Intel Pentium 4...: version 1.0?
Yield (2): SSE2 and optimizations
Unfortunately, Intel scarcely provides to us more information of utility, the only one laaarga series of tests without practical utility or on the yield in Internet (as if that had a lot to do with the microprocessor...). Very well, at least it gives us a few tests of Quake3... but slightly significant:
The "strange" question is that a resolution has been used of: 640x480 points with 256 colors!! Let's see: who in his healthy judgment uses a GeForce2 GTS with this configuration? And the fact of the Athlon, extracted of the page of AMD, also is to this resolution and with a GeForce2 GTS...
It is easy to explain: in both cases it has been done only to emphasize the difference between the mikes, since to 1024x768 ó 1280x1024 the yield difference would be almost indistinguishable, enclosed with this excellent graphic card (as they demonstrate in AnandTech, it would be I half-close 1 % or less).
In any case, this fact is relatively positive for the Pentium 4, although the Athlon is supported close thanks to the DDR and his new bus to 266 MHz (because if the Pentium 4 puts itself ahead it is only for two channels of RDRAM and the bus to 400 MHz, not precisely for his internal design).
And of this test we go to the most polemic matter on the yield of the Pentium 4: the hard mathematical applications. As we said in the corresponding paragraph of the architecture, with the classic method of handling these operations (the unit FPU), the yield of the Pentium 4 is HORRIFYING.
For example, Tom's Hardware (that is immersed in this polemic up to the neck) emphasizes that the same renderizado in 3D Studio Max 2 can take THE DOUBLE in a Pentium 4 to 1,4 GHz that in an Athlon of 1,2 GHz with DDR (and also more than in a Pentium III); and something similar happens in multimedia applications like video compression to MPEG4...
... UNTIL THE APPLICATION IS OPTIMIZED FOR SSE2. The question is that the Intel engineers have bet EVERYTHING on the yield with applications SSE2, neglecting the unit FPU. This does that in applications optimized (practically none of frequent use... for the time being) the yield rises very much, simultaneously that in many of the not optimized ones is almost ridiculous.
A very strong bet, it is me who would say; because it is possible that in one year many applications exist optimized (that it will be necessary to buy, skylight), but while a very slow mike will be had. And with a year of age a mike is old, be the one that is...
Some prices orientativos
Making the yield separate, the buy or not of any computer is a question if it turns out to be economically profitable. We are going to exhibit some information, and you will decide yourself if the Pentium 4 is worth it in this respect...
It is a question of average information obtained in PriceWatch. As comparative fact for the question of the memory, Crucial sells modules of 128 MB DDR-SDRAM for 129 $... although on the other hand the modules of 256 MB DDR-SDRAM are not easy to find still.
As for PCs finished, Dell, IBM, HP and Gateway have models from 1.999 $, but eye: normally to 1,4 GHz, only with 128 MB and without monitor.
Conclusion: version 1.0?
We are not going to walk ourselves for the branches, our conclusion is clear: presently, DO NOT BUY IT. NO. Not even madman, we go.
Motives? First, that it remains clear that WE ARE NOT ANTI-INTEL (although many people do not believe it, I write this from a PC equipped with a Pentium III, and very to taste). Simply:
And an incompatible mike with the new badges, and a memory that might be replaced (or not, already we will see) with the DDR-SDRAM... Too much risk for my taste, I prefer an Athlon with DDR or even a trustworthy Pentium III.
Of course, the Pentium 4 seems to have a lot of future. The SSE2 works and undoubtedly they will be implanted, the new design will allow to increase the GHz at big speed (2 GHz is foreseen at the end of 2001), in his moment it will appear in dual version, and it is still possible to reduce his consumption and to increase his useful life with a step to 0,13 microns. Brilliant future... that of the version 2.0.