libpfm_intel_hsw(3) support for Intel Haswell core PMU


#include <perfmon/pfmlib.h>

PMU name: hsw
PMU desc: Intel Haswell
PMU name: hsw_ep
PMU desc: Intel Haswell-EP


The library supports the Intel Haswell and Haswell-EP core PMU. It should be noted that this PMU model only covers each core's PMU and not the socket level PMU.

On Haswell, the number of generic counters depends on the Hyperthreading (HT) mode. When HT is on, then only 4 generic counters are available. When HT is off, then 8 generic counters are available. The pfm_get_pmu_info() function returns the maximum number of generic counters in num_cntrs.


The following modifiers are supported on Intel Haswell processors:
Measure at user level which includes privilege levels 1, 2, 3. This corresponds to PFM_PLM3. This is a boolean modifier.
Measure at kernel level which includes privilege level 0. This corresponds to PFM_PLM0. This is a boolean modifier.
Invert the meaning of the event. The counter will now count cycles in which the event is not occurring. This is a boolean modifier
Enable edge detection, i.e., count only when there is a state transition from no occurrence of the event to at least one occurrence. This modifier must be combined with a counter mask modifier (m) with a value greater or equal to one. This is a boolean modifier.
Set the counter mask value. The mask acts as a threshold. The counter will count the number of cycles in which the number of occurrences of the event is greater or equal to the threshold. This is an integer modifier with values in the range [0:255].
Measure on both threads at the same time assuming hyper-threading is enabled. This is a boolean modifier.
Pass a latency threshold to the MEM_TRANS_RETIRED:LOAD_LATENCY event. This is an integer attribute that must be in the range [3:65535]. It is required for this event. Note that the event must be used with precise sampling (PEBS).
Monitor the event only when executing inside a transactional memory region (in tx). Event does not count otherwise. This is a boolean modifiers. Default value is 0.
Do not count occurrences of the event when they are inside an aborted transactional memory region. This is a boolean modifier. Default value is 0.


Intel Haswell provides two offcore_response events. They are called OFFCORE_RESPONSE_0 and OFFCORE_RESPONSE_1.

Those events need special treatment in the performance monitoring infrastructure because each event uses an extra register to store some settings. Thus, in case multiple offcore_response events are monitored simultaneously, the kernel needs to manage the sharing of that extra register.

The offcore_response events are exposed as a normal events by the library. The extra settings are exposed as regular umasks. The library takes care of encoding the events according to the underlying kernel interface.

On Intel Haswell, the umasks are divided into three categories: request, supplier and snoop. The user must provide at least one umask for each category. The categories are shown in the umask descriptions.

There is also the special response umask called ANY_RESPONSE. When this umask is used then it overrides any supplier and snoop umasks. In other words, users can specify either ANY_RESPONSE OR any combinations of supplier + snoops.

In case no supplier or snoop is specified, the library defaults to using ANY_RESPONSE.

For instance, the following are valid event selections:


But the following are illegal:



Stephane Eranian <[email protected]>