Re: [PATCH 1/7] Simple Performance Counters: Core Piece

2007-08-17 Thread Mathieu Desnoyers
* Christoph Lameter ([EMAIL PROTECTED]) wrote: > On Fri, 17 Aug 2007, Mathieu Desnoyers wrote: > > > Actually, get_cycles() at least on some AMD cpus, do not synchronize the > > core, which can skew the results. You might want to use > > get_cycles_sync() there. > > get_cycle() results as used he

Re: [PATCH 1/7] Simple Performance Counters: Core Piece

2007-08-17 Thread Christoph Lameter
On Fri, 17 Aug 2007, Mathieu Desnoyers wrote: > Actually, get_cycles() at least on some AMD cpus, do not synchronize the > core, which can skew the results. You might want to use > get_cycles_sync() there. get_cycle() results as used here are bound to a single processor. If we end up on a differ

Re: [PATCH 1/7] Simple Performance Counters: Core Piece

2007-08-17 Thread Mathieu Desnoyers
Hi Christoph, Actually, get_cycles() at least on some AMD cpus, do not synchronize the core, which can skew the results. You might want to use get_cycles_sync() there. Mathieu * Christoph Lameter ([EMAIL PROTECTED]) wrote: > Simple performance counters are a way to measure the performance on cod

Re: [PATCH 1/7] Simple Performance Counters: Core Piece

2007-07-31 Thread Frank Ch. Eigler
Christoph Lameter <[EMAIL PROTECTED]> writes: > Simple performance counters are a way to measure the performance on > code paths in the Linux kernel. Code must be instrumented with calls > that signal the start and the stop of a measurement. [...] For what it's worth, this kind of measurement w

[PATCH 1/7] Simple Performance Counters: Core Piece

2007-07-31 Thread Christoph Lameter
Simple performance counters are a way to measure the performance on code paths in the Linux kernel. Code must be instrumented with calls that signal the start and the stop of a measurement. The beginning of a code path must have the following. Either: INIT_PC(var) or struct pc va