“Profiling is a formal summary or analysis of data, often in the
form of a graph or table, representing distinctive performance
features or characteristics. The profiling table provides the
percentage and number of samples collected for specified processor
events such as the number of cache line misses, Transition
Lookaside Buffer (TLB) misses, and so on.“OProfile is one of several profiling and performance monitoring
tools for Linux. It works on various architectures, including the
IA32, IA64, and AMD Athlon families, has a low overhead, and will
be included in the 2.6 version of the kernel.“OProfile can help you identify issues such as loop unrolling,
poor cache utilization, inefficient type conversion and redundant
operations, branch mispredictions, and so on. It collects
information about processor events including TLB misses, stalls,
memory references, total lines allocated in the DCU (Data Cache
Unit), the number of cycles of a DCU miss, and the number of
non-cacheable and cacheable instruction fetches. OProfile is
fine-grained and can collect samples for a set of instructions, or
for function, system call, or interrupt handlers. OProfile works by
sampling, and using the collected profile data, you can easily
identify performance problems…”
Related Story:
Linus
Torvalds: Linux 2.6.0-test6(Sep 28, 2003)