xref: /titanic_52/usr/src/data/perfmon/readme.txt (revision 53548f91e84cd97a638c23b5b295cc69089a5030)
1---------------------
2This package contains performance monitoring event lists for Intel processors, as well as a mapping file
3to help match event lists to processor Family/Model/Stepping codes.
4---------------------
5
6The event lists are available in 2 formats:
7	Tab delimited (.tsv)
8	Json (.json)
9
10Event lists are created per microarchitecture, and each has a version. Versions are listed in the event list
11name as well as the header for each file. For some microarchitectures, up to three different event lists will
12be available. These event lists correspond to the types of events that can be collected:
13
14core - Contains events counted from within a logical processor core.
15offcore - Contains matrix events counted from the core, but measuring responses that come from offcore.
16
17The event list filename indicates which type of list it contains, and follows this format:
18<microarchitecture-codename>_<core/offcore>_<version>
19
20New version releases will be announced in the mail list perfmon-announce@lists.01.org
21
22Different microarchitectures provide different performance monitoring capabilities, so field names and categories
23of events may vary.
24
25---------------------
26Event List Field Defitions:
27---------------------
28Below is a list of the fields/headers in the event files and a description of how SW tools should
29interpret these values. A particular event list from this package may not contain all the fields described
30below. For more detailed information of the Performance monitoring unit please refer to chapters 18 and 19
31of Intel� 64 and IA-32 Architectures Software Developer's Manual Volume 3B: System Programming Guide, Part 2.
32
33http://www.intel.com/content/www/us/en/processors/architectures-software-developer-manuals.html
34
35
36----EventCode----
37This field maps to the Event Select field in the IA32_PERFEVTSELx[7:0]MSRs. The set of values for this field
38is defined architecturally. Each value corresponds to an event logic unit and should be used with a unit
39mask value to obtain an architectural performance event.
40
41----UMask----
42This field maps to the Unit Mask filed in the IA32_PERFEVTSELx[15:8] MSRs. It further qualifies the event logic
43unit selected in the event select field to detect a specific micro-architectural condition.
44
45----EventName----
46It is a string of characters to identify the programming of an event.
47
48----BriefDescription----
49This field contains a description of what is being counted by a particular event.
50
51----PublicDescription----
52In some cases, this field will contain a more detailed description of what is counted by an event.
53
54----Counter----
55This field lists the fixed (PERF_FIXED_CTRX) or programmable (IA32_PMCX) counters that can be used to count the event.
56
57----CounterHTOff----
58This field lists the counters where this event can be sampled when Intel� Hyper-Threading Technology (Intel� HT Technology) is
59disabled. When Intel� HT Technology is disabled, some processor cores gain access to the programmable counters of the second
60thread, making a total of eight programmable counters available. The additional counters will be numbered 4,5,6,7. Fixed counter
61behavior remains unaffected.
62
63----PEBScounters----
64This field is only relevant to PEBS events. It lists the counters where the event can be sampled when it is programmed as a PEBS event.
65
66----SampleAfterValue----
67Sample After Value (SAV) is the value that can be pre-loaded into the counter registers to set the point at which they will overflow.
68To make the counter overflow after N occurrences of the event, it should be loaded with (0xFF..FF � N) or �(N-1). On overflow a
69hardware interrupt is generated through the Local APIC and additional architectural state can be collected in the interrupt handler.
70This is useful in event-based sampling. This field gives a recommended default overflow value, which may be adjusted based on
71workload or tool preference.
72
73----MSRIndex----
74Additional MSRs may be required for programming certain events. This field gives the address of such MSRS.
75Potential values are:
760x3F6: MSR_PEBS_LD_LAT - used to configure the Load Latency Perforamnce Monitoring Facility
770x1A6/0x1A7: MSR_OFFCORE_RSP_X - used to configure the offcore response events
78
79----MSRValue----
80When an MSRIndex is used (indicated by the MSRIndex column), this field will contain the value that needs to be loaded into the
81register whose address is given in MSRIndex column. For example, in the case of the load latency events, MSRValue defines the
82latency threshold value to write into the MSR defined in MSRIndex (0x3F6).
83
84----CollectPEBSRecord----
85Applies to processors that support both precise and non-precise events in Processor Event Based Sampling, such as Goldmont.
860: The event cannot be programmed to collect a PEBS record.
871: The event may be programmed to collect a PEBS record, but caution is advised.
88For instance, PEBS collection of this event may consume limited PEBS resources whereas interrupt-based sampling may be sufficient for the usage model.
892: The event may programmed to collect a PEBS record, and due to the nature of the event, PEBS collection may be preferred.  For instance,
90PEBS collection of Goldmont�s HW_INTERUPTS.RECIEVED event is recommended because the hardware interrupt being counted may lead to the masking of
91interrupts which would interfere with interrupt-based sampling.
92
93
94	----TakenAlone----
95This field is set for an event which can only be sampled or counted by itself, meaning that when this event is being collected,
96the remaining programmable counters are not available to count any other events.
97
98----CounterMask----
99This field maps to the Counter Mask (CMASK) field in IA32_PERFEVTSELx[31:24] MSR.
100
101----Invert----
102This field corresponds to the Invert Counter Mask (INV) field in IA32_PERFEVTSELx[23] MSR.
103
104----AnyThread----
105This field corresponds to the Any Thread (ANY) bit of IA32_PERFEVTSELx[21] MSR.
106
107----EdgeDetect----
108This field corresponds to the Edge Detect (E) bit of IA32_PERFEVTSELx[18] MSR.
109
110----PEBS----
111A '0' in this field means that the event cannot collect a PEBS record with a Precise IP.  A '1' in this field means that the event is a
112precise event and can be programmed in one of two ways � as a regular event or as a PEBS event. And a '2' in this field means
113that the event can only be programmed as a PEBS event.
114
115----PRECISE_STORE----
116A '1' in this field means the event uses the Precise Store feature and Bit 3 and bit 63 in IA32_PEBS_ENABLE MSR must be set
117to enable IA32_PMC3 as a PEBS counter and enable the precise store facility respectively. Processors based on SandyBridge and
118IvyBridge micro-architecture offer a precise store capability that provides a means to profile store memory references in
119the system.
120
121----DATA_LA----
122A '1' in this field means that when the event is configured as a PEBS event, the Data Linear Address facility is supported.
123The Data Linear Address facility is a new feature added to Haswell as a replacement or extension of the precise store facility
124in SNB.
125
126----L1_HIT_INDICATION----
127A '1' in this field means that when the event is configured as a PEBS event, the DCU hit field of the PEBS record is set to 1
128when the store hits in the L1 cache and 0 when it misses.
129
130----Errata----
131This field lists the known bugs that apply to the events. For the latest, up to date errata refer to
132
133Haswell:
134http://www.intel.com/content/dam/www/public/us/en/documents/specification-updates/4th-gen-core-family-mobile-specification-update.pdf
135
136IvyBridge:
137https://www-ssl.intel.com/content/dam/www/public/us/en/documents/specification-updates/3rd-gen-core-desktop-specification-update.pdf
138
139SandyBridge:
140https://www-ssl.intel.com/content/dam/www/public/us/en/documents/specification-updates/2nd-gen-core-family-mobile-specification-update.pdf
141
142----offcore----
143This field is specific to the json format. There is only 1 file for core and offcore events in this format. This field is set to 1 for offcore events
144and 0 for core events.
145
146---------------------
147For additional information:
148---------------------
149Intel Platform Monitoring Homepage
150http://software.intel.com/en-us/platform-monitoring/
151
152http://software.intel.com/en-us/articles/performance-monitoring-on-intel-xeon-processor-e5-family
153
154http://software.intel.com/en-us/articles/monitoring-integrated-memory-controller-requests-in-the-2nd-3rd-and-4th-generation-intel
155
156http://www.intel.com/content/dam/www/public/us/en/documents/specification-updates/4th-gen-core-family-desktop-specification-update.pdf
157
158---------------------
159For questions:
160---------------------
161email perfmon-discuss@lists.01.org
162
163---------------------
164Notices:
165---------------------
166INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL PRODUCTS. NO LICENSE, EXPRESS OR IMPLIED, BY ESTOPPEL OR OTHERWISE,
167TO ANY INTELLECTUAL PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. EXCEPT AS PROVIDED IN INTEL'S TERMS AND CONDITIONS OF SALE FOR SUCH
168PRODUCTS, INTEL ASSUMES NO LIABILITY WHATSOEVER AND INTEL DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY, RELATING TO SALE AND/OR USE OF
169INTEL PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT OF ANY
170PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.
171
172A "Mission Critical Application" is any application in which failure of the Intel Product could result, directly or indirectly, in
173personal injury or death. SHOULD YOU PURCHASE OR USE INTEL'S PRODUCTS FOR ANY SUCH MISSION CRITICAL APPLICATION, YOU SHALL INDEMNIFY
174AND HOLD INTEL AND ITS SUBSIDIARIES, SUBCONTRACTORS AND AFFILIATES, AND THE DIRECTORS, OFFICERS, AND EMPLOYEES OF EACH, HARMLESS AGAINST
175ALL CLAIMS COSTS, DAMAGES, AND EXPENSES AND REASONABLE ATTORNEYS' FEES ARISING OUT OF, DIRECTLY OR INDIRECTLY, ANY CLAIM OF PRODUCT
176LIABILITY, PERSONAL INJURY, OR DEATH ARISING IN ANY WAY OUT OF SUCH MISSION CRITICAL APPLICATION, WHETHER OR NOT INTEL OR ITS SUBCONTRACTOR
177WAS NEGLIGENT IN THE DESIGN, MANUFACTURE, OR WARNING OF THE INTEL PRODUCT OR ANY OF ITS PARTS.
178
179Intel may make changes to specifications and product descriptions at any time, without notice. Designers must not rely on the absence or
180characteristics of any features or instructions marked "reserved" or "undefined". Intel reserves these for future definition and shall have
181no responsibility whatsoever for conflicts or incompatibilities arising from future changes to them. The information here is subject to
182change without notice. Do not finalize a design with this information.
183
184The products described in this document may contain design defects or errors known as errata which may cause the product to deviate from
185published specifications. Current characterized errata are available on request.
186
187Contact your local Intel sales office or your distributor to obtain the latest specifications and before placing your product order.
188
189Copies of documents which have an order number and are referenced in this document, or other Intel literature, may be obtained by calling
1901-800-548-4725, or go to: http://www.intel.com/design/literature.htm
191
192Copyright � 2014 Intel Corporation. All rights reserved.