Surely Intel doesn't have one production line for the top-of-the-range Hyperthreading enabled processor and another for rest of its chips? It just wouldn't make sense. The hunt for a way to enable Hyperthreading on processors further down the marchitecture scale was on. Pin modification was the first thing that was tried but to no avail. Time passed.
One of the lovely things about the Internet is that it holds huge amounts of answers. One of the nasty things about the Internet is that, just as philosophic tradition would be happy to point out, knowing the question is often the biggest hindrance. For instance, here is a nice patent filed by Intel many years ago. On Feb 28 1995 to be exact. It outlines the idea of using some flash memory on a chip to control certain aspects of how the chip works and other details. It would normally contain updates for the microcode on the chip but could contain other information. For instance, it could be used to keep a record of the production line that the chip came from. Or to mark off areas of the cache that didn't work. Or to set the speed. Or to enable/disable features.
If you knew which questions to ask, this article would have answered some of your questions. Such as, is this flash memory mentioned in the patent present on current Intel Pentiums? And, can the flash memory be updated while it's sat inside a PC? The answers to both are 'yes.'
It also seems that someone managed to accidentally enable Hyperthreading on a processor that wasn't supposed to have it. But nobody seems to know exactly how this happened. So now the hunt is on for a way to update Intel processors using this carefully hidden little feature.
The questions now are:
Has anyone else managed to enable Hyperthreading on a non-HT P4?
Does anyone know the specific parameters used in the flash memory?
Is the Pentium 4 2.4GHz really a 3.06GHz HT in a changeable disguise?
One thing is for certain, Intel is going to remain tight-lipped. A few questions put its way all resulted in a simple "no comment." µ
L'INQS
Special thanks must go to the people of VR Forums who have been plugging away at this problem for months. You can
find the original thread
here.