Lines Matching +full:cache +full:- +full:time +full:- +full:ms

54 out at 25-35 tps in practice), and the user cpu utilization (us) should
60 be non-zero).  It is healthy for the paging demon to free pages when
70 is overloaded or imbalanced.  If you have several non-dma
72 that are doing high-speed non-buffered input/output, then the system
73 time may go high (60-80% or higher).
74 It is often possible to pin down the cause of high system time by
94 It also gives the percentage of time the system has
102 Also, every 100 ms,
118 non-zero average seek times.  Most modern disk drives should
119 exhibit an average seek time of 25-35 ms.
135 .B \-p
138 or by an alternate real time clock.
141 the time spent in the kernel clock routine.
156 The postprocessing propagates the time spent in each
159 sorted according to the time it uses
160 including the time of its call graph descendents.
163 A similar display above the routine shows how this routine's time and the
164 time of its descendents is propagated to its (direct) call graph parents.
166 A profiled system is about 5-10% larger in its text space because of
175 under normal load we see anywhere from 5-25%
176 of the system time spent in the profiling code.
190 system is operating.  Each trace record contains a time stamp
191 (taken from the VAX hardware time of day clock register), an
193 according to the event type.  Buffer cache operations, such as
198 buffer is normally configured to hold 256 16-byte trace records.\**
201 differ slightly from those described here.  The time stamp in the
202 distributed system is calculated from the kernel's time of day
203 variable instead of the VAX hardware register, and the buffer cache
223 the file system buffer cache.  The sampling program maintained a
224 history of read-ahead blocks and used the trace information to
225 calculate, for example, percentage of read-ahead blocks used.
232 context switching time are critically important in evaluating the