Some musing on the relation between bench and system testing of a wb.
In the unlikely event of true CPU temps have considered two(2) cases for the Cooltech WB75 cooling a 1sq cm Die( ~Duron):
a) Wattage calculator and/or stress program are crap(70%) and no significant secondary heat paths(e.g. thro mb and mb C/W = infinity))
b)Wattage calculator and stress program are 100% correct but there are significant secondary heat paths(e.g. mb C/W=0.5 and 0.7).
Possibly worth noting that heat loss by secondary paths delivers a double whammy to any temp changes.
.