CPU multipliers

[Mookow]

AMD, if I recall correctly, has their chips in their packaging before doing the final speed test (I assume that they are tested for functionality prior to this), and then they burn the L# bridges as appropriate to set the multiplier for that particular chip. How does Intel do this? Do they hard-wire it into the CPU and just toss the chips that would have been functional at a lower speed, but cannot function with their hare-coded multiplier? Does Intel have that much spare fab capacity to just waste the chips, or is there an advantage to doing it this way aside from preventing making remarkers job easier?

 

[pm]

There are electrical fuses on the die to permanently set the serial number, the processor ID, set any required redundant cache lines, and set the multiplier.

 

[Mookow]

when are they set, though? or has my talent for asking questions you arent allowed to answer found one of those questions again?

 

[pm]

No, my habit of replying to you "I can't answer that" has only been when you ask about future products. Multipliers and product ID's are set after binning - which near the end of the manufacturing flow.