moocow, as Reynardin pointed out, 72.4C is for Tcase max, not Tjmax. Tjmax is die temp, Tcase is IHS/outer casing temp.
At idle, undervolted, underclocked state, which is how I do my temperature testing, the two temps, die and case are within 1C. Temp difference between two, or gradient = TDP*theta or 4W*.35c/w or roughly 1.2C. Will actually be less than 1C with no heatink on.
However at load with heatsink on, there can be a high gradient or difference between tcase and die temps (tjmax), for example TDP of 50W*.35 = 17C (max rated load TD power E8400 = 65W). However this buffer or gradient is highly variable on cooling, ambients, load type (floating point etc), TDP and therefore can not be counted on.
another similar thread here:
http://forums.anandtech.com/messageview.aspx?catid=28&threadid=2195801
But that is why tjmax 95C, die temp, is higher than tcasemax (72.4C) casing temps. Intel does not expect me to be running at idle with no heatsink, pulling off IHS's etc, where die temp=IHS temp (within .5C). Normally only way to get temps that high is with load, hence with gradient, hence the tjmax higher than tcasemax.
But there is no sensor or protection at tcase. Only protection is cpu diodes ie tjmax, thus I can run my temp up to 95C on E8400 before throttling occurs on my cpu... at my risk, as I am indeed also at tcase 95C and way over safe temp...because I am running under conditions where tcase=die, as opposed to normal high temp conditions (load) where tj temps higher than tcase.. And I can run temp up to ~115C before cpu shutsdown.
Below is a pic of my E7200, throttling (prochot assertion at bottom in realtemp) has occurred, I am at 104C one core, 108C other core. Delta to tjmax is 7 bit, so can only register up to DTS=0, then goes to 127C and starts backcounting, ie is fubar'd. By the way, I killed my E7200, fried the die from bare die testing. Testing with IHS is pretty safe, with bare die not so much.
See intel specs quote, regarding TCC temp (at tjmax), and next quote is shutdown temp which is 20C hotter than TCC (tjmax)
http://download.intel.com/design/pro...s/31873201.pdf
"An external signal, PROCHOT# (processor hot), is asserted when the processor core temperature has reached its maximum operating temperature. If the Thermal Monitor
is enabled, the TCC will be active when PROCHOT# is asserted....The TCC causes the processor to adjust its operating frequency (via the bus multiplier) and input voltage (via the VID signals). This combination of reduced frequency and VID results in a reduction to the processor power consumption."
http://download.intel.com/design/pro...x/31873401.pdf
"In the event of a catastrophic cooling failure, the processor will automatically shut down when the silicon temperature has exceeded the TCC activation temperature by approximately 20 to 25 °C. At this point the system bus signal THERMTRIP# goes active and power must be removed from the processor."