Strategic thinking is of vital importance to business, but especially so in the hardware market where tremendous capital investments have to be made years in advance of full production. Intel is a company that has been criticized repeatedly for not always making the right strategic decisions in the past two years. Its approach was not bad, but the market did not develop according to the expectations of the marketing strategists. For over two years, Intel tried to convince almost all PC manufacturers of the efficacy of Rambus technology, but the result was that after Dell & Co initially sold such systems, they virtually disappeared from the market to be replaced by better price performance SKUs (stock keeping units). Rambus did not become widely adopted in PC systems because it was too pricey, and so it ended up being what many enthusiasts and technology lovers considered a mega-flop.

Looking back, it's clear that Intel introduced this new memory technology to the market much too soon - and ultimately ended up introducing DDR-SDRAM support , reluctantly, as an afterthought almost. Originally, after wrapping up the 440BX chipsets in 1999, Intel had planned to file away SDRAM technology and move to its higher price segments. If it had been up to Intel, it would have also meant the end of PC100 (100 MHz memory clock with SDRAM).

After that, PC133 established itself - Intel was powerless and felt forced to launch the 815 chipset. Everyone now knows what happened: the memory interface of the Intel 815 was slower at FSB 100 MHz clock speed than its 440BX predecessor. The reason for this was pure politics: ultimately, a respectable distance had to be kept from the 820 and 840 Rambus foundations. This was confirmed for us two years ago through numerous benchmarks and reviews . As DDR SDRAM memory established itself on the market, Intel held itself back. At best, all Intel could do was to sell its Pentium 4 systems based on the 845 chipset together with SDRAM memory - even though the performance of such systems turn out to be pretty weak.

Nevertheless, Intel succeeded in feeding large quantities of the P4 to the masses, the corporate market and selling them over-the-counter at discount chain stores. It wasn't until the end of last year that the P4 with DDR SDRAM was available from Intel directly, due to the fact that a contractual agreement with Rambus had expired. Intel had extricated itself from this contract, which prohibited the chipset support of DDR SDRAM until the specified date. But, here's what we want to talk about, beyond the politics, it must be said that Intel's flagship, the Pentium 4, only performs at full potential only when used in conjunction with Rambus memory.

Depending on the chipset, there's a 10% to 15% difference compared to DDR SDRAM. In view of the high processing power, many users are certain to ask themselves if one can really cope with the lesser performance of DDR SDRAM. However, the basic problem can be recognized by a more thorough analysis of the tests from Clash Of The OC'd Titans: Athlon XP 2300+ vs. Pentium 4/3000 : the overclocked Pentium 4/3000 is a good example of how the combination with DDR memory only offers a marginal increase in performance compared to the lastest top-of-the line model, P4/2200. In any case, we had expected a much faster performance in the individual tests, especially since the FSB and memory clocks were running at 138 MHz. By contrast, a further attempt with RDRAM showed how a Pentium 4/2600 can perform at its best. This CPU overtook the Pentium 4/3000 (based on the 845 chipset with DDR SDRAM) in a few of the benchmarks.