The problem of the chipset

Despite all his advantages, these mikes present an almost incredible problem: the chipset that was foreseen that was handling them, the Intel i820, takes almost two months of delay with regard to the due date of issue, and it is possible that it is still late another month more.

This would not have supposed any more than a small disadvantage of talking each other of other mikes (like original Pentium III), since the previous chipsets might be used; nevertheless, the speed of bus of 133 MHz that Pentium III EB incorporate and B it turns out to be almost impossible to use with the current chipsets BX or ZX, since it would imply a bus AGP working to 88 MHz (with a multiplier 2/3), 33 % over the normal speed, something very difficult to support for most of the graphic cards.

An engagement solution would be to use cards PCI, which they might use a multiplier 1/4 to work at the 33 MHz normal speed, or 1/3 to go to 44 MHz quite bearable (overclocking), but at present scarcely good graphic cards exist in PCI (excepting the Voodoo3).

In any case, another four solutions exist:

• to use badges based on the chipset i810E;
• or in the chipset ROUTE Apollo Pro 133;
• or in i840;
• or to use only Pentium III type E (without 133 MHz bus).

The first solution (chipset i810E) turns out to be aberrant for the sinsentido who supposes installing a microprocessor of the highest scale (with a price of more than 50.000 pts) in a motherboard based on a chipset destined for the average - low segment, with an integrated graphic card of very scarce yield and without groove AGP to update it. It would go out more economic to buy a rapid Celeron and to install in an economic badge BX together with a card 3D luxurious type type TNT2 Ultra, and the yield would be even better.

As for the second solution, it is not that it is bad: the fact is that it turns out to be fatal for Intel. For those who do not know it, ROUTE is one of the biggest rivals of Intel in the chipsets manufacture. In fact till now it had little in what to compete, was centred almost exclusively on the badges for the family of mikes AMD K6-2/III, since Intel was selling almost 95 % of the chipsets for Celeron, Pentium II and Pentium III... but the case is that they yes have a chipset prepared for new Pentium III, "Apollo Pro 133".

This chipset includes support for all the principal characteristics of i820, but instead of supporting the memory RDRAM there uses it much more sale PC133 (SDRAM to 133 MHz) or the PC100; considering the yield more that acceptable of the PC133 and the highest price of the RDRAM, it is not possible to consider a disadvantage.

The third option is not such, since the chipset i840 is thought for servants, as what it has many unnecessary characteristics in teams of dessert that get dearer his price up to turning it unacceptable like solution.

And finally, one of the solutions that more please Intel (at least until it appears i820): to use only the mikes type E (nucleus Coppermine but bus of 100 Mhz), whose yield is very similar to that of the EB, in badges BX or ZX. Let's remember that these mikes exist with 700, 650, 600, 550 and 500 MHz speeds.

The problem of the memory

If the delay of the chipset was small i820, also we have the added problem of that this chipset does not support of native form another memory that is not the RDRAM (Rambus DRAM).

The memory RDRAM is a type of very rapid memory (up to 400 Mhz with double band equivalent to 800 Mhz) sponsored by Intel, which according to all the indications offers an increase of the rather small yield, although it depends much of the application in particular.

In any case, the criticism does not come to him so much because he does not suppose a revolution as for yield, but because it is the most expensive, much more expensive than the memory PC133. Also, it turns out to be difficult to make: the problems of Intel with his chipset i820 owe precisely to the use of this memory.

And in addition, the chipset i820 is not foreseen that supports in any case the memory PC133, but only the PC100 (for commercial motives, that not technical staff), for what the only cheap yield alternative more that acceptable is again the chipset Apollo Pro 133 of ROUTE. Sincerely, it seems that Intel is pawned in benefiting his competitor...