xref: /freebsd/lib/libpmc/pmc.ivybridge.3 (revision ab2043b81eaba0d7d7769b4a58b2b6d17bc464a3)
1.\" Copyright (c) 2012 Fabien Thomas.  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.\" $FreeBSD$
25.\"
26.Dd August 24, 2012
27.Dt PMC.IVYBRIDGE 3
28.Os
29.Sh NAME
30.Nm pmc.ivybridge
31.Nd measurement events for
32.Tn Intel
33.Tn Ivy Bridge
34family CPUs
35.Sh LIBRARY
36.Lb libpmc
37.Sh SYNOPSIS
38.In pmc.h
39.Sh DESCRIPTION
40.Tn Intel
41.Tn "Ivy Bridge"
42CPUs contain PMCs conforming to version 2 of the
43.Tn Intel
44performance measurement architecture.
45These CPUs may contain up to three classes of PMCs:
46.Bl -tag -width "Li PMC_CLASS_IAP"
47.It Li PMC_CLASS_IAF
48Fixed-function counters that count only one hardware event per counter.
49.It Li PMC_CLASS_IAP
50Programmable counters that may be configured to count one of a defined
51set of hardware events.
52.El
53.Pp
54The number of PMCs available in each class and their widths need to be
55determined at run time by calling
56.Xr pmc_cpuinfo 3 .
57.Pp
58Intel Ivy Bridge PMCs are documented in
59.Rs
60.%B "Intel(R) 64 and IA-32 Architectures Software Developer's Manual"
61.%T "Volume 3B: System Programming Guide, Part 2"
62.%N "Order Number: 253669-043US"
63.%D May 2012
64.%Q "Intel Corporation"
65.Re
66.Ss IVYBRIDGE FIXED FUNCTION PMCS
67These PMCs and their supported events are documented in
68.Xr pmc.iaf 3 .
69.Ss IVYBRIDGE PROGRAMMABLE PMCS
70The programmable PMCs support the following capabilities:
71.Bl -column "PMC_CAP_INTERRUPT" "Support"
72.It Em Capability Ta Em Support
73.It PMC_CAP_CASCADE Ta \&No
74.It PMC_CAP_EDGE Ta Yes
75.It PMC_CAP_INTERRUPT Ta Yes
76.It PMC_CAP_INVERT Ta Yes
77.It PMC_CAP_READ Ta Yes
78.It PMC_CAP_PRECISE Ta \&No
79.It PMC_CAP_SYSTEM Ta Yes
80.It PMC_CAP_TAGGING Ta \&No
81.It PMC_CAP_THRESHOLD Ta Yes
82.It PMC_CAP_USER Ta Yes
83.It PMC_CAP_WRITE Ta Yes
84.El
85.Ss Event Qualifiers
86Event specifiers for these PMCs support the following common
87qualifiers:
88.Bl -tag -width indent
89.It Li rsp= Ns Ar value
90Configure the Off-core Response bits.
91.Bl -tag -width indent
92.It Li REQ_DMND_DATA_RD
93Counts the number of demand and DCU prefetch data reads of full and partial
94cachelines as well as demand data page table entry cacheline reads. Does not
95count L2 data read prefetches or instruction fetches.
96.It Li REQ_DMND_RFO
97Counts the number of demand and DCU prefetch reads for ownership (RFO)
98requests generated by a write to data cacheline. Does not count L2 RFO
99prefetches.
100.It Li REQ_DMND_IFETCH
101Counts the number of demand and DCU prefetch instruction cacheline reads.
102Does not count L2 code read prefetches.
103.It Li REQ_WB
104Counts the number of writeback (modified to exclusive) transactions.
105.It Li REQ_PF_DATA_RD
106Counts the number of data cacheline reads generated by L2 prefetchers.
107.It Li REQ_PF_RFO
108Counts the number of RFO requests generated by L2 prefetchers.
109.It Li REQ_PF_IFETCH
110Counts the number of code reads generated by L2 prefetchers.
111.It Li REQ_PF_LLC_DATA_RD
112L2 prefetcher to L3 for loads.
113.It Li REQ_PF_LLC_RFO
114RFO requests generated by L2 prefetcher
115.It Li REQ_PF_LLC_IFETCH
116L2 prefetcher to L3 for instruction fetches.
117.It Li REQ_BUS_LOCKS
118Bus lock and split lock requests.
119.It Li REQ_STRM_ST
120Streaming store requests.
121.It Li REQ_OTHER
122Any other request that crosses IDI, including I/O.
123.It Li RES_ANY
124Catch all value for any response types.
125.It Li RES_SUPPLIER_NO_SUPP
126No Supplier Information available.
127.It Li RES_SUPPLIER_LLC_HITM
128M-state initial lookup stat in L3.
129.It Li RES_SUPPLIER_LLC_HITE
130E-state.
131.It Li RES_SUPPLIER_LLC_HITS
132S-state.
133.It Li RES_SUPPLIER_LLC_HITF
134F-state.
135.It Li RES_SUPPLIER_LOCAL
136Local DRAM Controller.
137.It Li RES_SNOOP_SNPI_NONE
138No details on snoop-related information.
139.It Li RES_SNOOP_SNP_NO_NEEDED
140No snoop was needed to satisfy the request.
141.It Li RES_SNOOP_SNP_MISS
142A snoop was needed and it missed all snooped caches:
143-For LLC Hit, ReslHitl was returned by all cores
144-For LLC Miss, Rspl was returned by all sockets and data was returned from
145DRAM.
146.It Li RES_SNOOP_HIT_NO_FWD
147A snoop was needed and it hits in at least one snooped cache. Hit denotes a
148cache-line was valid before snoop effect. This includes:
149-Snoop Hit w/ Invalidation (LLC Hit, RFO)
150-Snoop Hit, Left Shared (LLC Hit/Miss, IFetch/Data_RD)
151-Snoop Hit w/ Invalidation and No Forward (LLC Miss, RFO Hit S)
152In the LLC Miss case, data is returned from DRAM.
153.It Li RES_SNOOP_HIT_FWD
154A snoop was needed and data was forwarded from a remote socket.
155This includes:
156-Snoop Forward Clean, Left Shared (LLC Hit/Miss, IFetch/Data_RD/RFT).
157.It Li RES_SNOOP_HITM
158A snoop was needed and it HitM-ed in local or remote cache. HitM denotes a
159cache-line was in modified state before effect as a results of snoop. This
160includes:
161-Snoop HitM w/ WB (LLC miss, IFetch/Data_RD)
162-Snoop Forward Modified w/ Invalidation (LLC Hit/Miss, RFO)
163-Snoop MtoS (LLC Hit, IFetch/Data_RD).
164.It Li RES_NON_DRAM
165Target was non-DRAM system address. This includes MMIO transactions.
166.El
167.It Li cmask= Ns Ar value
168Configure the PMC to increment only if the number of configured
169events measured in a cycle is greater than or equal to
170.Ar value .
171.It Li edge
172Configure the PMC to count the number of de-asserted to asserted
173transitions of the conditions expressed by the other qualifiers.
174If specified, the counter will increment only once whenever a
175condition becomes true, irrespective of the number of clocks during
176which the condition remains true.
177.It Li inv
178Invert the sense of comparison when the
179.Dq Li cmask
180qualifier is present, making the counter increment when the number of
181events per cycle is less than the value specified by the
182.Dq Li cmask
183qualifier.
184.It Li os
185Configure the PMC to count events happening at processor privilege
186level 0.
187.It Li usr
188Configure the PMC to count events occurring at privilege levels 1, 2
189or 3.
190.El
191.Pp
192If neither of the
193.Dq Li os
194or
195.Dq Li usr
196qualifiers are specified, the default is to enable both.
197.Ss Event Specifiers (Programmable PMCs)
198Ivy Bridge programmable PMCs support the following events:
199.Bl -tag -width indent
200.It Li LD_BLOCKS.STORE_FORWARD
201.Pq Event 03H , Umask 02H
202loads blocked by overlapping with store buffer that cannot be forwarded .
203.It Li MISALIGN_MEM_REF.LOADS
204.Pq Event 05H , Umask 01H
205Speculative cache-line split load uops dispatched to L1D.
206.It Li MISALIGN_MEM_REF.STORES
207.Pq Event 05H , Umask 02H
208Speculative cache-line split Store- address uops dispatched to L1D.
209.It Li LD_BLOCKS_PARTIAL.ADDRESS_ALIAS
210.Pq Event 07H , Umask 01H
211False dependencies in MOB due to partial compare on address.
212.It Li DTLB_LOAD_MISSES.DEMAND_LD_MISS_CAUSES_A_WALK
213.Pq Event 08H , Umask 81H
214Misses in all TLB levels that cause a page walk of any page size from demand loads.
215.It Li DTLB_LOAD_MISSES.DEMAND_LD_WALK_COMPLETED
216.Pq Event 08H , Umask 82H
217Misses in all TLB levels that caused page walk completed of any size by demand loads.
218.It Li DTLB_LOAD_MISSES.DEMAND_LD_WALK_DURATION
219.Pq Event 08H , Umask 84H
220Cycle PMH is busy with a walk due to demand loads.
221.It Li UOPS_ISSUED.ANY
222.Pq Event 0EH , Umask 01H
223Increments each cycle the # of Uops issued by the RAT to RS.
224Set Cmask = 1, Inv = 1to count stalled cycles.
225Set Cmask = 1, Inv = 1, Any= 1to count stalled cycles of this core.
226.It Li UOPS_ISSUED.FLAGS_MERGE
227.Pq Event 0EH , Umask 10H
228Number of flags-merge uops allocated. Such uops adds delay.
229.It Li UOPS_ISSUED.SLOW_LEA
230.Pq Event 0EH , Umask 20H
231Number of slow LEA or similar uops allocated. Such uop has 3 sources (e.g. 2
232sources + immediate) regardless if as a result of LEA instruction or not.
233.It Li UOPS_ISSUED.SINGLE_MUL
234.Pq Event 0EH , Umask 40H
235Number of multiply packed/scalar single precision uops allocated.
236.It Li ARITH.FPU_DIV_ACTIVE
237.Pq Event 14H , Umask 01H
238Cycles that the divider is active, includes INT and FP. Set 'edge =1,
239cmask=1' to count the number of divides.
240.It Li L2_RQSTS.DEMAND_DATA_RD_HIT
241.Pq Event 24H , Umask 01H
242Demand Data Read requests that hit L2 cache.
243.It Li L2_RQSTS.ALL_DEMAND_DATA_RD
244.Pq Event 24H , Umask 03H
245Counts any demand and L1 HW prefetch data load requests to L2.
246.It Li L2_RQSTS.RFO_HITS
247.Pq Event 24H , Umask 04H
248Counts the number of store RFO requests that hit the L2 cache.
249.It Li L2_RQSTS.RFO_MISS
250.Pq Event 24H , Umask 08H
251Counts the number of store RFO requests that miss the L2 cache.
252.It Li L2_RQSTS.ALL_RFO
253.Pq Event 24H , Umask 0CH
254Counts all L2 store RFO requests.
255.It Li L2_RQSTS.CODE_RD_HIT
256.Pq Event 24H , Umask 10H
257Number of instruction fetches that hit the L2 cache.
258.It Li L2_RQSTS.CODE_RD_MISS
259.Pq Event 24H , Umask 20H
260Number of instruction fetches that missed the L2 cache.
261.It Li L2_RQSTS.ALL_CODE_RD
262.Pq Event 24H , Umask 30H
263Counts all L2 code requests.
264.It Li L2_RQSTS.PF_HIT
265.Pq Event 24H , Umask 40H
266Counts all L2 HW prefetcher requests that hit L2.
267.It Li L2_RQSTS.PF_MISS
268.Pq Event 24H , Umask 80H
269Counts all L2 HW prefetcher requests that missed L2.
270.It Li L2_RQSTS.ALL_PF
271.Pq Event 24H , Umask C0H
272Counts all L2 HW prefetcher requests.
273.It Li L2_STORE_LOCK_RQSTS.MISS
274.Pq Event 27H , Umask 01H
275RFOs that miss cache lines.
276.It Li L2_STORE_LOCK_RQSTS.HIT_M
277.Pq Event 27H , Umask 08H
278RFOs that hit cache lines in M state.
279.It Li L2_STORE_LOCK_RQSTS.ALL
280.Pq Event 27H , Umask 0FH
281RFOs that access cache lines in any state.
282.It Li L2_L1D_WB_RQSTS.MISS
283.Pq Event 28H , Umask 01H
284Not rejected writebacks that missed LLC.
285.It Li L2_L1D_WB_RQSTS.HIT_E
286.Pq Event 28H , Umask 04H
287Not rejected writebacks from L1D to L2 cache lines in E state.
288.It Li L2_L1D_WB_RQSTS.HIT_M
289.Pq Event 28H , Umask 08H
290Not rejected writebacks from L1D to L2 cache lines in M state.
291.It Li L2_L1D_WB_RQSTS.ALL
292.Pq Event 28H , Umask 0FH
293Not rejected writebacks from L1D to L2 cache lines in any state.
294.It Li LONGEST_LAT_CACHE.REFERENCE
295.Pq Event 2EH , Umask 4FH
296This event counts requests originating from the core that reference a cache
297line in the last level cache.
298.It Li LONGEST_LAT_CACHE.MISS
299.Pq Event 2EH , Umask 41H
300This event counts each cache miss condition for references to the last level
301cache.
302.It Li CPU_CLK_UNHALTED.THREAD_P
303.Pq Event 3CH , Umask 00H
304Counts the number of thread cycles while the thread is not in a halt state.
305The thread enters the halt state when it is running the HLT instruction. The
306core frequency may change from time to time due to power or thermal
307throttling.
308.It Li CPU_CLK_THREAD_UNHALTED.REF_XCLK
309.Pq Event 3CH , Umask 01H
310Increments at the frequency of XCLK (100 MHz) when not halted.
311.It Li L1D_PEND_MISS.PENDING
312.Pq Event 48H , Umask 01H
313Increments the number of outstanding L1D misses every cycle. Set Cmaks = 1
314and Edge =1 to count occurrences.
315Counter 2 only.
316Set Cmask = 1 to count cycles.
317.It Li DTLB_STORE_MISSES.MISS_CAUSES_A_WALK
318.Pq Event 49H , Umask 01H
319Miss in all TLB levels causes an page walk of any page size (4K/2M/4M/1G).
320.It Li DTLB_STORE_MISSES.WALK_COMPLETED
321.Pq Event 49H , Umask 02H
322Miss in all TLB levels causes a page walk that completes of any page size
323(4K/2M/4M/1G).
324.It Li DTLB_STORE_MISSES.WALK_DURATION
325.Pq Event 49H , Umask 04H
326Cycles PMH is busy with this walk.
327.It Li DTLB_STORE_MISSES.STLB_HIT
328.Pq Event 49H , Umask 10H
329Store operations that miss the first TLB level but hit the second and do not
330cause page walks.
331.It Li LOAD_HIT_PRE.SW_PF
332.Pq Event 4CH , Umask 01H
333Non-SW-prefetch load dispatches that hit fill buffer allocated for S/W prefetch.
334.It Li LOAD_HIT_PRE.HW_PF
335.Pq Event 4CH , Umask 02H
336Non-SW-prefetch load dispatches that hit fill buffer allocated for H/W prefetch.
337.It Li L1D.REPLACEMENT
338.Pq Event 51H , Umask 01H
339Counts the number of lines brought into the L1 data cache.
340.It Li MOVE_ELIMINATION.INT_NOT_ELIMINATED
341.Pq Event 58H , Umask 01H
342Number of integer Move Elimination candidate uops that were not eliminated.
343.It Li MOVE_ELIMINATION.SIMD_NOT_ELIMINATED
344.Pq Event 58H , Umask 02H
345Number of SIMD Move Elimination candidate uops that were not eliminated.
346.It Li MOVE_ELIMINATION.INT_ELIMINATED
347.Pq Event 58H , Umask 04H
348Number of integer Move Elimination candidate uops that were eliminated.
349.It Li MOVE_ELIMINATION.SIMD_ELIMINATED
350.Pq Event 58H , Umask 08H
351Number of SIMD Move Elimination candidate uops that were eliminated.
352.It Li CPL_CYCLES.RING0
353.Pq Event 5CH , Umask 01H
354Unhalted core cycles when the thread is in ring 0.
355Use Edge to count transition.
356.It Li CPL_CYCLES.RING123
357.Pq Event 5CH , Umask 02H
358Unhalted core cycles when the thread is not in ring 0.
359.It Li RS_EVENTS.EMPTY_CYCLES
360.Pq Event 5EH , Umask 01H
361Cycles the RS is empty for the thread.
362.It Li TLB_ACCESS.LOAD_STLB_HIT
363.Pq Event 5FH , Umask 01H
364Counts load operations that missed 1st level DTLB but hit the 2nd level.
365.It Li OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD
366.Pq Event 60H , Umask 01H
367Offcore outstanding Demand Data Read transactions in SQ to uncore. Set
368Cmask=1 to count cycles.
369.It Li OFFCORE_REQUESTS_OUTSTANDING.DEMAND_CODE_RD
370.Pq Event 60H , Umask 02H
371Offcore outstanding Demand Code Read transactions in SQ to uncore. Set
372Cmask=1 to count cycles.
373.It Li OFFCORE_REQUESTS_OUTSTANDING.DEMAND_RFO
374.Pq Event 60H , Umask 04H
375Offcore outstanding RFO store transactions in SQ to uncore. Set Cmask=1 to
376count cycles.
377.It Li OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD
378.Pq Event 60H , Umask 08H
379Offcore outstanding cacheable data read transactions in SQ to uncore. Set
380Cmask=1 to count cycles.
381.It Li LOCK_CYCLES.SPLIT_LOCK_UC_LOCK_DURATION
382.Pq Event 63H , Umask 01H
383Cycles in which the L1D and L2 are locked, due to a UC lock or split lock.
384.It Li LOCK_CYCLES.CACHE_LOCK_DURATION
385.Pq Event 63H , Umask 02H
386Cycles in which the L1D is locked.
387.It Li IDQ.EMPTY
388.Pq Event 79H , Umask 02H
389Counts cycles the IDQ is empty.
390.It Li IDQ.MITE_UOPS
391.Pq Event 79H , Umask 04H
392Increment each cycle # of uops delivered to IDQ from MITE path.
393Can combine Umask 04H and 20H.
394Set Cmask = 1 to count cycles.
395.It Li IDQ.DSB_UOPS
396.Pq Event 79H , Umask 08H
397Increment each cycle. # of uops delivered to IDQ from DSB path.
398Can combine Umask 08H and 10H
399Set Cmask = 1 to count cycles.
400.It Li IDQ.MS_DSB_UOPS
401.Pq Event 79H , Umask 10H
402Increment each cycle # of uops delivered to IDQ when MS_busy by DSB. Set
403Cmask = 1 to count cycles. Add Edge=1 to count # of delivery.
404Can combine Umask 04H, 08H.
405.It Li IDQ.MS_MITE_UOPS
406.Pq Event 79H , Umask 20H
407Increment each cycle # of uops delivered to IDQ when MS_busy by MITE. Set
408Cmask = 1 to count cycles.
409Can combine Umask 04H, 08H.
410.It Li IDQ.MS_UOPS
411.Pq Event 79H , Umask 30H
412Increment each cycle # of uops delivered to IDQ from MS by either DSB or
413MITE. Set Cmask = 1 to count cycles.
414Can combine Umask 04H, 08H.
415.It Li IDQ.ALL_DSB_CYCLES_ANY_UOPS
416.Pq Event 79H , Umask 18H
417Counts cycles DSB is delivered at least one uops. Set Cmask = 1.
418.It Li IDQ.ALL_DSB_CYCLES_4_UOPS
419.Pq Event 79H , Umask 18H
420Counts cycles DSB is delivered four uops. Set Cmask = 4.
421.It Li IDQ.ALL_MITE_CYCLES_ANY_UOPS
422.Pq Event 79H , Umask 24H
423Counts cycles MITE is delivered at least one uops. Set Cmask = 1.
424.It Li IDQ.ALL_MITE_CYCLES_4_UOPS
425.Pq Event 79H , Umask 24H
426Counts cycles MITE is delivered four uops. Set Cmask = 4.
427.It Li IDQ.MITE_ALL_UOPS
428.Pq Event 79H , Umask 3CH
429# of uops delivered to IDQ from any path.
430.It Li ICACHE.MISSES
431.Pq Event 80H , Umask 02H
432Number of Instruction Cache, Streaming Buffer and Victim Cache Misses.
433Includes UC accesses.
434.It Li ITLB_MISSES.MISS_CAUSES_A_WALK
435.Pq Event 85H , Umask 01H
436Misses in all ITLB levels that cause page walks.
437.It Li ITLB_MISSES.WALK_COMPLETED
438.Pq Event 85H , Umask 02H
439Misses in all ITLB levels that cause completed page walks.
440.It Li ITLB_MISSES.WALK_DURATION
441.Pq Event 85H , Umask 04H
442Cycle PMH is busy with a walk.
443.It Li ITLB_MISSES.STLB_HIT
444.Pq Event 85H , Umask 10H
445Number of cache load STLB hits. No page walk.
446.It Li ILD_STALL.LCP
447.Pq Event 87H , Umask 01H
448Stalls caused by changing prefix length of the instruction.
449.It Li ILD_STALL.IQ_FULL
450.Pq Event 87H , Umask 04H
451Stall cycles due to IQ is full.
452.It Li BR_INST_EXEC.COND
453.Pq Event 88H , Umask 01H
454Qualify conditional near branch instructions executed, but not necessarily
455retired.
456Must combine with umask 40H, 80H.
457.It Li BR_INST_EXEC.DIRECT_JMP
458.Pq Event 88H , Umask 02H
459Qualify all unconditional near branch instructions excluding calls and
460indirect branches.
461Must combine with umask 80H.
462.It Li BR_INST_EXEC.INDIRECT_JMP_NON_CALL_RET
463.Pq Event 88H , Umask 04H
464Qualify executed indirect near branch instructions that are not calls nor
465returns.
466Must combine with umask 80H.
467.It Li BR_INST_EXEC.RETURN_NEAR
468.Pq Event 88H , Umask 08H
469Qualify indirect near branches that have a return mnemonic.
470Must combine with umask 80H.
471.It Li BR_INST_EXEC.DIRECT_NEAR_CALL
472.Pq Event 88H , Umask 10H
473Qualify unconditional near call branch instructions, excluding non call
474branch, executed.
475Must combine with umask 80H.
476.It Li BR_INST_EXEC.INDIRECT_NEAR_CALL
477.Pq Event 88H , Umask 20H
478Qualify indirect near calls, including both register and memory indirect,
479executed.
480Must combine with umask 80H.
481.It Li BR_INST_EXEC.NONTAKEN
482.Pq Event 88H , Umask 40H
483Qualify non-taken near branches executed.
484Applicable to umask 01H only.
485.It Li BR_INST_EXEC.TAKEN
486.Pq Event 88H , Umask 80H
487Qualify taken near branches executed. Must combine with 01H,02H, 04H, 08H,
48810H, 20H.
489.It Li BR_INST_EXEC.ALL_BRANCHES
490.Pq Event 88H , Umask FFH
491Counts all near executed branches (not necessarily retired).
492.It Li BR_MISP_EXEC.COND
493.Pq Event 89H , Umask 01H
494Qualify conditional near branch instructions mispredicted.
495Must combine with umask 40H, 80H.
496.It Li BR_MISP_EXEC.INDIRECT_JMP_NON_CALL_RET
497.Pq Event 89H , Umask 04H
498Qualify mispredicted indirect near branch instructions that are not calls
499nor returns.
500Must combine with umask 80H.
501.It Li BR_MISP_EXEC.RETURN_NEAR
502.Pq Event 89H , Umask 08H
503Qualify mispredicted indirect near branches that have a return mnemonic.
504Must combine with umask 80H.
505.It Li BR_MISP_EXEC.DIRECT_NEAR_CALL
506.Pq Event 89H , Umask 10H
507Qualify mispredicted unconditional near call branch instructions, excluding
508non call branch, executed.
509Must combine with umask 80H.
510.It Li BR_MISP_EXEC.INDIRECT_NEAR_CALL
511.Pq Event 89H , Umask 20H
512Qualify mispredicted indirect near calls, including both register and memory
513indirect, executed.
514Must combine with umask 80H.
515.It Li BR_MISP_EXEC.NONTAKEN
516.Pq Event 89H , Umask 40H
517Qualify mispredicted non-taken near branches executed.
518Applicable to umask 01H only.
519.It Li BR_MISP_EXEC.TAKEN
520.Pq Event 89H , Umask 80H
521Qualify mispredicted taken near branches executed. Must combine with
52201H,02H, 04H, 08H, 10H, 20H.
523.It Li BR_MISP_EXEC.ALL_BRANCHES
524.Pq Event 89H , Umask FFH
525Counts all near executed branches (not necessarily retired).
526.It Li IDQ_UOPS_NOT_DELIVERED.CORE
527.Pq Event 9CH , Umask 01H
528Count number of non-delivered uops to RAT per thread.
529Use Cmask to qualify uop b/w.
530.It Li UOPS_DISPATCHED_PORT.PORT_0
531.Pq Event A1H , Umask 01H
532Cycles which a Uop is dispatched on port 0.
533.It Li UOPS_DISPATCHED_PORT.PORT_1
534.Pq Event A1H , Umask 02H
535Cycles which a Uop is dispatched on port 1.
536.It Li UOPS_DISPATCHED_PORT.PORT_2_LD
537.Pq Event A1H , Umask 04H
538Cycles which a load uop is dispatched on port 2.
539.It Li UOPS_DISPATCHED_PORT.PORT_2_STA
540.Pq Event A1H , Umask 08H
541Cycles which a store address uop is dispatched on port 2.
542.It Li UOPS_DISPATCHED_PORT.PORT_2
543.Pq Event A1H , Umask 0CH
544Cycles which a Uop is dispatched on port 2.
545.It Li UOPS_DISPATCHED_PORT.PORT_3_LD
546.Pq Event A1H , Umask 10H
547Cycles which a load uop is dispatched on port 3.
548.It Li UOPS_DISPATCHED_PORT.PORT_3_STA
549.Pq Event A1H , Umask 20H
550Cycles which a store address uop is dispatched on port 3.
551.It Li UOPS_DISPATCHED_PORT.PORT_3
552.Pq Event A1H , Umask 30H
553Cycles which a Uop is dispatched on port 3.
554.It Li UOPS_DISPATCHED_PORT.PORT_4
555.Pq Event A1H , Umask 40H
556Cycles which a Uop is dispatched on port 4.
557.It Li UOPS_DISPATCHED_PORT.PORT_5
558.Pq Event A1H , Umask 80H
559Cycles which a Uop is dispatched on port 5.
560.It Li RESOURCE_STALLS.ANY
561.Pq Event A2H , Umask 01H
562Cycles Allocation is stalled due to Resource Related reason.
563.It Li RESOURCE_STALLS.RS
564.Pq Event A2H , Umask 04H
565Cycles stalled due to no eligible RS entry available.
566.It Li RESOURCE_STALLS.SB
567.Pq Event A2H , Umask 08H
568Cycles stalled due to no store buffers available. (not including draining
569form sync).
570.It Li RESOURCE_STALLS.ROB
571.Pq Event A2H , Umask 10H
572Cycles stalled due to re-order buffer full.
573.It Li DSB2MITE_SWITCHES.COUNT
574.Pq Event ABH , Umask 01H
575Number of DSB to MITE switches.
576.It Li DSB2MITE_SWITCHES.PENALTY_CYCLES
577.Pq Event ABH , Umask 02H
578Cycles DSB to MITE switches caused delay.
579.It Li DSB_FILL.EXCEED_DSB_LINES
580.Pq Event ACH , Umask 08H
581DSB Fill encountered > 3 DSB lines.
582.It Li ITLB.ITLB_FLUSH
583.Pq Event AEH , Umask 01H
584Counts the number of ITLB flushes, includes 4k/2M/4M pages.
585.It Li OFFCORE_REQUESTS.DEMAND_DATA_RD
586.Pq Event B0H , Umask 01H
587Demand data read requests sent to uncore.
588.It Li OFFCORE_REQUESTS.DEMAND_CODE_RD
589.Pq Event B0H , Umask 02H
590Demand code read requests sent to uncore.
591.It Li OFFCORE_REQUESTS.DEMAND_RFO
592.Pq Event B0H , Umask 04H
593Demand RFO read requests sent to uncore, including regular RFOs, locks,
594ItoM.
595.It Li OFFCORE_REQUESTS.ALL_DATA_RD
596.Pq Event B0H , Umask 08H
597Data read requests sent to uncore (demand and prefetch).
598.It Li UOPS_EXECUTED.THREAD
599.Pq Event B1H , Umask 01H
600Counts total number of uops to be executed per-thread each cycle. Set Cmask
601= 1, INV =1 to count stall cycles.
602.It Li UOPS_EXECUTED.CORE
603.Pq Event B1H , Umask 02H
604Counts total number of uops to be executed per-core each cycle.
605Do not need to set ANY.
606.It Li OFF_CORE_RESPONSE_0
607.Pq Event B7H , Umask 01H
608Off-core Response Performance Monitoring.
609PMC0 only.
610Requires programming MSR 01A6H.
611.It Li OFF_CORE_RESPONSE_1
612.Pq Event BBH , Umask 01H
613Off-core Response Performance Monitoring.
614PMC3 only.
615Requires programming MSR 01A7H.
616.It Li TLB_FLUSH.DTLB_THREAD
617.Pq Event BDH , Umask 01H
618DTLB flush attempts of the thread- specific entries.
619.It Li TLB_FLUSH.STLB_ANY
620.Pq Event BDH , Umask 20H
621Count number of STLB flush attempts.
622.It Li INST_RETIRED.ANY_P
623.Pq Event C0H , Umask 00H
624Number of instructions at retirement.
625.It Li INST_RETIRED.ALL
626.Pq Event C0H , Umask 01H
627Precise instruction retired event with HW to reduce effect of PEBS shadow in
628IP distribution.
629PMC1 only.
630Must quiesce other PMCs.
631.It Li OTHER_ASSISTS.AVX_STORE
632.Pq Event C1H , Umask 08H
633Number of assists associated with 256-bit AVX store operations.
634.It Li OTHER_ASSISTS.AVX_TO_SSE
635.Pq Event C1H , Umask 10H
636Number of transitions from AVX- 256 to legacy SSE when penalty applicable.
637.It Li OTHER_ASSISTS.SSE_TO_AVX
638.Pq Event C1H , Umask 20H
639Number of transitions from SSE to AVX-256 when penalty applicable.
640.It Li UOPS_RETIRED.ALL
641.Pq Event C2H , Umask 01H
642Counts the number of micro-ops retired, Use cmask=1 and invert to count
643active cycles or stalled cycles.
644Supports PEBS, use Any=1 for core granular.
645.It Li UOPS_RETIRED.RETIRE_SLOTS
646.Pq Event C2H , Umask 02H
647Counts the number of retirement slots used each cycle.
648.It Li MACHINE_CLEARS.MEMORY_ORDERING
649.Pq Event C3H , Umask 02H
650Counts the number of machine clears due to memory order conflicts.
651.It Li MACHINE_CLEARS.SMC
652.Pq Event C3H , Umask 04H
653Number of self-modifying-code machine clears detected.
654.It Li MACHINE_CLEARS.MASKMOV
655.Pq Event C3H , Umask 20H
656Counts the number of executed AVX masked load operations that refer to an
657illegal address range with the mask bits set to 0.
658.It Li BR_INST_RETIRED.ALL_BRANCHES
659.Pq Event C4H , Umask 00H
660Branch instructions at retirement.
661.It Li BR_INST_RETIRED.CONDITIONAL
662.Pq Event C4H , Umask 01H
663Counts the number of conditional branch instructions retired.
664Supports PEBS.
665.It Li BR_INST_RETIRED.NEAR_CALL
666.Pq Event C4H , Umask 02H
667Direct and indirect near call instructions retired.
668.It Li BR_INST_RETIRED.ALL_BRANCHES
669.Pq Event C4H , Umask 04H
670Counts the number of branch instructions retired.
671.It Li BR_INST_RETIRED.NEAR_RETURN
672.Pq Event C4H , Umask 08H
673Counts the number of near return instructions retired.
674.It Li BR_INST_RETIRED.NOT_TAKEN
675.Pq Event C4H , Umask 10H
676Counts the number of not taken branch instructions retired.
677.It Li BR_INST_RETIRED.NEAR_TAKEN
678.Pq Event C4H , Umask 20H
679Number of near taken branches retired.
680.It Li BR_INST_RETIRED.FAR_BRANCH
681.Pq Event C4H , Umask 40H
682Number of far branches retired.
683.It Li BR_MISP_RETIRED.ALL_BRANCHES
684.Pq Event C5H , Umask 00H
685Mispredicted branch instructions at retirement.
686.It Li BR_MISP_RETIRED.CONDITIONAL
687.Pq Event C5H , Umask 01H
688Mispredicted conditional branch instructions retired.
689Supports PEBS.
690.It Li BR_MISP_RETIRED.NEAR_CALL
691.Pq Event C5H , Umask 02H
692Direct and indirect mispredicted near call instructions retired.
693.It Li BR_MISP_RETIRED.ALL_BRANCHES
694.Pq Event C5H , Umask 04H
695Mispredicted macro branch instructions retired.
696.It Li BR_MISP_RETIRED.NOT_TAKEN
697.Pq Event C5H , Umask 10H
698Mispredicted not taken branch instructions retired.
699.It Li BR_MISP_RETIRED.TAKEN
700.Pq Event C5H , Umask 20H
701Mispredicted taken branch instructions retired.
702.It Li FP_ASSIST.X87_OUTPUT
703.Pq Event CAH , Umask 02H
704Number of X87 FP assists due to Output values.
705.It Li FP_ASSIST.X87_INPUT
706.Pq Event CAH , Umask 04H
707Number of X87 FP assists due to input values.
708.It Li FP_ASSIST.SIMD_OUTPUT
709.Pq Event CAH , Umask 08H
710Number of SIMD FP assists due to Output values.
711.It Li FP_ASSIST.SIMD_INPUT
712.Pq Event CAH , Umask 10H
713Number of SIMD FP assists due to input values.
714.It Li FP_ASSIST.ANY
715.Pq Event CAH , Umask 1EH
716Cycles with any input/output SSE* or FP assists.
717.It Li ROB_MISC_EVENTS.LBR_INSERTS
718.Pq Event CCH , Umask 20H
719Count cases of saving new LBR records by hardware.
720.It Li MEM_TRANS_RETIRED.LOAD_LATENCY
721.Pq Event CDH , Umask 01H
722Sample loads with specified latency threshold.
723PMC3 only.
724Specify threshold in MSR 0x3F6.
725.It Li MEM_TRANS_RETIRED.PRECISE_STORE
726.Pq Event CDH , Umask 02H
727Sample stores and collect precise store operation via PEBS record.
728PMC3 only.
729.It Li MEM_UOP_RETIRED.LOADS
730.Pq Event D0H , Umask 01H
731Qualify retired memory uops that are loads. Combine with umask 10H, 20H,
73240H, 80H.
733Supports PEBS.
734.It Li MEM_UOP_RETIRED.STORES
735.Pq Event D0H , Umask 02H
736Qualify retired memory uops that are stores. Combine with umask 10H, 20H,
73740H, 80H.
738.It Li MEM_UOP_RETIRED.STLB_MISS
739.Pq Event D0H , Umask 10H
740Qualify retired memory uops with STLB miss. Must combine with umask 01H,
74102H, to produce counts.
742.It Li MEM_UOP_RETIRED.LOCK
743.Pq Event D0H , Umask 20H
744Qualify retired memory uops with lock. Must combine with umask 01H, 02H, to
745produce counts.
746.It Li MEM_UOP_RETIRED.SPLIT
747.Pq Event D0H , Umask 40H
748Qualify retired memory uops with line split. Must combine with umask 01H,
74902H, to produce counts.
750.It Li MEM_UOP_RETIRED.ALL
751.Pq Event D0H , Umask 80H
752Qualify any retired memory uops. Must combine with umask 01H, 02H, to
753produce counts.
754.It Li MEM_LOAD_UOPS_RETIRED.L1_HIT
755.Pq Event D1H , Umask 01H
756Retired load uops with L1 cache hits as data sources.
757Supports PEBS.
758.It Li MEM_LOAD_UOPS_RETIRED.L2_HIT
759.Pq Event D1H , Umask 02H
760Retired load uops with L2 cache hits as data sources.
761.It Li MEM_LOAD_UOPS_RETIRED.LLC_HIT
762.Pq Event D1H , Umask 04H
763Retired load uops with LLC cache hits as data sources.
764.It Li MEM_LOAD_UOPS_RETIRED.HIT_LFB
765.Pq Event D1H , Umask 40H
766Retired load uops which data sources were load uops missed L1 but hit FB due
767to preceding miss to the same cache line with data not ready.
768.It Li MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_MISS
769.Pq Event D2H , Umask 01H
770Retired load uops which data sources were LLC hit and cross-core snoop
771missed in on-pkg core cache.
772Supports PEBS.
773.It Li MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_HIT
774.Pq Event D2H , Umask 02H
775Retired load uops which data sources were LLC and cross-core snoop hits in
776on-pkg core cache.
777Supports PEBS.
778.It Li MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_HITM
779.Pq Event D2H , Umask 04H
780Retired load uops which data sources were HitM responses from shared LLC.
781.It Li MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_NONE
782.Pq Event D2H , Umask 08H
783Retired load uops which data sources were hits in LLC without snoops
784required.
785.It Li MEM_LOAD_UOPS_LLC_MISS_RETIRED.LOCAL_DRAM
786.Pq Event D3H , Umask 01H
787Retired load uops which data sources missed LLC but serviced from local
788dram.
789Supports PEBS.
790.It Li L2_TRANS.DEMAND_DATA_RD
791.Pq Event F0H , Umask 01H
792Demand Data Read requests that access L2 cache.
793.It Li L2_TRANS.RFO
794.Pq Event F0H , Umask 02H
795RFO requests that access L2 cache.
796.It Li L2_TRANS.CODE_RD
797.Pq Event F0H , Umask 04H
798L2 cache accesses when fetching instructions.
799.It Li L2_TRANS.ALL_PF
800.Pq Event F0H , Umask 08H
801Any MLC or LLC HW prefetch accessing L2, including rejects.
802.It Li L2_TRANS.L1D_WB
803.Pq Event F0H , Umask 10H
804L1D writebacks that access L2 cache.
805.It Li L2_TRANS.L2_FILL
806.Pq Event F0H , Umask 20H
807L2 fill requests that access L2 cache.
808.It Li L2_TRANS.L2_WB
809.Pq Event F0H , Umask 40H
810L2 writebacks that access L2 cache.
811.It Li L2_TRANS.ALL_REQUESTS
812.Pq Event F0H , Umask 80H
813Transactions accessing L2 pipe.
814.It Li L2_LINES_IN.I
815.Pq Event F1H , Umask 01H
816L2 cache lines in I state filling L2.
817Counting does not cover rejects.
818.It Li L2_LINES_IN.S
819.Pq Event F1H , Umask 02H
820L2 cache lines in S state filling L2.
821Counting does not cover rejects.
822.It Li L2_LINES_IN.E
823.Pq Event F1H , Umask 04H
824L2 cache lines in E state filling L2.
825Counting does not cover rejects.
826.It Li L2_LINES_IN.ALL
827.Pq Event F1H , Umask 07H
828L2 cache lines filling L2.
829Counting does not cover rejects.
830.It Li L2_LINES_OUT.DEMAND_CLEAN
831.Pq Event F2H , Umask 01H
832Clean L2 cache lines evicted by demand.
833.It Li L2_LINES_OUT.DEMAND_DIRTY
834.Pq Event F2H , Umask 02H
835Dirty L2 cache lines evicted by demand.
836.It Li L2_LINES_OUT.PF_CLEAN
837.Pq Event F2H , Umask 04H
838Clean L2 cache lines evicted by the MLC prefetcher.
839.It Li L2_LINES_OUT.PF_DIRTY
840.Pq Event F2H , Umask 08H
841Dirty L2 cache lines evicted by the MLC prefetcher.
842.El
843.Sh SEE ALSO
844.Xr pmc 3 ,
845.Xr pmc.atom 3 ,
846.Xr pmc.core 3 ,
847.Xr pmc.iaf 3 ,
848.Xr pmc.ucf 3 ,
849.Xr pmc.k7 3 ,
850.Xr pmc.k8 3 ,
851.Xr pmc.p4 3 ,
852.Xr pmc.p5 3 ,
853.Xr pmc.p6 3 ,
854.Xr pmc.corei7 3 ,
855.Xr pmc.corei7uc 3 ,
856.Xr pmc.sandybridge 3 ,
857.Xr pmc.sandybridgeuc 3 ,
858.Xr pmc.westmere 3 ,
859.Xr pmc.westmereuc 3 ,
860.Xr pmc.soft 3 ,
861.Xr pmc.tsc 3 ,
862.Xr pmc_cpuinfo 3 ,
863.Xr pmclog 3 ,
864.Xr hwpmc 4
865.Sh HISTORY
866The
867.Nm pmc
868library first appeared in
869.Fx 6.0 .
870.Sh AUTHORS
871The
872.Lb libpmc
873library was written by
874.An "Joseph Koshy"
875.Aq jkoshy@FreeBSD.org .
876The support for the Ivy Bridge
877microarchitecture was written by
878.An "Fabien Thomas"
879.Aq fabient@FreeBSD.org .
880