Thermal Solutions
40
Thermal/Mechanical Design Guide
5.6.2
PECI Averaging and Catastrophic Thermal Management
By averaging DTS over PECI, thermal solution failure can be detected and a soft
shutdown can be initiated to help prevent loss of data.
Thermal data is averaged over a rolling window of 256mS by default (X=8):
AVG
N
= AVG
N-1
* (1 – 1/2
X
) + Temperature * 1/2
X
Using a smaller averaging constant could cause premature detection of failure.
The Critical Temperature threshold generally triggers somewhere between PECI of
-0.75 and -0.50. To avoid false shutdowns, initiate soft shutdown at -0.25.
Since customer designs, boundary conditions, and failure scenarios differ, above
guidance should be tested in the customer’s system to prevent loss of data during
shutdown.
5.6.3
Intel® Turbo Boost Technology
Intel® Turbo Boost Technology (Intel®
TBT) is a new feature available on certain
processor SKUs that opportunistically, and automatically, allows the processor to run
faster than the marked frequency if the part is operating below its power, temperature
and current limits.
Heatsink performance (lower
Ψ
CA
as described in
Section 5.5.1
) is one of several
factors that can impact the amount of Intel® TBT frequency benefit. Intel® TBT
performance is also constrained by ICC, and VCC limits.
Increased IMON accuracy may provide more Intel® TBT benefit on TDP limited
applications, as compared to lower
Ψ
CA
, as temperature is not typically the limiter for
these workloads.
With Intel® TBT enabled, the processor may run more consistently at higher power
levels (but still within TDP), and be more likely to operate above T
CONTROL
, as
compared to when Intel® TBT is disabled. This may result in higher acoustics.
With Intel® TBT enabled, processors with dual thermal profiles (described in
Section 5.5.2
, have greater potential for performance delta between Profile A and
Profile B platforms, as compared to previous platforms.
5.7
Thermal Guidance
5.7.1
Thermal Excursion Power for 95 W Processor
Under fan failure or other anomalous thermal excursions, Tcase may exceed Thermal
Profile B for a duration totaling less than 360 hours per year without affecting long term
reliability (life) of the processor. For more typical thermal excursions, Thermal Monitor
is expected to control the processor power level as long as conditions do not allow the
Tcase to exceed the temperature at which Thermal Control Circuit (TCC) activation
initially occurred. Under more severe anomalous thermal excursions when the
processor temperature cannot be controlled at or below this Tcase level by TCC
activation, then data integrity is not assured. At some higher threshold, THERMTRIP#
will enable a shut down in an attempt to prevent permanent damage to the processor.
Thermal Test Vehicle (TTV) may be used to check anomalous thermal excursion
Summary of Contents for Xeon 5500 Series
Page 8: ...8 Thermal Mechanical Design Guide ...
Page 12: ...Introduction 12 Thermal Mechanical Design Guide ...
Page 26: ...Independent Loading Mechanism ILM 26 Thermal Mechanical Design Guide ...
Page 42: ...Thermal Solutions 42 Thermal Mechanical Design Guide ...
Page 50: ...Component Suppliers 50 Thermal Mechanical Design Guide ...
Page 78: ...Mechanical Drawings 78 Thermal Mechanical Design Guide ...
Page 84: ...Socket Mechanical Drawings 84 Thermal Mechanical Design Guide ...
