- The current situation

- The future of the memory: PC133, RDRAM, DDR...?

- i810: a too economic (?) chipset

- i820: a chipset (and a mike) late

- You tackle for AMD K7 Athlon

- The ROUTE chipset for Pentium III

- Conclusions

Released page

Juan Herrerías Rey

What is... the motherboard?

What is... the chipset?

What is... the memory RAM?

Pentium III: an immature mike?

principal page
what is...
to update
Contact Us
forum and chat



The distribution of the present pages is prohibited without the express assent of the authors.

All the marks are registered by his respective owners.

For more legal details, pulsate in the following icon


Hardware for the summer - autumn 1999


The future of the memory: PC133, RDRAM, DDR...?

From the arrival of Pentium II and his derivatives (Celeron, Celeron Mendocino and Pentium III), the memory "reigns" for his use like principal memory or "memory RAM" has been the SDRAM. Of this one fundamentally two versions have existed, SDRAM to 66 MHz (sometimes call PC66) and SDRAM to 100 MHz or PC100 (to tell truth, "PC100" is a norm of Intel that defines the characteristics that should have the 100 MHz SDRAM to work correctly... although unfortunately many of the memoirs that sell as PC100 they do not fulfill this norm to 100 %).

The modules PC133 are identical to the PC100 like this one

The fact of qualifying to the memoirs SDRAM with a number of megahertzes (MHz) is due to the fact that these memoirs work at the same speed as the bus of the mike (SDRAM = Synchronous DRAM) that is to say, to this 66 ó 100 MHz. This offers diverse advantages as for yield, although it complicates the securing of memoirs of high speed, therefore some of them saw difficult to overcome 100 MHz. But the technology almost always overcomes the forecasts, and we already have available memoirs SDRAM to 133 MHz or PC133.

Nevertheless, between these fortune-tellers who were distrusting the PC133 it is neither more nor less than the giant Intel in person, who prefers a different solution called RDRAM or DRDRAM (Direct Rambus DRAM). In normal conditions, Intel would have left with his and the PC133 would not have been successful (as it happened for example with the BEDO - with B, not the EDO-); but weight motives against the RDRAM have prevented this:

  1. The RDRAM developed like a proprietary design (for which they must be paid royalties), and not like a free standard. Property of whom? Of Rambus Inc... and of Intel.
  2. The previous point turns out to be aggravated by the fact that the GIANT Intel controls most of the market of CPUs and chipsets, apart from taking strong investments in other components as finished basic badges. The manufacturers are nervous for this (they are right in it being...) and do not wish Intel to gain all the battles.
  3. The memory RDRAM is very difficult to make, between others motives because it works at speeds of up to 400 MHz. Many of the companies that at present control the business of the memory RAM have found problems to overcome 300 MHz.
  4. In addition, the yield does not seem to overcome for much to that of the PC133, at least with the current generation of microprocessors (included Pentium III) either; this is due, between other causes, to the fact that the breadth of his bus of information is only of 16 bits, opposite to 64 bits of the SDRAM.
  5. The new mike of AMD, the K7 or Athlon, bet for the PC133, at least for the time being. A few years ago this would not have had the minor importance, but it turns out that presently it is the mike for more rapid PC of the planet...
  6. The memory RDRAM is very expensive, condenadamente face.
As all the cynics will have foreseen (we will have, I include myself), the most important factor of all the previous ones is not different that the last one: the money. Although it turns out to be almost impossible to buy memory RDRAM (why to buy anything that with the current chipsets still cannot be used?), his price is supposed more of 5 times more expensive than that of the PC100, while the PC133 scarcely costs a little more than the PC100 (as it happened with the step of PC66 to PC100). Logically, nobody is ready to pay 50.000 pts for 64 MB of memory, for rapid that is.

Nevertheless, being sincere we must give him to the RDRAM a certain recognition; with all his defects, the true thing is that it offers up to 1,6 GB/s of breadth of bus (16 bits x 400 MHz x Double use of every cycle of clock / 8 bits for byte), opposite to "only 1,06 GB/s of the PC133 (64 bits x 133 MHz / 8 bits for byte). The future mikes, with 133 MHz buses or more, combined with the AGP to 4x, they can benefit something of the extra band... or not.

Modules RIMM

Not even in Intel they see clear what type of memory is more adapted for the current moment (good, they would like that it was RDRAM... but there are almost the only ones). A pair of times changes opinion at least a month, although in the last week some movements in Intel seem to indicate that they accept the reality and they will use PC133, at least for the time being.

In fact, the RDRAM perhaps has future, especially if the prices go down drastically, but there seems to be absolutely no technical need to change to the RDRAM in the current moment, but only marketing motives. In a few months we will see the solution to this foul-up, although I would bet for the short-term PC133 and for the long-term RDRAM... or almost better for the DDR-SDRAM, a SDRAM with technology of double use of every cycle of clock (like the RDRAM)... you are going to know; but to buy actions of Rambus seems risky for the time being.

(Ah, certainly: the modules of memory RDRAM will look alike very much to the DIMM of SDRAM, although they will have 184 pines and will be named RIMM... originally: truth?)


To return

Pulsate in the picture desplegable or in "Following"

Following - Sitemap - Contact us Best website of 2009 & 2010 voted by us are:
, , , , , ,,