Method and apparatus for determining CC-NUMA intra-processor delays
AssigneeBull HN Information Systems Inc.
InventorsCharles P. Ryan, Eric E. Conway
Life of this patent
Estimating the term…
Loading patent…
AssigneeBull HN Information Systems Inc.
InventorsCharles P. Ryan, Eric E. Conway
Estimating the term…
In a Cache-Coherent Non-Uniform Memory Architecture (CC-NUMA), the time as measured in cycles that it takes for cache control signals to travel between processors (92) sharing an L2 cache (94) differs from the time it takes for those signals to travel between processors (92) not sharing the same L2 cache (94). This difference, or DELTA, is dynamically computed by computing (332) the time it takes for a invalidate cache line cache command to travel between a local processor (92) and a master processor (92). This computation (334) is then made for the time it takes the signal to travel between a remote processor (92) and the master processor (92). The difference (336) is the DELTA value in cycles. This DELTA value can then be utilized to bias delay values when exhaustively testing the interactions among multiple processors in a CC-NUMA environment (180).
In a data processing system having a plurality of processors, a method of dynamically calculating a difference between a time it takes a signal to travel between a first test processor and a second test processor and a time it takes a signal to travel between the first test processor and a third test processor, the first test processor being one of a first pair of test processors, the second test processor being one of the first pair of test processors, and the third test processor being one of a second pair of test processors,
In a data processing system having a plurality of processors and containing software stored in a Computer Software Storage Medium for dynamically calculating a difference between a time it takes a signal to travel between a first test processor and a second test processor and a time it takes a signal to travel between the first test processor and a third test processor, the first test processor being one of a first pair of test processors, the second test processor being one of the first pair of test processors, and the third test processor being one of a second pair of test processors,
Dependencies are inferred from the claim text.