Coppermine: Pentium III EB/E
Half a year ago we were already publishing a titled article "Pentium III: an immature mike?" in which, on having described the most recent Intel microprocessor, was commenting on his scarce technological innovations with regard to Pentium II, which were making him an almost inevitable candidate for a prompt update. Well, the moment has come: we present the Pentium III Coppermine.
What has changed?
In the new remittance of Pentium III four innovations have got, fundamentally:
Nevertheless, four changes have not been applied to any new mike simultaneously, but we meet mikes Slot1 with 133 MHz bus but I surprised the ancient one, mikes with modern cache memory but 100 MHz bus, mikes PGA370 but with bus to 100 MHz... the classification is an authentic madness that we will treat further on in a specific paragraph.
Now we are going to center on what each of the changes contributes:
Technology of 0,18 microns manufacture
This factor is of supreme importance, and one of the most difficult to carry out since it implies a very important technological progress. Up to the date, the microprocessors were made by a 0,25 microns process (that is to say, that the smallest parts of the mike were measuring 0,25 microns); the 0,18 microns process would be equivalent to 600 times less than the thickness of a human hair.
On having used this technology, we manage:
The 0,25 microns process in Pentium III had come to his practical limit with Pentium III to 600 MHz, who even without overclocking needed a voltage of 2,05V instead of 2,00V normal to work in a stable way:
It turns out to be certainly impressive: 74 % less heat! (less enclosed heat than a Celeron to 333 MHz); and also they seem to have aptitude to be overclockeados up to approximately 750-800 MHz without many problems... the whole progress, undoubtedly.
256 KB of cache memory L2 integrated, at the same speed as the mike
If the previous factor was that it was making possible to make the mikes, this one can award almost the whole merit of the increase of yield. The idea of including the secondary cache memory or L2 in the encapsulated proper one of the mike it is not new, it already does more than one year that there appeared the Celeron with cache memory or Mendocino, with 128 KB of cache memory L2 integrated at the same speed as the mike that they make practically equal to him of rapid that a Pentium II (and even that a Pentium III in many applications).
The innovation takes root, apart from in using 256 KB instead of 128, in which Intel has used a bus for the cache memory L2 of 256 bits wide, 4 times more wide than the used one in Pentium II and in the previous Pentium III. This new type of cache memory (who Intel calls "Advanced Transfer Cache"), of big speed and big band, is of benefit to many applications, especially "professionals" (CAD, image treatment, Windows NT) and is the best trick of Pentium III Coppermine.
Speed of 133 MHz bus
The increase of the bus from 100 up to 133 Mhz, although it is quite more 33 % (we do not explain the mathematics necessary for this calculation for his complexity), does not imply a very big increase of the real yield, especially if we compare it with the increase that generates the new design of the cache memory; also, any microprocessor with a rapid cache memory does not depend much of the speed on the bus, as it demonstrates the rapid Celeron with his "scarce" 66 Mhz.
Anyway, in some very technical applications that need bands big, like the servants or applications with intensive calculations, it can be beneficial; but it it would be more with a type of memory of major band than the PC100, something that as we will see is not so simple to manage.
Connector type socket FC-PGA370
If we have a mike with the cache memory L2 integrated in the same encapsulated that the nucleus, and also his size is very small thanks to the 0,18 microns process: what advantages does he it have to install in a cartridge? Absolutely none.
Intel has decided that the most logical thing (and the cheapest thing) is to leave the format in Slot1 type groove that began using with Pentium II and turning to the outstanding figure format in socket, format that we think many that it retired principally because the Slot1 cannot use without license... a license that never gave to AMD. This abandonment will be done gradually, but it is possible to affirm that in less than one year mikes will not be already made for Slot1.