xref: /freebsd/lib/libpmc/pmc.core.3 (revision 82d6d46d0d1dbdaa32c754f97e94a772d01cb26c)
1.\" Copyright (c) 2008 Joseph Koshy.  All rights reserved.
2.\"
3.\" Redistribution and use in source and binary forms, with or without
4.\" modification, are permitted provided that the following conditions
5.\" are met:
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.
11.\"
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
22.\" SUCH DAMAGE.
23.\"
24.Dd November 12, 2008
25.Dt PMC.CORE 3
26.Os
27.Sh NAME
28.Nm pmc.core
29.Nd measurement events for
30.Tn Intel
31.Tn Core Solo
32and
33.Tn Core Duo
34family CPUs
35.Sh LIBRARY
36.Lb libpmc
37.Sh SYNOPSIS
38.In pmc.h
39.Sh DESCRIPTION
40.Tn Intel
41.Tn "Core Solo"
42and
43.Tn "Core Duo"
44CPUs contain PMCs conforming to version 1 of the
45.Tn Intel
46performance measurement architecture.
47.Pp
48These PMCs are documented in
49.Rs
50.%B IA-32 Intel\(rg Architecture Software Developer's Manual
51.%T Volume 3: System Programming Guide
52.%N Order Number 253669-027US
53.%D July 2008
54.%Q Intel Corporation
55.Re
56.Ss PMC Features
57CPUs conforming to version 1 of the
58.Tn Intel
59performance measurement architecture contain two programmable PMCs of
60class
61.Li PMC_CLASS_IAP .
62The PMCs are 40 bits width and offer the following capabilities:
63.Bl -column "PMC_CAP_INTERRUPT" "Support"
64.It Em Capability Ta Em Support
65.It PMC_CAP_CASCADE Ta \&No
66.It PMC_CAP_EDGE Ta Yes
67.It PMC_CAP_INTERRUPT Ta Yes
68.It PMC_CAP_INVERT Ta Yes
69.It PMC_CAP_READ Ta Yes
70.It PMC_CAP_PRECISE Ta \&No
71.It PMC_CAP_SYSTEM Ta Yes
72.It PMC_CAP_TAGGING Ta \&No
73.It PMC_CAP_THRESHOLD Ta Yes
74.It PMC_CAP_USER Ta Yes
75.It PMC_CAP_WRITE Ta Yes
76.El
77.Ss Event Qualifiers
78Event specifiers for these PMCs support the following common
79qualifiers:
80.Bl -tag -width indent
81.It Li cmask= Ns Ar value
82Configure the PMC to increment only if the number of configured
83events measured in a cycle is greater than or equal to
84.Ar value .
85.It Li edge
86Configure the PMC to count the number of de-asserted to asserted
87transitions of the conditions expressed by the other qualifiers.
88If specified, the counter will increment only once whenever a
89condition becomes true, irrespective of the number of clocks during
90which the condition remains true.
91.It Li inv
92Invert the sense of comparison when the
93.Dq Li cmask
94qualifier is present, making the counter increment when the number of
95events per cycle is less than the value specified by the
96.Dq Li cmask
97qualifier.
98.It Li os
99Configure the PMC to count events happening at processor privilege
100level 0.
101.It Li usr
102Configure the PMC to count events occurring at privilege levels 1, 2
103or 3.
104.El
105.Pp
106If neither of the
107.Dq Li os
108or
109.Dq Li usr
110qualifiers are specified, the default is to enable both.
111.Pp
112Events that require core-specificity to be specified use a
113additional qualifier
114.Dq Li core= Ns Ar value ,
115where argument
116.Ar value
117is one of:
118.Bl -tag -width indent -compact
119.It Li all
120Measure event conditions on all cores.
121.It Li this
122Measure event conditions on this core.
123.El
124The default is
125.Dq Li this .
126.Pp
127Events that require an agent qualifier to be specified use an
128additional qualifier
129.Dq Li agent= Ns value ,
130where argument
131.Ar value
132is one of:
133.Bl -tag -width indent -compact
134.It Li this
135Measure events associated with this bus agent.
136.It Li any
137Measure events caused by any bus agent.
138.El
139The default is
140.Dq Li this .
141.Pp
142Events that require a hardware prefetch qualifier to be specified use an
143additional qualifier
144.Dq Li prefetch= Ns Ar value ,
145where argument
146.Ar value
147is one of:
148.Bl -tag -width "exclude" -compact
149.It Li both
150Include all prefetches.
151.It Li only
152Only count hardware prefetches.
153.It Li exclude
154Exclude hardware prefetches.
155.El
156The default is
157.Dq Li both .
158.Pp
159Events that require a cache coherence qualifier to be specified use an
160additional qualifier
161.Dq Li cachestate= Ns Ar value ,
162where argument
163.Ar value
164contains one or more of the following letters:
165.Bl -tag -width indent -compact
166.It Li e
167Count cache lines in the exclusive state.
168.It Li i
169Count cache lines in the invalid state.
170.It Li m
171Count cache lines in the modified state.
172.It Li s
173Count cache lines in the shared state.
174.El
175The default is
176.Dq Li eims .
177.Ss Event Specifiers
178The following event names are case insensitive.
179Whitespace, hyphens and underscore characters in these names are
180ignored.
181.Pp
182Core PMCs support the following events:
183.Bl -tag -width indent
184.It Li BAClears
185.Pq Event E6H , Umask 00H
186The number of BAClear conditions asserted.
187.It Li BTB_Misses
188.Pq Event E2H , Umask 00H
189The number of branches for which the branch table buffer did not
190produce a prediction.
191.It Li Br_BAC_Missp_Exec
192.Pq Event 8AH , Umask 00H
193The number of branch instructions executed that were mispredicted at
194the front end.
195.It Li Br_Bogus
196.Pq Event E4H , Umask 00H
197The number of bogus branches.
198.It Li Br_Call_Exec
199.Pq Event 92H , Umask 00H
200The number of
201.Li CALL
202instructions executed.
203.It Li Br_Call_Missp_Exec
204.Pq Event 93H , Umask 00H
205The number of
206.Li CALL
207instructions executed that were mispredicted.
208.It Li Br_Cnd_Exec
209.Pq Event 8BH , Umask 00H
210The number of conditional branch instructions executed.
211.It Li Br_Cnd_Missp_Exec
212.Pq Event 8CH , Umask 00H
213The number of conditional branch instructions executed that were mispredicted.
214.It Li Br_Ind_Call_Exec
215.Pq Event 94H , Umask 00H
216The number of indirect
217.Li CALL
218instructions executed.
219.It Li Br_Ind_Exec
220.Pq Event 8DH , Umask 00H
221The number of indirect branches executed.
222.It Li Br_Ind_Missp_Exec
223.Pq Event 8EH , Umask 00H
224The number of indirect branch instructions executed that were mispredicted.
225.It Li Br_Inst_Exec
226.Pq Event 88H , Umask 00H
227The number of branch instructions executed including speculative branches.
228.It Li Br_Instr_Decoded
229.Pq Event E0H , Umask 00H
230The number of branch instructions decoded.
231.It Li Br_Instr_Ret
232.Pq Event C4H , Umask 00H
233.Pq Alias Qq "Branch Instruction Retired"
234The number of branch instructions retired.
235This is an architectural performance event.
236.It Li Br_MisPred_Ret
237.Pq Event C5H , Umask 00H
238.Pq Alias Qq "Branch Misses Retired"
239The number of mispredicted branch instructions retired.
240This is an architectural performance event.
241.It Li Br_MisPred_Taken_Ret
242.Pq Event CAH , Umask 00H
243The number of taken and mispredicted branches retired.
244.It Li Br_Missp_Exec
245.Pq Event 89H , Umask 00H
246The number of branch instructions executed and mispredicted at
247execution including branches that were not predicted.
248.It Li Br_Ret_BAC_Missp_Exec
249.Pq Event 91H , Umask 00H
250The number of return branch instructions that were mispredicted at the
251front end.
252.It Li Br_Ret_Exec
253.Pq Event 8FH , Umask 00H
254The number of return branch instructions executed.
255.It Li Br_Ret_Missp_Exec
256.Pq Event 90H , Umask 00H
257The number of return branch instructions executed that were mispredicted.
258.It Li Br_Taken_Ret
259.Pq Event C9H , Umask 00H
260The number of taken branches retired.
261.It Li Bus_BNR_Clocks
262.Pq Event 61H , Umask 00H
263The number of external bus cycles while BNR (bus not ready) was asserted.
264.It Li Bus_DRDY_Clocks Op ,agent= Ns Ar agent
265.Pq Event 62H , Umask 00H
266The number of external bus cycles while DRDY was asserted.
267.It Li Bus_Data_Rcv
268.Pq Event 64H , Umask 40H
269.\" XXX Using the description in Core2 PMC documentation.
270The number of cycles during which the processor is busy receiving data.
271.It Li Bus_Locks_Clocks Op ,core= Ns Ar core
272.Pq Event 63H
273The number of external bus cycles while the bus lock signal was asserted.
274.It Li Bus_Not_In_Use Op ,core= Ns Ar core
275.Pq Event 7DH
276The number of cycles when there is no transaction from the core.
277.It Li Bus_Req_Outstanding Xo
278.Op ,agent= Ns Ar agent
279.Op ,core= Ns Ar core
280.Xc
281.Pq Event 60H
282The weighted cycles of cacheable bus data read requests
283from the data cache unit or hardware prefetcher.
284.It Li Bus_Snoop_Stall
285.Pq Event 7EH , Umask 00H
286The number bus cycles while a bus snoop is stalled.
287.It Li Bus_Snoops Xo
288.Op ,agent= Ns Ar agent
289.Op ,cachestate= Ns Ar mesi
290.Xc
291.Pq Event 77H
292.\" XXX Using the description in Core2 PMC documentation.
293The number of snoop responses to bus transactions.
294.It Li Bus_Trans_Any Op ,agent= Ns Ar agent
295.Pq Event 70H
296The number of completed bus transactions.
297.It Li Bus_Trans_Brd Op ,core= Ns Ar core
298.Pq Event 65H
299The number of read bus transactions.
300.It Li Bus_Trans_Burst Op ,agent= Ns Ar agent
301.Pq Event 6EH
302The number of completed burst transactions.
303Retried transactions may be counted more than once.
304.It Li Bus_Trans_Def Op ,core= Ns Ar core
305.Pq Event 6DH
306The number of completed deferred transactions.
307.It Li Bus_Trans_IO Xo
308.Op ,agent= Ns Ar agent
309.Op ,core= Ns Ar core
310.Xc
311.Pq Event 6CH
312The number of completed I/O transactions counting both reads and
313writes.
314.It Li Bus_Trans_Ifetch Xo
315.Op ,agent= Ns Ar agent
316.Op ,core= Ns Ar core
317.Xc
318.Pq Event 68H
319Completed instruction fetch transactions.
320.It Li Bus_Trans_Inval Xo
321.Op ,agent= Ns Ar agent
322.Op ,core= Ns Ar core
323.Xc
324.Pq Event 69H
325The number completed invalidate transactions.
326.It Li Bus_Trans_Mem Op ,agent= Ns Ar agent
327.Pq Event 6FH
328The number of completed memory transactions.
329.It Li Bus_Trans_P Xo
330.Op ,agent= Ns Ar agent
331.Op ,core= Ns Ar core
332.Xc
333.Pq Event 6BH
334The number of completed partial transactions.
335.It Li Bus_Trans_Pwr Xo
336.Op ,agent= Ns Ar agent
337.Op ,core= Ns Ar core
338.Xc
339.Pq Event 6AH
340The number of completed partial write transactions.
341.It Li Bus_Trans_RFO Xo
342.Op ,agent= Ns Ar agent
343.Op ,core= Ns Ar core
344.Xc
345.Pq Event 66H
346The number of completed read-for-ownership transactions.
347.It Li Bus_Trans_WB Op ,agent= Ns Ar agent
348.Pq Event 67H
349The number of completed write-back transactions from the data cache
350unit, excluding L2 write-backs.
351.It Li Cycles_Div_Busy
352.Pq Event 14H , Umask 00H
353The number of cycles the divider is busy.
354The event is only available on PMC0.
355.It Li Cycles_Int_Masked
356.Pq Event C6H , Umask 00H
357The number of cycles while interrupts were disabled.
358.It Li Cycles_Int_Pending_Masked
359.Pq Event C7H , Umask 00H
360The number of cycles while interrupts were disabled and interrupts
361were pending.
362.It Li DCU_Snoop_To_Share Op ,core= Ns core
363.Pq Event 78H
364The number of data cache unit snoops to L1 cache lines in the shared
365state.
366.It Li DCache_Cache_Lock Op ,cachestate= Ns Ar mesi
367.\" XXX needs clarification
368.Pq Event 42H
369The number of cacheable locked read operations to invalid state.
370.It Li DCache_Cache_LD Op ,cachestate= Ns Ar mesi
371.Pq Event 40H
372The number of cacheable L1 data read operations.
373.It Li DCache_Cache_ST Op ,cachestate= Ns Ar mesi
374.Pq Event 41H
375The number cacheable L1 data write operations.
376.It Li DCache_M_Evict
377.Pq Event 47H , Umask 00H
378The number of M state data cache lines that were evicted.
379.It Li DCache_M_Repl
380.Pq Event 46H , Umask 00H
381The number of M state data cache lines that were allocated.
382.It Li DCache_Pend_Miss
383.Pq Event 48H , Umask 00H
384The weighted cycles an L1 miss was outstanding.
385.It Li DCache_Repl
386.Pq Event 45H , Umask 0FH
387The number of data cache line replacements.
388.It Li Data_Mem_Cache_Ref
389.Pq Event 44H , Umask 02H
390The number of cacheable read and write operations to L1 data cache.
391.It Li Data_Mem_Ref
392.Pq Event 43H , Umask 01H
393The number of L1 data reads and writes, both cacheable and
394un-cacheable.
395.It Li Dbus_Busy Op ,core= Ns Ar core
396.Pq Event 22H
397The number of core cycles during which the data bus was busy.
398.It Li Dbus_Busy_Rd Op ,core= Ns Ar core
399.Pq Event 23H
400The number of cycles during which the data bus was busy transferring
401data to a core.
402.It Li Div
403.Pq Event 13H , Umask 00H
404The number of divide operations including speculative operations for
405integer and floating point divides.
406This event can only be counted on PMC1.
407.It Li Dtlb_Miss
408.Pq Event 49H , Umask 00H
409The number of data references that missed the TLB.
410.It Li ESP_Uops
411.Pq Event D7H , Umask 00H
412The number of ESP folding instructions decoded.
413.It Li EST_Trans Op ,trans= Ns Ar transition
414.Pq Event 3AH
415Count the number of Intel Enhanced SpeedStep transitions.
416The argument
417.Ar transition
418can be one of the following values:
419.Bl -tag -width indent -compact
420.It Li any
421(Umask 00H) Count all transitions.
422.It Li frequency
423(Umask 01H) Count frequency transitions.
424.El
425The default is
426.Dq Li any .
427.It Li FP_Assist
428.Pq Event 11H , Umask 00H
429The number of floating point operations that required microcode
430assists.
431The event is only available on PMC1.
432.It Li FP_Comp_Instr_Ret
433.Pq Event C1H , Umask 00H
434The number of X87 floating point compute instructions retired.
435The event is only available on PMC0.
436.It Li FP_Comps_Op_Exe
437.Pq Event 10H , Umask 00H
438The number of floating point computational instructions executed.
439.It Li FP_MMX_Trans
440.Pq Event CCH , Umask 01H
441The number of transitions from X87 to MMX.
442.It Li Fused_Ld_Uops_Ret
443.Pq Event DAH , Umask 01H
444The number of fused load uops retired.
445.It Li Fused_St_Uops_Ret
446.Pq Event DAH , Umask 02H
447The number of fused store uops retired.
448.It Li Fused_Uops_Ret
449.Pq Event DAH , Umask 00H
450The number of fused uops retired.
451.It Li HW_Int_Rx
452.Pq Event C8H , Umask 00H
453The number of hardware interrupts received.
454.It Li ICache_Misses
455.Pq Event 81H , Umask 00H
456The number of instruction fetch misses in the instruction cache and
457streaming buffers.
458.It Li ICache_Reads
459.Pq Event 80H , Umask 00H
460The number of instruction fetches from the instruction cache and
461streaming buffers counting both cacheable and un-cacheable fetches.
462.It Li IFU_Mem_Stall
463.Pq Event 86H , Umask 00H
464The number of cycles the instruction fetch unit was stalled while
465waiting for data from memory.
466.It Li ILD_Stall
467.Pq Event 87H , Umask 00H
468The number of instruction length decoder stalls.
469.It Li ITLB_Misses
470.Pq Event 85H , Umask 00H
471The number of instruction TLB misses.
472.It Li Instr_Decoded
473.Pq Event D0H , Umask 00H
474The number of instructions decoded.
475.It Li Instr_Ret
476.Pq Event C0H , Umask 00H
477.Pq Alias Qq "Instruction Retired"
478The number of instructions retired.
479This is an architectural performance event.
480.It Li L1_Pref_Req
481.Pq Event 4FH , Umask 00H
482The number of L1 prefetch request due to data cache misses.
483.It Li L2_ADS Op ,core= Ns core
484.Pq Event 21H
485The number of L2 address strobes.
486.It Li L2_IFetch Xo
487.Op ,cachestate= Ns Ar mesi
488.Op ,core= Ns Ar core
489.Xc
490.Pq Event 28H
491The number of instruction fetches by the instruction fetch unit from
492L2 cache including speculative fetches.
493.It Li L2_LD Xo
494.Op ,cachestate= Ns Ar mesi
495.Op ,core= Ns Ar core
496.Xc
497.Pq Event 29H
498The number of L2 cache reads.
499.It Li L2_Lines_In Xo
500.Op ,core= Ns Ar core
501.Op ,prefetch= Ns Ar prefetch
502.Xc
503.Pq Event 24H
504The number of L2 cache lines allocated.
505.It Li L2_Lines_Out Xo
506.Op ,core= Ns Ar core
507.Op ,prefetch= Ns Ar prefetch
508.Xc
509.Pq Event 26H
510The number of L2 cache lines evicted.
511.It Li L2_M_Lines_In Op ,core= Ns Ar core
512.Pq Event 25H
513The number of L2 M state cache lines allocated.
514.It Li L2_M_Lines_Out Xo
515.Op ,core= Ns Ar core
516.Op ,prefetch= Ns Ar prefetch
517.Xc
518.Pq Event 27H
519The number of L2 M state cache lines evicted.
520.It Li L2_No_Request_Cycles Xo
521.Op ,cachestate= Ns Ar mesi
522.Op ,core= Ns Ar core
523.Op ,prefetch= Ns Ar prefetch
524.Xc
525.Pq Event 32H
526The number of cycles there was no request to access L2 cache.
527.It Li L2_Reject_Cycles Xo
528.Op ,cachestate= Ns Ar mesi
529.Op ,core= Ns Ar core
530.Op ,prefetch= Ns Ar prefetch
531.Xc
532.Pq Event 30H
533The number of cycles the L2 cache was busy and rejecting new requests.
534.It Li L2_Rqsts Xo
535.Op ,cachestate= Ns Ar mesi
536.Op ,core= Ns Ar core
537.Op ,prefetch= Ns Ar prefetch
538.Xc
539.Pq Event 2EH
540The number of L2 cache requests.
541.It Li L2_ST Xo
542.Op ,cachestate= Ns Ar mesi
543.Op ,core= Ns Ar core
544.Xc
545.Pq Event 2AH
546The number of L2 cache writes including speculative writes.
547.It Li LD_Blocks
548.Pq Event 03H , Umask 00H
549The number of load operations delayed due to store buffer blocks.
550.It Li LLC_Misses
551.Pq Event 2EH , Umask 41H
552The number of cache misses for references to the last level cache,
553excluding misses due to hardware prefetches.
554This is an architectural performance event.
555.It Li LLC_Reference
556The number of references to the last level cache,
557excluding those due to hardware prefetches.
558This is an architectural performance event.
559.Pq Event 2EH , Umask 4FH
560This is an architectural performance event.
561.It Li MMX_Assist
562.Pq Event CDH , Umask 00H
563The number of EMMX instructions executed.
564.It Li MMX_FP_Trans
565.Pq Event CCH , Umask 00H
566The number of transitions from MMX to X87.
567.It Li MMX_Instr_Exec
568.Pq Event B0H , Umask 00H
569The number of MMX instructions executed excluding
570.Li MOVQ
571and
572.Li MOVD
573stores.
574.It Li MMX_Instr_Ret
575.Pq Event CEH , Umask 00H
576The number of MMX instructions retired.
577.It Li Misalign_Mem_Ref
578.Pq Event 05H , Umask 00H
579The number of misaligned data memory references, counting loads and
580stores.
581.It Li Mul
582.Pq Event 12H , Umask 00H
583The number of multiply operations include speculative floating point
584and integer multiplies.
585This event is available on PMC1 only.
586.It Li NonHlt_Ref_Cycles
587.Pq Event 3CH , Umask 01H
588.Pq Alias Qq "Unhalted Reference Cycles"
589The number of non-halted bus cycles.
590This is an architectural performance event.
591.It Li Pref_Rqsts_Dn
592.Pq Event F8H , Umask 00H
593The number of hardware prefetch requests issued in backward streams.
594.It Li Pref_Rqsts_Up
595.Pq Event F0H , Umask 00H
596The number of hardware prefetch requests issued in forward streams.
597.It Li Resource_Stall
598.Pq Event A2H , Umask 00H
599The number of cycles where there is a resource related stall.
600.It Li SD_Drains
601.Pq Event 04H , Umask 00H
602The number of cycles while draining store buffers.
603.It Li SIMD_FP_DP_P_Ret
604.Pq Event D8H , Umask 02H
605The number of SSE/SSE2 packed double precision instructions retired.
606.It Li SIMD_FP_DP_P_Comp_Ret
607.Pq Event D9H , Umask 02H
608The number of SSE/SSE2 packed double precision compute instructions
609retired.
610.It Li SIMD_FP_DP_S_Ret
611.Pq Event D8H , Umask 03H
612The number of SSE/SSE2 scalar double precision instructions retired.
613.It Li SIMD_FP_DP_S_Comp_Ret
614.Pq Event D9H , Umask 03H
615The number of SSE/SSE2 scalar double precision compute instructions
616retired.
617.It Li SIMD_FP_SP_P_Comp_Ret
618.Pq Event D9H , Umask 00H
619The number of SSE/SSE2 packed single precision compute instructions
620retired.
621.It Li SIMD_FP_SP_Ret
622.Pq Event D8H , Umask 00H
623The number of SSE/SSE2 scalar single precision instructions retired,
624both packed and scalar.
625.It Li SIMD_FP_SP_S_Ret
626.Pq Event D8H , Umask 01H
627The number of SSE/SSE2 scalar single precision instructions retired.
628.It Li SIMD_FP_SP_S_Comp_Ret
629.Pq Event D9H , Umask 01H
630The number of SSE/SSE2 single precision compute instructions retired.
631.It Li SIMD_Int_128_Ret
632.Pq Event D8H , Umask 04H
633The number of SSE2 128-bit integer instructions retired.
634.It Li SIMD_Int_Pari_Exec
635.Pq Event B3H , Umask 20H
636The number of SIMD integer packed arithmetic instructions executed.
637.It Li SIMD_Int_Pck_Exec
638.Pq Event B3H , Umask 04H
639The number of SIMD integer pack operations instructions executed.
640.It Li SIMD_Int_Plog_Exec
641.Pq Event B3H , Umask 10H
642The number of SIMD integer packed logical instructions executed.
643.It Li SIMD_Int_Pmul_Exec
644.Pq Event B3H , Umask 01H
645The number of SIMD integer packed multiply instructions executed.
646.It Li SIMD_Int_Psft_Exec
647.Pq Event B3H , Umask 02H
648The number of SIMD integer packed shift instructions executed.
649.It Li SIMD_Int_Sat_Exec
650.Pq Event B1H , Umask 00H
651The number of SIMD integer saturating instructions executed.
652.It Li SIMD_Int_Upck_Exec
653.Pq Event B3H , Umask 08H
654The number of SIMD integer unpack instructions executed.
655.It Li SMC_Detected
656.Pq Event C3H , Umask 00H
657The number of times self-modifying code was detected.
658.It Li SSE_NTStores_Miss
659.Pq Event 4BH , Umask 03H
660The number of times an SSE streaming store instruction missed all caches.
661.It Li SSE_NTStores_Ret
662.Pq Event 07H , Umask 03H
663The number of SSE streaming store instructions executed.
664.It Li SSE_PrefNta_Miss
665.Pq Event 4BH , Umask 00H
666The number of times
667.Li PREFETCHNTA
668missed all caches.
669.It Li SSE_PrefNta_Ret
670.Pq Event 07H , Umask 00H
671The number of
672.Li PREFETCHNTA
673instructions retired.
674.It Li SSE_PrefT1_Miss
675.Pq Event 4BH , Umask 01H
676The number of times
677.Li PREFETCHT1
678missed all caches.
679.It Li SSE_PrefT1_Ret
680.Pq Event 07H , Umask 01H
681The number of
682.Li PREFETCHT1
683instructions retired.
684.It Li SSE_PrefT2_Miss
685.Pq Event 4BH , Umask 02H
686The number of times
687.Li PREFETCHNT2
688missed all caches.
689.It Li SSE_PrefT2_Ret
690.Pq Event 07H , Umask 02H
691The number of
692.Li PREFETCHT2
693instructions retired.
694.It Li Seg_Reg_Loads
695.Pq Event 06H , Umask 00H
696The number of segment register loads.
697.It Li Serial_Execution_Cycles
698.Pq Event 3CH , Umask 02H
699The number of non-halted bus cycles of this code while the other core
700was halted.
701.It Li Thermal_Trip
702.Pq Event 3BH , Umask C0H
703The duration in a thermal trip based on the current core clock.
704.It Li Unfusion
705.Pq Event DBH , Umask 00H
706The number of unfusion events.
707.It Li Unhalted_Core_Cycles
708.Pq Event 3CH , Umask 00H
709The number of core clock cycles when the clock signal on a specific
710core is not halted.
711This is an architectural performance event.
712.It Li Uops_Ret
713.Pq Event C2H , Umask 00H
714The number of micro-ops retired.
715.El
716.Ss Event Name Aliases
717The following table shows the mapping between the PMC-independent
718aliases supported by
719.Lb libpmc
720and the underlying hardware events used.
721.Bl -column "branch-mispredicts" "Description"
722.It Em Alias Ta Em Event
723.It Li branches Ta Li Br_Instr_Ret
724.It Li branch-mispredicts Ta Li Br_MisPred_Ret
725.It Li dc-misses Ta (unsupported)
726.It Li ic-misses Ta Li ICache_Misses
727.It Li instructions Ta Li Instr_Ret
728.It Li interrupts Ta Li HW_Int_Rx
729.It Li unhalted-cycles Ta (unsupported)
730.El
731.Sh PROCESSOR ERRATA
732The following errata affect performance measurement on these
733processors.
734These errata are documented in
735.Rs
736.%B Specification Update
737.%T Intel\(rg CoreTM Duo Processor and Intel\(rg CoreTM Solo Processor on 65 nm Process
738.%N Order Number 309222-017
739.%D July 2008
740.%Q Intel Corporation
741.Re
742.Bl -tag -width indent -compact
743.It AE19
744Data prefetch performance monitoring events can only be enabled
745on a single core.
746.It AE25
747Performance monitoring counters that count external bus events
748may report incorrect values after processor power state transitions.
749.It AE28
750Performance monitoring events for retired floating point operations
751(C1H) may not be accurate.
752.It AE29
753DR3 address match on MOVD/MOVQ/MOVNTQ memory store
754instruction may incorrectly increment performance monitoring count
755for saturating SIMD instructions retired (Event CFH).
756.It AE33
757Hardware prefetch performance monitoring events may be counted
758inaccurately.
759.It AE36
760The
761.Li CPU_CLK_UNHALTED
762performance monitoring event (Event 3CH) counts
763clocks when the processor is in the C1/C2 processor power states.
764.It AE39
765Certain performance monitoring counters related to bus, L2 cache
766and power management are inaccurate.
767.It AE51
768Performance monitoring events for retired instructions (Event C0H) may
769not be accurate.
770.It AE67
771Performance monitoring event
772.Li FP_ASSIST
773may not be accurate.
774.It AE78
775Performance monitoring event for hardware prefetch requests (Event
7764EH) and hardware prefetch request cache misses (Event 4FH) may not be
777accurate.
778.It AE82
779Performance monitoring event
780.Li FP_MMX_TRANS_TO_MMX
781may not count some transitions.
782.El
783.Sh SEE ALSO
784.Xr pmc 3 ,
785.Xr pmc.atom 3 ,
786.Xr pmc.core2 3 ,
787.Xr pmc.iaf 3 ,
788.Xr pmc.k8 3 ,
789.Xr pmc.soft 3 ,
790.Xr pmc.tsc 3 ,
791.Xr pmclog 3 ,
792.Xr hwpmc 4
793.Sh HISTORY
794The
795.Nm pmc
796library first appeared in
797.Fx 6.0 .
798.Sh AUTHORS
799The
800.Lb libpmc
801library was written by
802.An Joseph Koshy Aq Mt jkoshy@FreeBSD.org .
803