Intel’s original 64bit extensions for x86

Introduction

In the late 1900s, Intel was fully invested in Itanium (IA-64) being their future 64bit architecture.

At that time (allegedly around 1997-1998), some Intel engineers (that were more cautious than their management at that time) built a 64bit extension for x86 as a backstop, which never shipped:

Intel’s Pentium 4 had our own internal version of x86–64. But you could not use it: we were forced to “fuse it off”, meaning that even though the functionality was in there, it could not be exercised by a user.
This was a marketing decision by Intel — they believed, probably rightly, that bringing out a new 64-bit feature in the x86 would be perceived as betting against their own native-64-bit Itanium, and might well severely damage Itanium’s chances.
– Bob Colwell¹

Only a few years later, AMD announced (in 1999) and subsequently shipped (in 2003) their own 64bit extension for x86, condemning both Itanium and Intel’s own x86 extension to obscurity.

Intel implemented AMD’s design in Project Yamhill and shipped its first processors in 2004.

How did Intel’s design look like?

While AMD’s 64bit extension design that repurposed inc/dec instructions as the REX prefix is well-documented, there is very little known about Intel’s own 64bit extension design.

Here is what can be reconstructed from Intel’s patent applications from 2000 and 2003²:

An instruction having [a] format […] — where the mode field is 01B, the R/M field is 100B, the index field is 100B — has addressing mode information that is currently unsupported in the IA-32 architecture regardless of the value of the scale field.
An instruction of [that] format […] thereby includes heretofore unused bit fields (e.g., the two bits of scale field, bits of displacement, etc.), that can support an expanded logical register set for existing instructions formats and legacy operands.

Material from the Bristol Community College³ also mentions this specific combination of bits:

Note that this addressing mode does not allow the use of the ESP register as an index register.
Presumably, Intel left this particular mode undefined to provide the ability to extend the addressing modes in a future version of the CPU.

Conclusion

Sadly, there is no definitive information on how close the patents were to what was shipped (fused-off) in Intel processors of that specific time period.