Meanwhile Intel continues to keep the wraps on the already demonstrated system that was said to be based on a desktop chip with two processing engines, the analyst, who laid down assumptions that Intel had not really showcased any dual-core desktop central processing unit, has issued a yet another commentary explaining one more possibility of Intel’s shady demonstration.
“Since I published the report, some of my readers have suggested that I used too broad a definition of “engineering prototype.” They argue that a dual-core prototype must, at a minimum, include two processor cores on a single piece of silicon, and that Intel would be misleading its audience if it tried to fit any other implementation under the “dual-core” label. They argue further that Intel would not be willing to risk its reputation for high integrity, just for the sake of a product demo at a conference,” principal analyst for Insight 64, Nathan Brookwood, said.
In order to find out another explanation of what had happened during the presentation in early September Mr. Brookwood questioned other experts at Insight 64 and came up with the idea that Intel did not “dice” two processing cores used in today’s Intel Pentium 4 processors after they were made, but placed them onto a special package that allowed them to work in pair.
“Intel manufactures its processors on large (300mm diameter) wafers that hold approximately 500 individual processors. After it completes some basic tests on each processor on the wafer, Intel separates the individual dice (a process known as singulation), and packages those chips that it believes will operate properly. […] On any wafer, there will be many instances where two, four or even eight contiguous processors are all defect free. Intel needs only to keep these pairs of known good die together, dividing the wafer into 250 dyadic chips, rather than 500 individual die. It would need a new package that to accommodate this large chip (just a little more than 2x the size of the single core processor shipping today). The modified package would have the same footprint and “pinout” as the current chip, but would have internal connections that distribute bus signals and power to both cores. The package would work with Intel’s standard Pentium 4 motherboards,” Mr. Brookwood explained.
Insight 64 believes Intel could have used the described approach, or something similar, to create the chips it used for the demo, and still legitimately claim that it had displayed an “engineering prototype of a dual-core processor”.
At the same time, when asked whether the demonstration of the desktop dual-core chip involved a single-die dual-core product, Intel declined to comment.
“We will disclose more details later. We cannot do it before official launch,” an Intel’s representative said.