background image

Vol. 3B 19-17

PERFORMANCE-MONITORING EVENTS

9CH

01H

IDQ_UOPS_NOT_DELIVERED.CO

RE

Count issue pipeline slots where no uop was 

delivered from the front end to the back end when 

there is no back end stall. 

Use Cmask to qualify uop 

b/w.

A1H

01H

UOPS_DISPATCHED_PORT.PORT

_0

Counts the number of cycles in which a uop is 

dispatched to port 0.

Set AnyThread to count 

per core.

A1H

02H

UOPS_DISPATCHED_PORT.PORT

_1

Counts the number of cycles in which a uop is 

dispatched to port 1.

Set AnyThread to count 

per core.

A1H

04H

UOPS_DISPATCHED_PORT.PORT

_2

Counts the number of cycles in which a uop is 

dispatched to port 2.

Set AnyThread to count 

per core.

A1H

08H

UOPS_DISPATCHED_PORT.PORT

_3

Counts the number of cycles in which a uop is 

dispatched to port 3.

Set AnyThread to count 

per core.

A1H

10H

UOPS_DISPATCHED_PORT.PORT

_4

Counts the number of cycles in which a uop is 

dispatched to port 4.

Set AnyThread to count 

per core.

A1H

20H

UOPS_DISPATCHED_PORT.PORT

_5

Counts the number of cycles in which a uop is 

dispatched to port 5.

Set AnyThread to count 

per core.

A1H

40H

UOPS_DISPATCHED_PORT.PORT

_6

Counts the number of cycles in which a uop is 

dispatched to port 6.

Set AnyThread to count 

per core.

A1H

80H

UOPS_DISPATCHED_PORT.PORT

_7

Counts the number of cycles in which a uop is 

dispatched to port 7.

Set AnyThread to count 

per core.

A2H

01H

RESOURCE_STALLS.ANY

Cycles Allocation is stalled due to resource related 

reason. 

A2H

04H

RESOURCE_STALLS.RS

Cycles stalled due to no eligible RS entry available. 

A2H

08H

RESOURCE_STALLS.SB

Cycles stalled due to no store buffers available (not 

including draining form sync).

A2H

10H

RESOURCE_STALLS.ROB

Cycles stalled due to re-order buffer full.

A8H

01H

LSD.UOPS

Number of uops delivered by the LSD.

ABH

02H

DSB2MITE_SWITCHES.PENALTY

_CYCLES

Cycles of delay due to Decode Stream Buffer to MITE 

switches.

AEH

01H

ITLB.ITLB_FLUSH

Counts the number of ITLB flushes; includes 

4k/2M/4M pages.

B0H

01H

OFFCORE_REQUESTS.DEMAND_

DATA_RD

Demand data read requests sent to uncore. 

Use only when HTT is 

off.

B0H

02H

OFFCORE_REQUESTS.DEMAND_

CODE_RD

Demand code read requests sent to uncore. 

Use only when HTT is 

off.

B0H

04H

OFFCORE_REQUESTS.DEMAND_

RFO

Demand RFO read requests sent to uncore, including 

regular RFOs, locks, ItoM.

Use only when HTT is 

off.

B0H

08H

OFFCORE_REQUESTS.ALL_DATA

_RD

Data read requests sent to uncore (demand and 

prefetch).

Use only when HTT is 

off.

B1H

01H

UOPS_EXECUTED.THREAD

Counts total number of uops to be executed per-

logical-processor each cycle.

Use Cmask to count stall 

cycles.

B1H

02H

UOPS_EXECUTED.CORE

Counts total number of uops to be executed per-core 

each cycle.

Do not need to set ANY.

B7H

01H

OFF_CORE_RESPONSE_0

See Section 18.9.5, “Off-core Response Performance 

Monitoring”.

Requires MSR 01A6H.

Table 19-5.  Non-Architectural Performance Events of the Processor Core Supported by Broadwell 

Microarchitecture (Contd.)

Event

Num.

Umask

Value

Event Mask Mnemonic

Description

Comment