Intel Pentium 4...: version 1.0?
The chipset...: and the memory!
We enter one of the most polemic topics with regard to the new Intel mike: for many months, it will only be free for memory RDRAM (Rambus). Or, more correctly: for many months there will only be available basic badges with the chipset Intel 850, who only supports memory RDRAM.
And what does this memory have of villain? (If really he wonders it, the fact is that it does not usually follow us). well, it has two problems: that it is not a free standard (it is necessary to pay royalties to his creator, Rambus Inc). and his high price (for the commented earlier and because it is difficult to make).
It would be much more desirable to use memory DDR-SDRAM, but Intel is tied of hands by a contract with Rambus that it prevents; he has already declared that the whole year repents, that will make chipsets for DDR as soon as the above mentioned contract expires and that will give licenses to other manufacturers to design the suitable chipsets... but that will take a lot of time, probably.
In honor to the truth, the Rambus has begun shining with this chip, although thanks to a small trick: Intel has used TWO channels of memory RDRAM, reaching 3,2 GB/s of theoretical maximum transference (1,6 GB/s for channel).
This is more enough than 2,1 GB/s of the DDR, or wretched 1 GB/s of the PC133; also, Intel "gives" two modules of 64 MB of memory RDRAM with every set of badge and mike... But there is nothing free in the life, and less in the world of the hardware: to the being two channels: the grooves of two must deal in two with identical modules!!.
This complicates seriously the possibilities of expansion and makes the Intel offer almost useless, so if this mike is useful in some ambience it is in the professional, where they usually use 512 MB or more. As the basic badges only have 4 grooves RIMM, the only solution would be to resort to the modules of 256 MB RDRAM , which price is COMPLETELY PROHIBITIVE.
And in addition, not the whole RDRAM is equal, not much less. The only one that really is worth it is the RDRAM PC800; but enough people are selling teams with memory RDRAM PC600, which yield is 50 % minor (even with two channels, scarcely 2,1 GB/s it reaches of a lot of more economic DDR with a channel). Anyway...
Architecture, design... where they are good. Since in this case, there is some graph... but certainly I miss.
The first one is an Intel courteousness; to be exact, we have joined the information of two pages that Intel has been in charge of keeping well separated, because they do not stop in good place to the new microprocessor.
Impressed? And if I say to him that we are speaking about the yield in Word, Excel, PowerPoint, CorelDraw, Netscape Communicator, Photoshop...? It seems clear that the new architecture of the Pentium 4 has some problem, very probably for the longest pipeline of 20 stages.
But good, the thing can improve a little if we pass to the professional operating system (and more expensive) of Microsoft, the Windows 2000:
(Ah, for the suspicious ones: the configuration of the computers is IDENTICAL up to where it it can be (a luck, the truth). Only they differ in the mike, the motherboard and the memory...: and Pentium III do not even use memory DDR, only PC133!!)
We might deduce that the Pentium 4 is not thought to improve the yield in this class of "simple (?)" tasks; and in multimedia? Well, let's see how one gives him the MPEG2 (the format of compression of video of the DVD):
Very well, much better, we begin noticing the force of the GHz, the 400 MHz bus and the memory RDRAM.
But apart from observing that it is in the multimedia applications of audio and video, renderizados and games 3D, where the Pentium 4 can be worth it , a simple division indicates us that the design is much less powerful than that of Pentium III as regards the MHz (or GHz):
Namely that a future Pentium III of 1,2 GHz would be more rapid than the Pentium 4 of 1,5 GHz in the same task.
... And an Athlon of 1,2 GHz with memory DDR-SDRAM would be MUCH ahead of both, and in this case we do not speak about a future mike, but about a reality that is on the market.