC or use it in the lounge like silent multimedia breeding animal, only in spite of investing in a good refrigeration), and especially they are cheaper enough than the Intel for an equal or top yield (except in very concrete tasks like the video compression, and even there everything depends on how much we have spent).
The myth of the GHz is increasingly a myth and less reality (enclosed Intel, pressed for not being able to raise them so rapidly as earlier, leaves the numeration based on GHz and passes to "comparative" numerations), and the case is that, to price equality, in almost any task (if he considers the price, even in those of video) the AMD are more rapid than the Intel, despite going to less enough GHz.
My recommendations for AMD would be the following ones:
The one that approaches: mikes of double nucleus!
Very much original PC has spent from 4,77 MHz (yes, 4 eats 77 MegaHz) of the IBM up to almost 3 ó 4 GigaHz of the current mikes. But although both Intel and AMD have realized spectacular prowesses to support the speed increase rhythm, the limits of the process go being reached in silicon; in the same Intel one speaks about a maximum of approximately 20 GHz, but only with radically different designs and, especially, increasing to a much minor rhythm and at the cost of technical solutions hardly approachable for the tremendous investments that they suppose. To happen from 130 nanómetros to 90 has not been by no means a bargain sale, and every time it is difficult to keep on reducing more everything...
For it, and to keep on selling as it is (ah, the free market...), both companies are going to throw the same year (Intel will do it in any moment) his first CPUs with "double nucleus", which to understand us is something like putting two microprocessors in the same pill (technically it is a question of tremendous complexity, in addition with approaches and very different solutions, like major or minor interoperability grades and compartición of resources between the nuclei, conflicts resolution... but to understand us that thing about is enough "2 in 1").
Publicitariamente, can be the whole revolution; for the yield... one will see already. Although in a servant's environment, a mike with double nucleus (or a badge with two mikes) can suppose a yield increase of up to 100 % (in ideal conditions, obviously), the true thing is that most of users do not realize simultaneously any more than one task, and that the programs software must be re-designed to be able to make use of the double nucleus.
In addition, the mikes with double nucleus will have minor speed of clock (less GHz), complicating moreover the situation. It will always be useful to have two nuclei instead of one, or even one "partible in two", as it makes modern 4 the technology Hyper-Threading de los Intel Pentium , but this technology has demonstrated us since the success very much of the type depends on used software and on the work of the programmer, with yield changes between-1 % and +20 % (often 0 % and other times more than 10 %)...: and that without reducing the speed of clock, eye!
Anyway, there approaches a rare situation, full of suspense, even more difficult to evaluate that the current one. The one that waits for us...
To know more on the CPUs...
... Investigate in Internet. In addition to the pages of two manufacturers (www.intel.com and www.amd.com), where he will not need very much English knowledge to locate information as the GHz or the size of cache memory of a mike, I recommend to him to throw a glance to the following articles about the always excellent web sites AnandTech, Tom's Hardware and Ace's Hardware:
... Observe a computer science magazine to spend the filter of the price to the tests of yield of the above mentioned articles and you will be able to decide yourself what he needs and what not. Although if he prefers a little more of theory in Spanish, he keeps on reading to know how there is the topic of the basic badges and the memory.