1 .\" Copyright (c) 2003-2008 Joseph Koshy. All rights reserved.
3 .\" Redistribution and use in source and binary forms, with or without
4 .\" modification, are permitted provided that the following conditions
6 .\" 1. Redistributions of source code must retain the above copyright
7 .\" notice, this list of conditions and the following disclaimer.
8 .\" 2. Redistributions in binary form must reproduce the above copyright
9 .\" notice, this list of conditions and the following disclaimer in the
10 .\" documentation and/or other materials provided with the distribution.
12 .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
13 .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
14 .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
15 .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
16 .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
17 .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
18 .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
19 .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
20 .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
21 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 .Nd measurement events for
41 Intel P4 PMCs are present in Intel
45 processors that use the
49 These PMCs are documented in
51 .%B "IA-32 Intel(R) Architecture Software Developer's Manual"
52 .%T "Volume 3: System Programming Guide"
53 .%N "Order Number 245472-012"
55 .%Q "Intel Corporation"
57 Further information about using these PMCs may be found in
59 .%B "IA-32 Intel(R) Architecture Optimization Guide"
61 .%N "Order Number 248966-009"
62 .%Q "Intel Corporation"
64 Some of these events are affected by processor errata described in
66 .%B "Intel(R) Pentium(R) 4 Processor Specification Update"
67 .%N "Document Number: 249199-059"
69 .%Q "Intel Corporation"
72 Intel Pentium 4 PMCs are 40 bits wide.
73 Each CPU contains 18 PMCs, divided into 4 groups with 4, 4, 4 and 6
75 On processors with hyperthreading support, PMC resources are shared
76 between logical processors.
77 These PMCs support the following capabilities:
78 .Bl -column "PMC_CAP_INTERRUPT" "Support"
79 .It Em Capability Ta Em Support
80 .It PMC_CAP_CASCADE Ta Yes
81 .It PMC_CAP_EDGE Ta Yes
82 .It PMC_CAP_INTERRUPT Ta Yes
83 .It PMC_CAP_INVERT Ta Yes
84 .It PMC_CAP_READ Ta Yes
85 .It PMC_CAP_PRECISE Ta Unimplemented
86 .It PMC_CAP_SYSTEM Ta Yes
87 .It PMC_CAP_TAGGING Ta Yes
88 .It PMC_CAP_THRESHOLD Ta Yes
89 .It PMC_CAP_USER Ta Yes
90 .It PMC_CAP_WRITE Ta Yes
94 Event specifiers for Intel P4 PMCs can have the following common
96 .Bl -tag -width indent
97 .It Li active= Ns Ar choice
98 (On P4 HTT CPUs) Filter event counting based on which logical
99 processors are active.
100 The allowed values of
104 .Bl -tag -width indent -compact
106 Count when either logical processor is active.
108 Count when both logical processors are active.
110 Count only when neither logical processor is active.
112 Count only when one logical processor is active.
118 Configure the PMC to cascade onto its partner.
120 .Sx "Cascading P4 PMCs"
121 below for more information.
123 Configure the counter to count false to true transitions of the threshold
125 This qualifier only takes effect if a threshold qualifier has also been
128 Configure the counter to increment only when the event count seen is
129 less than the threshold qualifier value specified.
130 .It Li mask= Ns Ar qualifier
131 Many event specifiers for Intel P4 PMCs need to be additionally
132 qualified using a mask qualifier.
133 The allowed syntax for these qualifiers is event specific and is
134 described along with the events.
136 Configure the PMC to count when the CPL of the processor is 0.
138 Select precise event based sampling.
139 Precise sampling is supported by the hardware for a limited set of
141 .It Li tag= Ns Ar value
142 Configure the PMC to tag the internal uop selected by the other
143 fields in this event specifier with value
145 This feature is used when cascading PMCs.
146 .It Li threshold= Ns Ar value
147 Configure the PMC to increment only when the event counts seen are
148 greater than the specified threshold value
151 Configure the PMC to count when the CPL of the processor is 1, 2 or 3.
158 qualifiers are specified, the default is to enable both.
160 On Intel Pentium 4 processors with HTT, events are
161 divided into two classes:
163 .Bl -tag -width indent -compact
165 are those where hardware can differentiate between events
166 generated on one logical processor from those generated on the
169 are those where hardware cannot differentiate between events
170 generated by multiple logical processors in a package.
173 Only TS events are allowed for use with process-mode PMCs on
176 The event specifiers supported by Intel P4 PMCs are:
177 .Bl -tag -width indent
178 .It Li p4-128bit-mmx-uop Op Li ,mask= Ns Ar flags
180 Count integer SIMD SSE2 instructions that operate on 128 bit SIMD
184 can take the following value (which is also the default):
186 .Bl -tag -width indent -compact
188 Count all uops operating on 128 bit SIMD integer operands in memory or
192 If an instruction contains more than one 128 bit MMX uop, then each
194 .It Li p4-64bit-mmx-uop Op Li ,mask= Ns Ar flags
196 Count MMX instructions that operate on 64 bit SIMD operands.
199 can take the following value (which is also the default):
201 .Bl -tag -width indent -compact
203 Count all uops operating on 64 bit SIMD integer operands in memory or
207 If an instruction contains more than one 64 bit MMX uop, then each
211 Count back-to-back bus cycles.
212 Further documentation for this event is unavailable.
215 Count bus-not-ready conditions.
216 Further documentation for this event is unavailable.
217 .It Li p4-bpu-fetch-request Op Li ,mask= Ns Ar qualifier
219 Count instruction fetch requests qualified by additional
222 At this point only one flag is supported:
224 .Bl -tag -width indent -compact
226 Count trace cache lookup misses.
229 The default qualifier is also
231 .It Li p4-branch-retired Op Li ,mask= Ns Ar flags
233 Counts retired branches.
236 is a list of the following
240 .Bl -tag -width indent -compact
242 Count branches not-taken and predicted.
244 Count branches not-taken and mis-predicted.
246 Count branches taken and predicted.
248 Count branches taken and mis-predicted.
251 The default qualifier counts all four kinds of branches.
252 .It Li p4-bsq-active-entries Op Li ,mask= Ns Ar qualifier
254 Count the number of entries (clipped at 15) currently active in the
260 separated set of the following flags:
262 .Bl -tag -width indent -compact
263 .It Li req-type0 , Li req-type1
264 Forms a 2-bit number used to select the request type encoding:
266 .Bl -tag -width indent -compact
268 reads excluding read invalidate
272 writes other than writebacks
279 is the MSB for this two bit number.
280 .It Li req-len0 , Li req-len1
281 Forms a two-bit number that specifies the request length encoding:
283 .Bl -tag -width indent -compact
294 is the MSB for this two bit number.
296 Count requests that are input or output requests.
298 Count requests that lock the bus.
299 .It Li req-lock-cache
300 Count requests that lock the cache.
301 .It Li req-split-type
302 Count requests that is a bus 8-byte chunk that is split across an
305 Count requests that are demand (not prefetches) if set.
306 Count requests that are prefetches if not set.
308 Count requests that are ordered.
309 .It Li mem-type0 , Li mem-type1 , Li mem-type2
310 Forms a 3-bit number that specifies a memory type encoding:
312 .Bl -tag -width indent -compact
327 is the MSB of this 3-bit number.
330 The default qualifier has all the above bits set.
332 Edge triggering using the
334 qualifier should not be used with this event when counting cycles.
335 .It Li p4-bsq-allocation Op Li ,mask= Ns Ar qualifier
337 Count allocations in the bus sequence unit according to the flags
342 separated set of the following flags:
344 .Bl -tag -width indent -compact
345 .It Li req-type0 , Li req-type1
346 Forms a 2-bit number used to select the request type encoding:
348 .Bl -tag -width indent -compact
350 reads excluding read invalidate
354 writes other than writebacks
361 is the MSB for this two bit number.
362 .It Li req-len0 , Li req-len1
363 Forms a two-bit number that specifies the request length encoding:
365 .Bl -tag -width indent -compact
376 is the MSB for this two bit number.
378 Count requests that are input or output requests.
380 Count requests that lock the bus.
381 .It Li req-lock-cache
382 Count requests that lock the cache.
383 .It Li req-split-type
384 Count requests that is a bus 8-byte chunk that is split across an
387 Count requests that are demand (not prefetches) if set.
388 Count requests that are prefetches if not set.
390 Count requests that are ordered.
391 .It Li mem-type0 , Li mem-type1 , Li mem-type2
392 Forms a 3-bit number that specifies a memory type encoding:
394 .Bl -tag -width indent -compact
409 is the MSB of this 3-bit number.
412 The default qualifier has all the above bits set.
414 This event is usually used along with the
416 qualifier to avoid multiple counting.
417 .It Li p4-bsq-cache-reference Op Li ,mask= Ns Ar qualifier
419 Count cache references as seen by the bus unit (2nd or 3rd level
425 separated list of the following keywords:
427 .Bl -tag -width indent -compact
429 Count 2nd level cache hits in the shared state.
431 Count 2nd level cache hits in the exclusive state.
433 Count 2nd level cache hits in the modified state.
435 Count 3rd level cache hits in the shared state.
437 Count 3rd level cache hits in the exclusive state.
439 Count 3rd level cache hits in the modified state.
441 Count 2nd level cache misses.
443 Count 3rd level cache misses.
445 Count write-back lookups from the data access cache that miss the 2nd
449 The default is to count all the above events.
450 .It Li p4-execution-event Op Li ,mask= Ns Ar flags
452 Count the retirement of tagged uops selected through the execution
456 can contain the following strings separated by
460 .Bl -tag -width indent -compact
461 .It Li nbogus0 , Li nbogus1 , Li nbogus2 , Li nbogus3
462 The marked uops are not bogus.
463 .It Li bogus0 , Li bogus1 , Li bogus2 , Li bogus3
464 The marked uops are bogus.
467 This event requires additional (upstream) events to be allocated to
468 perform the desired uop tagging.
469 The default is to set all the above flags.
470 This event can be used for precise event based sampling.
471 .It Li p4-front-end-event Op Li ,mask= Ns Ar flags
473 Count the retirement of tagged uops selected through the front-end
477 can contain the following strings separated by
481 .Bl -tag -width indent -compact
483 The marked uops are not bogus.
485 The marked uops are bogus.
488 This event requires additional (upstream) events to be allocated to
489 perform the desired uop tagging.
490 The default is to select both kinds of events.
491 This event can be used for precise event based sampling.
492 .It Li p4-fsb-data-activity Op Li ,mask= Ns Ar flags
494 Count each DBSY or DRDY event selected by qualifier
500 separated set of the following flags:
502 .Bl -tag -width indent -compact
504 Count when this processor is driving data onto the bus.
506 Count when this processor is reading data from the bus.
508 Count when data is on the bus but not being sampled by this processor.
510 Count when this processor reserves the bus for use in the next cycle
511 in order to drive data.
513 Count when some agent reserves the bus for use in the next bus cycle
514 to drive data that this processor will sample.
516 Count when some agent reserves the bus for use in the next bus cycle
517 to drive data that this processor will not sample.
524 are mutually exclusive.
529 are mutually exclusive.
530 The default value for
533 .Dq Li drdy-drv+drdy-own+dbsy-drv+dbsy-own .
534 .It Li p4-global-power-events Op Li ,mask= Ns Ar flags
536 Count cycles during which the processor is not stopped.
539 can take the following value (which is also the default):
541 .Bl -tag -width indent -compact
543 Count cycles when the processor is active.
546 .It Li p4-instr-retired Op Li ,mask= Ns Ar flags
548 Count instructions retired during a clock cycle.
551 comprises of the following strings separated by
555 .Bl -tag -width indent -compact
557 Count non-bogus instructions that are not tagged.
559 Count non-bogus instructions that are tagged.
561 Count bogus instructions that are not tagged.
563 Count bogus instructions that are tagged.
566 The default qualifier counts all the above kinds of instructions.
567 .It Li p4-ioq-active-entries Xo
568 .Op Li ,mask= Ns Ar qualifier
569 .Op Li ,busreqtype= Ns Ar req-type
572 Count the number of entries (clipped at 15) in the IOQ that are
574 The event masks are specified by qualifier
583 separated set of the following flags:
585 .Bl -tag -width indent -compact
591 Count entries accessing un-cacheable memory.
593 Count entries accessing write-combining memory.
595 Count entries accessing write-through memory.
597 Count entries accessing write-protected memory
599 Count entries accessing write-back memory.
601 Count store requests driven by the processor (i.e., not by other
602 processors or by DMA).
604 Count store requests driven by other processors or by DMA.
606 Include hardware and software prefetch requests in the count.
609 The default value for
611 is to enable all the above flags.
615 qualifier is a 5-bit number can be additionally used to select a
616 specific bus request type.
621 qualifier should not be used when counting cycles with this event.
622 The exact behavior of this event depends on the processor revision.
623 .It Li p4-ioq-allocation Xo
624 .Op Li ,mask= Ns Ar qualifier
625 .Op Li ,busreqtype= Ns Ar req-type
628 Count various types of transactions on the bus matching the flags set
638 separated set of the following flags:
640 .Bl -tag -width indent -compact
646 Count entries accessing un-cacheable memory.
648 Count entries accessing write-combining memory.
650 Count entries accessing write-through memory.
652 Count entries accessing write-protected memory
654 Count entries accessing write-back memory.
656 Count store requests driven by the processor (i.e., not by other
657 processors or by DMA).
659 Count store requests driven by other processors or by DMA.
661 Include hardware and software prefetch requests in the count.
664 The default value for
666 is to enable all the above flags.
670 qualifier is a 5-bit number can be additionally used to select a
671 specific bus request type.
676 qualifier is normally used with this event to prevent multiple
678 The exact behavior of this event depends on the processor revision.
679 .It Li p4-itlb-reference Op mask= Ns Ar qualifier
681 Count translations using the instruction translation look-aside
685 argument is a list of the following strings separated by
689 .Bl -tag -width indent -compact
695 Count un-cacheable ITLB hits.
700 is specified the default is to count all the three kinds of ITLB
702 .It Li p4-load-port-replay Op Li ,mask= Ns Ar qualifier
704 Count replayed events at the load port.
707 can take on one value:
709 .Bl -tag -width indent -compact
714 The default value for
718 .It Li p4-mispred-branch-retired Op Li ,mask= Ns Ar flags
720 Count mispredicted IA-32 branch instructions.
723 can take the following value (which is also the default):
725 .Bl -tag -width indent -compact
727 Count non-bogus retired branch instructions.
729 .It Li p4-machine-clear Op Li ,mask= Ns Ar flags
731 Count the number of pipeline clears seen by the processor.
734 is a list of the following strings separated by
738 .Bl -tag -width indent -compact
740 Count for a portion of the many cycles when the machine is being
741 cleared for any reason.
743 Count machine clears due to memory ordering issues.
745 Count machine clears due to self-modifying code.
750 to get a count of occurrences of machine clears.
751 The default qualifier is
753 .It Li p4-memory-cancel Op Li ,mask= Ns Ar event-list
755 Count the canceling of various kinds of requests in the data cache
756 address control unit of the CPU.
759 is a list of the following strings separated by
763 .Bl -tag -width indent -compact
765 Requests cancelled because no store request buffer was available.
767 Requests that conflict due to 64K aliasing.
772 is not specified, then the default is to count both kinds of events.
773 .It Li p4-memory-complete Op Li ,mask= Ns Ar event-list
775 Count the completion of load split, store split, un-cacheable split and
776 un-cacheable load operations selected by qualifier
782 separated list of the following flags:
784 .Bl -tag -width indent -compact
786 Count load splits completed, excluding loads from un-cacheable or
787 write-combining areas.
789 Count any split stores completed.
792 The default is to count both kinds of operations.
793 .It Li p4-mob-load-replay Op Li ,mask= Ns Ar qualifier
795 Count load replays triggered by the memory order buffer.
800 separated list of the following flags:
802 .Bl -tag -width indent -compact
804 Count replays because of unknown store addresses.
806 Count replays because of unknown store data.
808 Count replays because of partially overlapped data accesses between
809 load and store operations.
811 Count replays because of mismatches in the lower 4 bits of load and
815 The default qualifier is
816 .Ar no-sta+no-std+partial-data+unalgn-addr .
817 .It Li p4-packed-dp-uop Op Li ,mask= Ns Ar flags
819 Count packed double-precision uops.
822 can take the following value (which is also the default):
824 .Bl -tag -width indent -compact
826 Count all uops operating on packed double-precision operands.
828 .It Li p4-packed-sp-uop Op Li ,mask= Ns Ar flags
830 Count packed single-precision uops.
833 can take the following value (which is also the default):
835 .Bl -tag -width indent -compact
837 Count all uops operating on packed single-precision operands.
839 .It Li p4-page-walk-type Op Li ,mask= Ns Ar qualifier
841 Count page walks performed by the page miss handler.
846 separated list of the following keywords:
848 .Bl -tag -width indent -compact
850 Count page walks for data TLB misses.
852 Count page walks for instruction TLB misses.
855 The default value for
858 .Dq Li dtmiss+itmiss .
859 .It Li p4-replay-event Op Li ,mask= Ns Ar flags
861 Count the retirement of tagged uops selected through the replay
867 separated set of the following strings:
869 .Bl -tag -width indent -compact
871 The marked uops are not bogus.
873 The marked uops are bogus.
876 This event requires additional (upstream) events to be allocated to
877 perform the desired uop tagging.
878 The default qualifier counts both kinds of uops.
879 This event can be used for precise event based sampling.
880 .It Li p4-resource-stall Op Li ,mask= Ns Ar flags
882 Count the occurrence or latency of stalls in the allocator.
885 can take the following value (which is also the default):
887 .Bl -tag -width indent -compact
889 A stall due to the lack of store buffers.
893 Count different types of responses.
894 Further documentation on this event is not available.
895 .It Li p4-retired-branch-type Op Li ,mask= Ns Ar flags
897 Count branches retired.
902 separated list of strings:
904 .Bl -tag -width indent -compact
906 Count conditional jumps.
908 Count direct and indirect call branches.
910 Count return branches.
912 Count returns, indirect calls or indirect jumps.
915 The default qualifier counts all the above branch types.
916 .It Li p4-retired-mispred-branch-type Op Li ,mask= Ns Ar flags
918 Count mispredicted branches retired.
923 separated list of strings:
925 .Bl -tag -width indent -compact
927 Count conditional jumps.
929 Count indirect call branches.
931 Count return branches.
933 Count returns, indirect calls or indirect jumps.
936 The default qualifier counts all the above branch types.
937 .It Li p4-scalar-dp-uop Op Li ,mask= Ns Ar flags
939 Count the number of scalar double-precision uops.
942 can take the following value (which is also the default):
944 .Bl -tag -width indent -compact
946 Count the number of scalar double-precision uops.
948 .It Li p4-scalar-sp-uop Op Li ,mask= Ns Ar flags
950 Count the number of scalar single-precision uops.
953 can take the following value (which is also the default):
955 .Bl -tag -width indent -compact
957 Count all uops operating on scalar single-precision operands.
962 Further documentation on this event is not available.
963 .It Li p4-sse-input-assist Op Li ,mask= Ns Ar flags
965 Count the number of times an assist is required to handle problems
966 with the operands for SSE and SSE2 operations.
969 can take the following value (which is also the default):
971 .Bl -tag -width indent -compact
973 Count assists for all SSE and SSE2 uops.
975 .It Li p4-store-port-replay Op Li ,mask= Ns Ar qualifier
977 Count events replayed at the store port.
980 can take on one value:
982 .Bl -tag -width indent -compact
987 The default value for
991 .It Li p4-tc-deliver-mode Op Li ,mask= Ns Ar qualifier
993 Count the duration in cycles of operating modes of the trace cache and
995 The desired operating mode is selected by
997 which is a list of the following strings separated by
1001 .Bl -tag -width indent -compact
1003 Both logical processors are in deliver mode.
1005 Logical processor 0 is in deliver mode while logical processor 1 is in
1008 Logical processor 0 is in deliver mode while logical processor 1 is
1009 halted, or in machine clear, or transitioning to a long microcode
1012 Logical processor 0 is in build mode while logical processor 1 is in
1015 Both logical processors are in build mode.
1017 Logical processor 0 is in build mode while logical processor 1 is
1018 halted, or in machine clear or transitioning to a long microcode
1021 Logical processor 0 is halted, or in machine clear or transitioning to
1022 a long microcode flow while logical processor 1 is in deliver mode.
1024 Logical processor 0 is halted, or in machine clear or transitioning to
1025 a long microcode flow while logical processor 1 is in build mode.
1028 If there is only one logical processor in the processor package then
1029 the qualifier for logical processor 1 is ignored.
1030 If no qualifier is specified, the default qualifier is
1031 .Dq Li DD+DB+DI+BD+BB+BI+ID+IB .
1032 .It Li p4-tc-ms-xfer Op Li ,mask= Ns Ar flags
1034 Count the number of times uop delivery changed from the trace cache to
1038 can take the following value (which is also the default):
1040 .Bl -tag -width indent -compact
1042 Count TC to MS transfers.
1044 .It Li p4-uop-queue-writes Op Li ,mask= Ns Ar flags
1046 Count the number of valid uops written to the uop queue.
1049 is a list of the following strings, separated by
1053 .Bl -tag -width indent -compact
1054 .It Li from-tc-build
1055 Count uops being written from the trace cache in build mode.
1056 .It Li from-tc-deliver
1057 Count uops being written from the trace cache in deliver mode.
1059 Count uops being written from microcode ROM.
1062 The default qualifier counts all the above kinds of uops.
1063 .It Li p4-uop-type Op Li ,mask= Ns Ar flags
1065 This event is used in conjunction with the front-end at-retirement
1066 mechanism to tag load and store uops.
1069 comprises the following strings separated by
1073 .Bl -tag -width indent -compact
1075 Mark uops that are load operations.
1077 Mark uops that are store operations.
1080 The default qualifier counts both kinds of uops.
1081 .It Li p4-uops-retired Op Li ,mask= Ns Ar flags
1083 Count uops retired during a clock cycle.
1086 comprises the following strings separated by
1090 .Bl -tag -width indent -compact
1092 Count marked uops that are not bogus.
1094 Count marked uops that are bogus.
1097 The default qualifier counts both kinds of uops.
1098 .It Li p4-wc-buffer Op Li ,mask= Ns Ar flags
1100 Count write-combining buffer operations.
1103 contains the following strings separated by
1107 .Bl -tag -width indent -compact
1109 WC buffer evictions due to any cause.
1110 .It Li wcb-full-evict
1111 WC buffer evictions due to no WC buffer being available.
1114 The default qualifier counts both kinds of evictions.
1115 .It Li p4-x87-assist Op Li ,mask= Ns Ar flags
1117 Count the retirement of x87 instructions that required special
1121 contains the following strings separated by
1125 .Bl -tag -width indent -compact
1127 Count instructions that saw an FP stack underflow.
1129 Count instructions that saw an FP stack overflow.
1131 Count instructions that saw an x87 output overflow.
1133 Count instructions that saw an x87 output underflow.
1135 Count instructions that needed an x87 input assist.
1138 The default qualifier counts all the above types of instruction
1140 .It Li p4-x87-fp-uop Op Li ,mask= Ns Ar flags
1142 Count x87 floating-point uops.
1145 can take the following value (which is also the default):
1147 .Bl -tag -width indent -compact
1149 Count all x87 floating-point uops.
1152 If an instruction contains more than one x87 floating-point uops, then
1153 all x87 floating-point uops will be counted.
1154 This event does not count x87 floating-point data movement operations.
1155 .It Li p4-x87-simd-moves-uop Op Li ,mask= Ns Ar flags
1157 Count each x87 FPU, MMX, SSE, or SSE2 uops that load data or store
1158 data or perform register-to-register moves.
1159 This event does not count integer move uops.
1162 may contain the following keywords separated by
1166 .Bl -tag -width indent -compact
1168 Count all x87 and SIMD store and move uops.
1170 Count all x87 and SIMD load uops.
1173 The default is to count all uops.
1175 This event may be affected by processor errata N43.
1177 .Ss "Cascading P4 PMCs"
1178 PMC cascading support is currently poorly implemented.
1179 While individual event counters may be allocated with a
1181 qualifier, the current API does not offer the ability
1182 to name and allocate all the resources needed for a
1183 cascaded event counter pair in a single operation.
1184 .Ss "Precise Event Based Sampling"
1185 Support for precise event based sampling is currently
1187 .Ss Event Name Aliases
1188 The following table shows the mapping between the PMC-independent
1189 aliases supported by
1191 and the underlying hardware events used.
1192 .Bl -column "branch-mispredicts" "Description"
1193 .It Em Alias Ta Em Event
1194 .It Li branches Ta Li p4-branch-retired,mask=mmtp+mmtm
1195 .It Li branch-mispredicts Ta Li p4-mispred-branch-retired
1196 .It Li dc-misses Ta (unsupported)
1197 .It Li ic-misses Ta (unsupported)
1198 .It Li instructions Ta Li p4-instr-retired,mask=nbogusntag+nbogustag
1199 .It Li interrupts Ta Li (unsupported)
1200 .It Li unhalted-cycles Ta Li p4-global-power-events
1219 library first appeared in
1224 library was written by
1226 .Aq jkoshy@FreeBSD.org .
projects / FreeBSD / releng / 9.2.git / blob