Hello,
I am experimenting with IBS on AMD Shanghai processor.
Currently my code is based on the IBS example in libpfm3.9. I only capture
the load operation, and play with the sampling period.
When using small sampling period, the number of captured data is larger, but
the distribution is not ver
Hi, all
I do not sure some events description and their result. Take
event "BRANCH_INSTRUCTIONS_RETIRED" for example. `showevtinfo`
says this event counts the retirement of the last micro-op of
a branch instruction.
When I sample an application with event "BRANCH_INSTRUCTIONS_RETIRED",
I will
Hi,
What you're observing is a known side effect of interrupt-based
sampling. There is skid. Let me explain.
What the kernel captures is the address of the instruction at the time
of the PMU interrupt. That instruction may be far away from the instruction
that cause the counter to overflow, i.e.,
Is there any more informations about 'perf' ? I learn perfmon2, because of I
want to get the event of cpu by myself.
I have developed a tool which need this event, so I must to write these codes
by myself.
Sometimes you mustn't update the kernel due to the system used to production
line.
The
On 03/16/2011 07:35 AM, Easley wrote:
> Is there any more informations about 'perf' ? I learn perfmon2, because of I
> want to get the event of cpu by myself.
> I have developed a tool which need this event, so I must to write these codes
> by myself.
> Sometimes you mustn't update the kernel du
Hi, Stephane
> The PMU takes some variable amount of time to interrupt after an overflow.
> During that time, execution continues.
So the skid has nothing to do with "micro-op"?
> This "imprecision" cannot be corrected by software but only thru hardware.
> That's why you have Intel PEBS, for
Hi,
> In summary, except for cycles you should do not expect profiles to point at
> instructions that generated the occurrences of the sampling event.
Even "instructions" event has the skid?
Regards,
chenwj
--
Wei-Ren Chen (陳韋任)
Parallel Processing Lab, Institute of Information Science,
Acad
On Wed, Mar 16, 2011 at 7:07 PM, 陳韋任 wrote:
> You mean PEBS makes the skid disappered? Does `perf` or libpfm4 support this?
> The example I showed was ran on a x86 machine, it seems `perf` does not use
> PEBS
> feature.
>
It depends on the suffix. Try perf record -e branch-misses:p
PEBS is kn
EventName(2CH) UnitMask(07H) UNC_QMC_NORMAL_READS.ANY
Counts the number of Quickpath Memory Controller medium and low priority read
requests. The QMC normal read occupancy divided by this count provides the
average QMC read latency.
-
EventName(2FH) UnitMask(07H) UNC_Q
Hi,
> How does I take use of EventName and UnitMask to get UNC_QMC_NORMAL_READS.ANY
> or UNC_QMC_WRITES.FULL.ANY evnets? And Which are these 2 events control bits?
I guess you are using pfmon, try "-e UNC_QMC_NORMAL_READS:ANY".
Regards,
chenwj
--
Wei-Ren Chen (陳韋任)
Parallel Processing Lab,
10 matches
Mail list logo