xref: /freebsd/lib/libpmc/pmc.sandybridge.3 (revision 8ddb146abcdf061be9f2c0db7e391697dafad85c)
1.\" Copyright (c) 2012 Davide Italiano <davide@FreeBSD.org>
2.\" All rights reserved.
3.\"
4.\" Redistribution and use in source and binary forms, with or without
5.\" modification, are permitted provided that the following conditions
6.\" are met:
7.\" 1. Redistributions of source code must retain the above copyright
8.\"    notice, this list of conditions and the following disclaimer.
9.\" 2. Redistributions in binary form must reproduce the above copyright
10.\"    notice, this list of conditions and the following disclaimer in the
11.\"    documentation and/or other materials provided with the distribution.
12.\"
13.\" THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
14.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
15.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
16.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
17.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
18.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
19.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
20.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
21.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
22.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
23.\" SUCH DAMAGE.
24.\"
25.\" $FreeBSD$
26.\"
27.Dd October 19, 2012
28.Dt PMC.SANDYBRIDGE 3
29.Os
30.Sh NAME
31.Nm pmc.sandybridge
32.Nd measurement events for
33.Tn Intel
34.Tn Sandy Bridge
35family CPUs
36.Sh LIBRARY
37.Lb libpmc
38.Sh SYNOPSIS
39.In pmc.h
40.Sh DESCRIPTION
41.Tn Intel
42.Tn "Sandy Bridge"
43CPUs contain PMCs conforming to the version 3 of the
44.Tn Intel
45performance measurement architecture.
46These CPUs may contain up to three classes of PMCs:
47.Bl -tag -width "Li PMC_CLASS_IAP"
48.It Li PMC_CLASS_IAF
49Fixed-function counters that count only one hardware event per counter.
50.It Li PMC_CLASS_IAP
51Programmable counters that may be configured to count one of a defined
52set of hardware events.
53.It Li PMC_CLASS_TSC
54These PMCs are documented in
55.Xr pmc.tsc 3 .
56.El
57.Pp
58The number of PMCs available in each class and their widths need to be
59determined at run time by calling
60.Xr pmc_cpuinfo 3 .
61.Pp
62Intel Sandy Bridge PMCs are documented in
63.Rs
64.%B "Intel(R) 64 and IA-32 Architectures Software Developers Manual"
65.%T "Volume 3B: System Programming Guide, Part 2"
66.%N "Order Number: 253669-039US"
67.%D May 2011
68.%Q "Intel Corporation"
69.Re
70.Ss SANDY BRIDGE FIXED FUNCTION PMCS
71These PMCs and their supported events are documented in
72.Xr pmc.iaf 3 .
73.Ss SANDY BRIDGE PROGRAMMABLE PMCS
74The programmable PMCs support the following capabilities:
75.Bl -column "PMC_CAP_INTERRUPT" "Support"
76.It Em Capability Ta Em Support
77.It PMC_CAP_CASCADE Ta \&No
78.It PMC_CAP_EDGE Ta Yes
79.It PMC_CAP_INTERRUPT Ta Yes
80.It PMC_CAP_INVERT Ta Yes
81.It PMC_CAP_READ Ta Yes
82.It PMC_CAP_PRECISE Ta \&No
83.It PMC_CAP_SYSTEM Ta Yes
84.It PMC_CAP_TAGGING Ta \&No
85.It PMC_CAP_THRESHOLD Ta Yes
86.It PMC_CAP_USER Ta Yes
87.It PMC_CAP_WRITE Ta Yes
88.El
89.Ss Event Qualifiers
90Event specifiers for these PMCs support the following common
91qualifiers:
92.Bl -tag -width indent
93.It Li rsp= Ns Ar value
94Configure the Off-core Response bits.
95.Bl -tag -width indent
96.It Li REQ_DMND_DATA_RD
97Counts the number of demand and DCU prefetch data reads of full and partial
98cachelines as well as demand data page table entry cacheline reads.
99Does not count L2 data read prefetches or instruction fetches.
100.It Li REQ_DMND_RFO
101Counts the number of demand and DCU prefetch reads for ownership (RFO)
102requests generated by a write to data cacheline.
103Does not count L2 RFO prefetches.
104.It Li REQ_DMND_IFETCH
105Counts the number of demand and DCU prefetch instruction cacheline reads.
106Does not count L2 code read prefetches.
107.It Li REQ_WB
108Counts the number of writeback (modified to exclusive) transactions.
109.It Li REQ_PF_DATA_RD
110Counts the number of data cacheline reads generated by L2 prefetchers.
111.It Li REQ_PF_RFO
112Counts the number of RFO requests generated by L2 prefetchers.
113.It Li REQ_PF_IFETCH
114Counts the number of code reads generated by L2 prefetchers.
115.It Li REQ_PF_LLC_DATA_RD
116L2 prefetcher to L3 for loads.
117.It Li REQ_PF_LLC_RFO
118RFO requests generated by L2 prefetcher
119.It Li REQ_PF_LLC_IFETCH
120L2 prefetcher to L3 for instruction fetches.
121.It Li REQ_BUS_LOCKS
122Bus lock and split lock requests.
123.It Li REQ_STRM_ST
124Streaming store requests.
125.It Li REQ_OTHER
126Any other request that crosses IDI, including I/O.
127.It Li RES_ANY
128Catch all value for any response types.
129.It Li RES_SUPPLIER_NO_SUPP
130No Supplier Information available.
131.It Li RES_SUPPLIER_LLC_HITM
132M-state initial lookup stat in L3.
133.It Li RES_SUPPLIER_LLC_HITE
134E-state.
135.It Li RES_SUPPLIER_LLC_HITS
136S-state.
137.It Li RES_SUPPLIER_LLC_HITF
138F-state.
139.It Li RES_SUPPLIER_LOCAL
140Local DRAM Controller.
141.It Li RES_SNOOP_SNP_NONE
142No details on snoop-related information.
143.It Li RES_SNOOP_SNP_NO_NEEDED
144No snoop was needed to satisfy the request.
145.It Li RES_SNOOP_SNP_MISS
146A snoop was needed and it missed all snooped caches:
147-For LLC Hit, ReslHitl was returned by all cores
148-For LLC Miss, Rspl was returned by all sockets and data was returned from
149DRAM.
150.It Li RES_SNOOP_HIT_NO_FWD
151A snoop was needed and it hits in at least one snooped cache.
152Hit denotes a cache-line was valid before snoop effect.
153This includes:
154-Snoop Hit w/ Invalidation (LLC Hit, RFO)
155-Snoop Hit, Left Shared (LLC Hit/Miss, IFetch/Data_RD)
156-Snoop Hit w/ Invalidation and No Forward (LLC Miss, RFO Hit S)
157In the LLC Miss case, data is returned from DRAM.
158.It Li RES_SNOOP_HIT_FWD
159A snoop was needed and data was forwarded from a remote socket.
160This includes:
161-Snoop Forward Clean, Left Shared (LLC Hit/Miss, IFetch/Data_RD/RFT).
162.It Li RES_SNOOP_HITM
163A snoop was needed and it HitM-ed in local or remote cache.
164HitM denotes a cache-line was in modified state before effect as a results of snoop.
165This includes:
166-Snoop HitM w/ WB (LLC miss, IFetch/Data_RD)
167-Snoop Forward Modified w/ Invalidation (LLC Hit/Miss, RFO)
168-Snoop MtoS (LLC Hit, IFetch/Data_RD).
169.It Li RES_NON_DRAM
170Target was non-DRAM system address.
171This includes MMIO transactions.
172.El
173.It Li cmask= Ns Ar value
174Configure the PMC to increment only if the number of configured
175events measured in a cycle is greater than or equal to
176.Ar value .
177.It Li edge
178Configure the PMC to count the number of de-asserted to asserted
179transitions of the conditions expressed by the other qualifiers.
180If specified, the counter will increment only once whenever a
181condition becomes true, irrespective of the number of clocks during
182which the condition remains true.
183.It Li inv
184Invert the sense of comparison when the
185.Dq Li cmask
186qualifier is present, making the counter increment when the number of
187events per cycle is less than the value specified by the
188.Dq Li cmask
189qualifier.
190.It Li os
191Configure the PMC to count events happening at processor privilege
192level 0.
193.It Li usr
194Configure the PMC to count events occurring at privilege levels 1, 2
195or 3.
196.El
197.Pp
198If neither of the
199.Dq Li os
200or
201.Dq Li usr
202qualifiers are specified, the default is to enable both.
203.Ss Event Specifiers (Programmable PMCs)
204Sandy Bridge programmable PMCs support the following events:
205.Bl -tag -width indent
206.It Li LD_BLOCKS.DATA_UNKNOWN
207.Pq EVENT_03H, Umask 01H
208Blocked loads due to store buffer blocks with unknown data.
209.It Li LD_BLOCKS.STORE_FORWARD
210.Pq Event 03H, Umask 02H
211Loads blocked by overlapping with store buffer that cannot be forwarded.
212.It Li LD_BLOCKS.NO_SR
213.Pq Event 03H, Umask 08H
214# of Split loads blocked due to resource not available.
215.It Li LD_BLOCKS.ALL_BLOCK
216.Pq EVENT_03H, Umask 10H
217Number of cases where any load is blocked but has no DCU miss.
218.It Li  MISALIGN_MEM_REF.LOADS
219.Pq Event 05H, Umask  01H
220Speculative cache-line split load uops dispatched to L1D.
221.It Li MISALIGN_MEM_REF.STORES
222.Pq Event 05H, Umask  02H
223Speculative cache-line split Store-address uops dispatched to L1D.
224.It Li LD_BLOCKS_PARTIAL.ADDRESS_ALIAS
225.Pq Event 07H, Umask  01H
226False dependencies in MOB due to partial compare on address.
227.It Li LD_BLOCKS_PARTIAL.ALL_STA_BLOCK
228.Pq Event 07H, Umask 08H
229The number of times that load operations are temporarily blocked because of
230older stores, with addresses that are not yet known.
231A load operation may incur more than one block of this type.
232.It LI DTLB_LOAD_MISSES.MISS_CAUSES_A_WALK
233.Pq Event 08H, Umask 01H
234Misses in all TLB levels that cause a page walk of any page size.
235.It Li DTLB_LOAD_MISSES.WALK_COMPLETED
236.Pq Event 08H, Umask 02H
237Misses in all TLB levels that caused page walk completed of any size.
238.It Li DTLB_LOAD_MISSES.WALK_DURATION
239.Pq Event 08H, Umask 04H
240Cycle PMH is busy with a walk.
241.It Li DTLB_LOAD_MISSES.STLB_HIT
242.Pq Event 08H, Umask 10H
243Number of cache load STLB hits.
244No page walk.
245.It Li INT_MISC.RECOVERY_CYCLES
246.Pq Event 0DH, Umask 03H
247Cycles waiting to recover after Machine Clears or JEClear.
248Set Cmask = 1.
249Set Edge to count occurrences
250.It Li INT_MISC.RAT_STALL_CYCLES
251.Pq Event 0DH, Umask 40H
252Cycles RAT external stall is sent to IDQ for this thread.
253.It Li UOPS_ISSUED.ANY
254.Pq Event 0EH, Umask 01H
255Increments each cycle the # of Uops issued by the RAT to RS.
256Set Cmask = 1, Inv = 1, Any= 1 to count stalled cycles of this core.
257Set Cmask = 1, Inv = 1 to count stalled cycles
258.It Li FP_COMP_OPS_EXE.X87
259.Pq Event 10H, Umask 01H
260Counts number of X87 uops executed.
261.It Li FP_COMP_OPS_EXE.SSE_FP_PACKED_DOUBLE
262.Pq Event 10H, Umask 10H
263Counts number of SSE* double precision FP packed uops executed.
264.It Li FP_COMP_OPS_EXE.SSE_FP_SCALAR_SINGLE
265.Pq Event 10H, Umask 20H
266Counts number of SSE* single precision FP scalar uops executed.
267.It Li FP_COMP_OPS_EXE.SSE_PACKED_SINGLE
268.Pq Event 10H, Umask 40H
269Counts number of SSE* single precision FP packed uops executed.
270.It LiFP_COMP_OPS_EXE.SSE_SCALAR_DOUBLE
271.Pq Event 10H, Umask 80H
272Counts number of SSE* double precision FP scalar uops executed.
273.It Li SIMD_FP_256.PACKED_SINGLE
274.Pq Event 11H, Umask 01H
275Counts 256-bit packed single-precision floating-point instructions.
276.It Li SIMD_FP_256.PACKED_DOUBLE
277.Pq Event 11H, Umask 02H
278Counts 256-bit packed double-precision floating-point instructions.
279.It Li ARITH.FPU_DIV_ACTIVE
280.Pq Event 14H, Umask 01H
281Cycles that the divider is active, includes INT and FP.
282Set 'edge =1, cmask=1' to count the number of divides.
283.It Li INSTS_WRITTEN_TO_IQ.INSTS
284.Pq Event 17H, Umask 01H
285Counts the number of instructions written into the IQ every cycle.
286.It Li L2_RQSTS.DEMAND_DATA_RD_HIT
287.Pq Event 24H, Umask 01H
288Demand Data Read requests that hit L2 cache.
289.It Li L2_RQSTS.ALL_DEMAND_DATA_RD
290.Pq Event 24H, Umask 03H
291Counts any demand and L1 HW prefetch data load requests to L2.
292.It Li L2_RQSTS.RFO_HITS
293.Pq Event 24H, Umask 04H
294Counts the number of store RFO requests that hit the L2 cache.
295.It Li L2_RQSTS.RFO_MISS
296.Pq Event 24H, Umask 08H
297Counts the number of store RFO requests that miss the L2 cache.
298.It Li L2_RQSTS.ALL_RFO
299.Pq Event 24H, Umask 0CH
300Counts all L2 store RFO requests.
301.It Li L2_RQSTS.CODE_RD_HIT
302.Pq Event 24H, Umask 10H
303Number of instruction fetches that hit the L2 cache.
304.It Li L2_RQSTS.CODE_RD_MISS
305.Pq Event 24H, Umask 20H
306Number of instruction fetches that missed the L2 cache.
307.It Li L2_RQSTS.ALL_CODE_RD
308.Pq Event 24H, Umask 30H
309Counts all L2 code requests.
310.It Li L2_RQSTS.PF_HIT
311.Pq Event 24H, Umask 40H
312Requests from L2 Hardware prefetcher that hit L2.
313.It Li L2_RQSTS.PF_MISS
314.Pq Event 24H, Umask 80H
315Requests from L2 Hardware prefetcher that missed L2.
316.It Li L2_RQSTS.ALL_PF
317.Pq Event 24H, Umask C0H
318Any requests from L2 Hardware prefetchers.
319.It Li L2_STORE_LOCK_RQSTS.MISS
320.Pq Event 27H, Umask 01H
321RFOs that miss cache lines.
322.It Li L2_STORE_LOCK_RQSTS.HIT_E
323.Pq Event 27H, Umask 04H
324RFOs that hit cache lines in E state.
325.It Li L2_STORE_LOCK_RQSTS.HIT_M
326.Pq EVENT_27H, Umask 08H
327RFOs that hit cache lines in M state.
328.It Li L2_STORE_LOCK_RQSTS.ALL
329.Pq EVENT_27H, Umask 0FH
330RFOs that access cache lines in any state.
331.It Li L2_L1D_WB_RQSTS.HIT_E
332.Pq Event 28H, Umask 04H
333Not rejected writebacks from L1D to L2 cache lines in E state.
334.It Li L2_L1D_WB_RQSTS.HIT_M
335.Pq Event 28H, Umask 08H
336Not rejected writebacks from L1D to L2 cache lines in M state.
337.It Li LONGEST_LAT_CACHE.REFERENCE
338.Pq Event 2EH, Umask 4FH
339This event counts requests originating from the core that reference a cache
340line in the last level cache.
341.It Li LONGEST_LAT_CACHE.MISS
342.Pq Event 2EH, Umask 41H
343This event counts each cache miss condition for references to the last level
344cache.
345.It Li CPU_CLK_UNHALTED.THREAD_P
346.Pq Event 3CH, Umask 00H
347Counts the number of thread cycles while the thread is not in a halt state.
348The thread enters the halt state when it is running the HLT instruction.
349The core frequency may change from time to time due to power or thermal
350throttling.
351.It Li CPU_CLK_THREAD_UNHALTED.REF_XCLK
352.Pq Event 3CH, Umask 01H
353Increments at the frequency of XCLK (100 MHz) when not halted.
354.It Li L1D_PEND_MISS.PENDING
355.Pq Event 48H, Umask 01H
356Increments the number of outstanding L1D misses every cycle.
357Set Cmask = 1 and Edge =1  to count occurrences.
358Counter 2 only; Set Cmask = 1 to count cycles.
359.It Li DTLB_STORE_MISSES.MISS_CAUSES_A_WALK
360.Pq Event 49H, Umask 01H Miss in all TLB levels causes an page walk of any
361page size (4K/2M/4M/1G).
362.It Li DTLB_STORE_MISSES.WALK_COMPLETED
363.Pq Event 49H, Umask 02H
364Miss in all TLB levels causes a page walk that completes of any page size
365(4K/2M/4M/1G).
366.It Li DTLB_STORE_MISSES.WALK_DURATION
367.Pq Event 49H, Umask 04H
368Cycles PMH is busy with this walk.
369.It Li DTLB_STORE_MISSES.STLB_HIT
370.Pq Event 49H, Umask 10H
371Store operations that miss the first TLB level but hit the second and do not
372cause page walks.
373.It Li LOAD_HIT_PRE.SW_PF
374.Pq Event 4CH, Umask 01H
375Not SW-prefetch load dispatches that hit fill buffer allocated for S/W prefetch.
376.It Li LOAD_HIT_PER.HW_PF
377.Pq Event 4CH, Umask 02H
378Not SW-prefetch load dispatches that hit fill buffer allocated for H/W prefetch.
379.It Li HW_PRE_REQ.DL1_MISS
380.Pq Event 4EH, Umask 02H
381Hardware Prefetch requests that miss the L1D cache.
382A request is being counted each time it access the cache & miss it, including
383if a block is applicable or if hit the Fill Buffer for example.
384This accounts for both L1 streamer and IP-based (IPP) HW prefetchers.
385.It Li L1D.REPLACEMENT
386.Pq Event 51H, Umask 01H
387Counts the number of lines brought into the L1 data cache.
388.It Li L1D.ALLOCATED_IN_M
389.Pq Event 51H, Umask 02H
390Counts the number of allocations of modified L1D cache lines.
391.It Li L1D.EVICTION
392.Pq Event 51H, Umask 04H
393Counts the number of modified lines evicted from the L1 data cache due to
394replacement.
395.It Li L1D.ALL_M_REPLACEMENT
396.Pq Event 51H, Umask 08H
397Cache lines in M state evicted out of L1D due to Snoop HitM or dirty line
398replacement.
399.It Li PARTIAL_RAT_STALLS.FLAGS_MERGE_UOP
400.Pq Event 59H, Umask 20H
401Increments the number of flags-merge uops in flight each cycle.
402Set Cmask = 1 to count cycles.
403.It Li PARTIAL_RAT_STALLS.SLOW_LEA_WINDOW
404.Pq Event 59H, Umask 40H
405Cycles with at least one slow LEA uop allocated.
406.It Li PARTIAL_RAT_STALLS.MUL_SINGLE_UOP
407.Pq Event 59H, Umask 80H
408Number of Multiply packed/scalar single precision uops allocated.
409.It Li RESOURCE_STALLS2.ALL_FL_EMPTY
410.Pq Event 5BH, Umask 0CH
411Cycles stalled due to free list empty.
412.It Li RESOURCE_STALLS2.ALL_PRF_CONTROL
413.Pq Event 5BH, Umask 0FH
414Cycles stalled due to control structures full for physical registers.
415.It Li RESOURCE_STALLS2.BOB_FULL
416.Pq Event 5BH, Umask 40H
417Cycles Allocator is stalled due to Branch Order Buffer.
418.It Li RESOURCE_STALLS2.OOO_RSRC
419.Pq Event 5BH, Umask 4FH
420Cycles stalled due to out of order resources full.
421.It Li CPL_CYCLES.RING0
422.Pq Event 5CH, Umask 01H
423Unhalted core cycles when the thread is in ring 0.
424Use Edge to count transition
425.It Li CPL_CYCLES.RING123
426.Pq Event 5CH, Umask 02H
427Unhalted core cycles when the thread is not in ring 0.
428.It Li RS_EVENTS.EMPTY_CYCLES
429.Pq Event 5EH, Umask 01H
430Cycles the RS is empty for the thread.
431.It Li OFFCORE_REQUESTS_OUTSTANDING.DEMAND_DATA_RD
432.Pq Event 60H, Umask 01H
433Offcore outstanding Demand Data Read transactions in SQ to uncore.
434Set Cmask=1 to count cycles.
435.It Li OFFCORE_REQUESTS_OUTSTANDING.DEMAND_RFO
436.Pq Event 60H, Umask 04H
437Offcore outstanding RFO store transactions in SQ to uncore.
438Set Cmask=1 to count cycles.
439.It Li OFFCORE_REQUESTS_OUTSTANDING.ALL_DATA_RD
440.Pq Event 60H, Umask 08H
441Offcore outstanding cacheable data read transactions in SQ to uncore.
442Set Cmask=1 to count cycles.
443.It Li LOCK_CYCLES.SPLIT_LOCK_UC_LOCK_DURATION
444.Pq Event 63H, Umask 01H
445Cycles in which the L1D and L2 are locked, due to a UC lock or split lock.
446.It Li LOCK_CYCLES.CACHE_LOCK_DURATION
447.Pq Event 63H, Umask 02H
448Cycles in which the L1D is locked.
449.It Li IDQ.EMPTY
450.Pq Event 79H, Umask 02H
451Counts cycles the IDQ is empty.
452.It Li IQD.MITE_UOPS
453.Pq Event 79H, Umask 04H
454Increment each cycle # of uops delivered to IDQ from MITE path.
455Set Cmask = 1 to count cycles.
456Can combine Umask 04H and 20H
457.It Li IDQ.DSB_UOPS
458.Pq Event 79H, Umask 08H
459Increment each cycle.
460# of uops delivered to IDQ from DSB path.
461Set Cmask = 1 to count cycles.
462Can combine Umask 08H and 10H
463.It Li IDQ.MS_DSB_UOPS
464.Pq Event 79H, Umask 10H
465Increment each cycle # of uops delivered to IDQ when MS busy by DSB.
466Set Cmask = 1 to count cycles MS is busy.
467Set Cmask=1 and Edge=1 to count MS activations.
468Can combine Umask 08H and 10H
469.It Li IDQ.MS_MITE_UOPS
470.Pq Event 79H, Umask 20H
471Increment each cycle # of uops delivered to IDQ when MS is busy by MITE.
472Set Cmask = 1 to count cycles.
473Can combine Umask 04H and 20H
474.It Li IDQ.MS_UOPS
475.Pq Event 79H, Umask 30H
476Increment each cycle # of uops delivered to IDQ from MS by either DSB or MITE.
477Set Cmask = 1 to count cycles.
478Can combine Umask 04H, 08H and 30H
479.It Li ICACHE.MISSES
480.Pq Event 80H, Umask 02H
481Number of Instruction Cache, Streaming Buffer and Victim Cache Misses.
482Includes UC accesses.
483.It Li ITLB_MISSES.MISS_CAUSES_A_WALK
484.Pq Event 85H, Umask 01H
485Misses in all ITLB levels that cause page walks.
486.It Li ITLB_MISSES.WALK_COMPLETED
487.Pq Event 85H, Umask 02H
488Misses in all ITLB levels that cause completed page walks.
489.It Li ITLB_MISSES.WALK_DURATION
490.Pq Event 85H, Umask 04H
491Cycle PMH is busy with a walk.
492.It Li ITLB_MISSES.STLB_HIT
493.Pq Event 85H, Umask 10H
494Number of cache load STLB hits.
495No page walk.
496.It Li ILD_STALL.LCP
497.Pq Event 87H, Umask 01H
498Stalls caused by changing prefix length of the instruction.
499.It Li ILD_STALL.IQ_FULL
500.Pq Event 87H, Umask 04H
501Stall cycles due to IQ is full.
502.It Li BR_INST_EXEC.NONTAKEN_COND
503.Pq Event 88H , Umask 41H
504Count conditional near branch instructions that were executed (but not
505necessarily retired) and not taken.
506.It Li BR_INST_EXEC.TAKEN_COND
507.Pq Event 88H , Umask 81H
508Count conditional near branch instructions that were executed (but not
509necessarily retired) and taken.
510.It Li BR_INST_EXEC.DIRECT_JMP
511.Pq Event 88H , Umask 82H
512Count all unconditional near branch instructions excluding calls and
513indirect branches.
514.It Li BR_INST_EXEC.INDIRECT_JMP_NON_CALL_RET
515.Pq Event 88H , Umask 84H
516Count executed indirect near branch instructions that are not calls nor
517returns.
518.It Li BR_INST_EXEC.RETURN_NEAR
519.Pq Event 88H , Umask 88H
520Count indirect near branches that have a return mnemonic.
521.It Li BR_INST_EXEC.DIRECT_NEAR_CALL
522.Pq Event 88H , Umask 90H
523Count unconditional near call branch instructions, excluding non call
524branch, executed.
525.It Li BR_INST_EXEC.INDIRECT_NEAR_CALL
526.Pq Event 88H , Umask A0H
527Count indirect near calls, including both register and memory indirect,
528executed.
529.It Li BR_INST_EXEC.ALL_BRANCHES
530.Pq Event 88H , Umask FFH
531Counts all near executed branches (not necessarily retired).
532.It Li BR_MISP_EXEC.NONTAKEN_COND
533.Pq Event 89H , Umask 41H
534Count conditional near branch instructions mispredicted as nontaken.
535.It Li BR_MISP_EXEC.TAKEN_COND
536.Pq Event 89H , Umask 81H
537Count conditional near branch instructions mispredicted as taken.
538.It Li BR_MISP_EXEC.INDIRECT_JMP_NON_CALL_RET
539.Pq Event 89H , Umask 84H
540Count mispredicted indirect near branch instructions that are not calls
541nor returns.
542.It Li BR_MISP_EXEC.RETURN_NEAR
543.Pq Event 89H , Umask 88H
544Count mispredicted indirect near branches that have a return mnemonic.
545.It Li BR_MISP_EXEC.DIRECT_NEAR_CALL
546.Pq Event 89H , Umask 90H
547Count mispredicted unconditional near call branch instructions, excluding
548non call branch, executed.
549.It Li BR_MISP_EXEC.INDIRECT_NEAR_CALL
550.Pq Event 89H , Umask A0H
551Count mispredicted indirect near calls, including both register and memory
552indirect, executed.
553.It Li BR_MISP_EXEC.ALL_BRANCHES
554.Pq Event 89H , Umask FFH
555Counts all mispredicted near executed branches (not necessarily retired).
556.It Li IDQ_UOPS_NOT_DELIVERED.CORE
557.Pq Event 9CH, Umask 01H
558Count number of non-delivered uops to RAT per thread.
559Use Cmask to qualify uop b/w
560.It Li UOPS_DISPATCHED_PORT.PORT_0
561.Pq Event A1H, Umask 01H
562Cycles which a Uop is dispatched on port 0.
563.It Li UOPS_DISPATCHED_PORT.PORT_1
564.Pq Event A1H, Umask 02H
565Cycles which a Uop is dispatched on port 1.
566.It Li UOPS_DISPATCHED_PORT.PORT_2_LD
567.Pq Event A1H, Umask 04H
568Cycles which a load uop is dispatched on port 2.
569.It Li UOPS_DISPATCHED_PORT.PORT_2_STA
570.Pq Event A1H, Umask 08H
571Cycles which a store address uop is dispatched on port 2.
572.It Li UOPS_DISPATCHED_PORT.PORT_2
573.Pq Event A1H, Umask 0CH
574Cycles which a Uop is dispatched on port 2.
575.It Li UOPS_DISPATCHED_PORT.PORT_3_LD
576.Pq Event A1H, Umask 10H
577Cycles which a load uop is dispatched on port 3.
578.It Li UOPS_DISPATCHED_PORT.PORT_3_STA
579.Pq Event A1H, Umask 20H
580Cycles which a store address uop is dispatched on port 3.
581.It Li UOPS_DISPATCHED_PORT.PORT_3
582.Pq Event A1H, Umask 30H
583.Pq Cycles which a Uop is dispatched on port 3.
584.It Li UOPS_DISPATCHED_PORT.PORT_4
585.Pq Event A1H, Umask 40H
586Cycles which a Uop is dispatched on port 4.
587.It Li UOPS_DISPATCHED_PORT.PORT_5
588.Pq Event A1H, Umask 80H
589Cycles which a Uop is dispatched on port 5.
590.It Li RESOURCE_STALLS.ANY
591.Pq Event A2H, Umask 01H
592Cycles Allocation is stalled due to Resource Related reason.
593.It Li RESOURCE_STALLS.LB
594.Pq Event A2H, Umask 02H
595Counts the cycles of stall due to lack of load buffers.
596.It Li RESOURCE_STALLS.LB
597.Pq Event A2H, Umask 04H
598Cycles stalled due to no eligible RS entry available.
599.It Li RESOURCE_STALLS.SB
600.Pq Event A2H, Umask 08H
601Cycles stalled due to no store buffers available.
602(not including draining form sync)
603.It Li RESOURCE_STALLS.ROB
604.Pq Event A2H, Umask 10H
605Cycles stalled due to re-order buffer full.
606.It Li RESOURCE_STALLS.FCSW
607.Pq Event A2H, Umask 20H
608Cycles stalled due to writing the FPU control word.
609.It Li RESOURCE_STALLS.MXCSR
610.Pq Event A2H, Umask 40H
611Cycles stalled due to the MXCSR register rename occurring to close to a previous
612MXCSR rename.
613.It Li RESOURCE_STALLS.OTHER
614.Pq Event A2H, Umask 80H
615Cycles stalled while execution was stalled due to other resource issues.
616.It Li DSB2MITE_SWITCHES.COUNT
617.Pq Event ABH, Umask 01H
618Number of DSB to MITE switches.
619.It Li DSB2MITE_SWITCHES.PENALTY_CYCLES
620.Pq Event ABH, Umask 02H
621Cycles DSB to MITE switches caused delay.
622.It Li DSB_FILL.OTHER_CANCEL
623.Pq Event ACH, Umask 02H
624Cases of cancelling valid DSB fill not because of exceeding way limit.
625.It Li DSB_FILL.EXCEED_DSB_LINES
626.Pq Event ACH, Umask 08H
627DSB Fill encountered > 3 DSB lines.
628.It Li DSB_FILL.ALL_CANCEL
629.Pq Event ACH, Umask 0AH
630Cases of cancelling valid Decode Stream Buffer (DSB) fill not because of exceeding
631way limit.
632.It Li ITLB.ITLB_FLUSH
633.Pq Event AEH, Umask 01H
634Counts the number of ITLB flushes, includes 4k/2M/4M pages.
635.It Li OFFCORE_REQUESTS.DEMAND_DATA_RD
636.Pq Event B0H, Umask 01H
637Demand data read requests sent to uncore.
638.It Li OFFCORE_REQUESTS.DEMAND_RFO
639.Pq Event B0H, Umask 04H
640Demand RFO read requests sent to uncore, including regular RFOs, locks, ItoM.
641.It Li OFFCORE_REQUESTS.ALL_DATA_RD
642.Pq Event B0H, Umask 08H
643Data read requests sent to uncore (demand and prefetch).
644.It Li UOPS_DISPATCHED.THREAD
645.Pq Event B1H, Umask 01H
646Counts total number of uops to be dispatched per-thread each cycle.
647Set Cmask = 1, INV =1 to count stall cycles.
648.It Li UOPS_DISPATCHED.CORE
649.Pq Event B1H, Umask 02H
650Counts total number of uops to be dispatched per-core each cycle.
651Do not need to set ANY
652.It Li OFFCORE_REQUESTS_BUFFER.SQ_FULL
653.Pq Event B2H, Umask 01H
654Offcore requests buffer cannot take more entries for this thread core.
655.It Li AGU_BYPASS_CANCEL.COUNT
656.Pq Event B6H, Umask 01H
657Counts executed load operations with all the following traits: 1. addressing
658of the format [base + offset], 2. the offset is between 1 and 2047, 3. the
659address specified in the base register is in one page and the address
660[base+offset] is in another page.
661.It Li OFF_CORE_RESPONSE_0
662.Pq Event B7H, Umask 01H
663Off-core Response Performance Monitoring; PMC0 only.
664Requires programming MSR 01A6H
665.It Li OFF_CORE_RESPONSE_1
666.Pq Event BBH, Umask 01H
667Off-core Response Performance Monitoring.
668PMC3 only.
669Requires programming MSR 01A7H
670.It Li TLB_FLUSH.DTLB_THREAD
671.Pq Event BDH, Umask 01H
672DTLB flush attempts of the thread-specific entries.
673.It Li TLB_FLUSH.STLB_ANY
674.Pq Event BDH, Umask 20H
675Count number of STLB flush attempts.
676.It Li L1D_BLOCKS.BANK_CONFLICT_CYCLES
677.Pq Event BFH, Umask 05H
678Cycles when dispatched loads are cancelled due to L1D bank conflicts with other
679load ports.
680cmask=1
681.It Li INST_RETIRED.ANY_P
682.Pq Event C0H, Umask 00H
683Number of instructions at retirement.
684.It Li INST_RETIRED.PREC_DIST
685.Pq Event C0H, Umask 01H
686Precise instruction retired event with HW to reduce effect of PEBS shadow in IP
687distribution PMC1 only; Must quiesce other PMCs.
688.It Li INST_RETIRED.X87
689.Pq Event C0H, Umask 02H
690X87 instruction retired event.
691.It Li OTHER_ASSISTS.ITLB_MISS_RETIRED
692.Pq Event C1H, Umask 02H
693Instructions that experienced an ITLB miss.
694.It Li OTHER_ASSISTS.AVX_STORE
695.Pq Event C1H, Umask 08H
696Number of assists associated with 256-bit AVX store operations.
697.It Li OTHER_ASSISTS.AVX_TO_SSE
698.Pq Event C1H, Umask 10H
699Number of transitions from AVX256 to legacy SSE when penalty applicable.
700.It Li OTHER_ASSISTS.SSE_TO_AVX
701.Pq Event C1H, Umask 20H
702Number of transitions from SSE to AVX-256 when penalty applicable.
703.It Li UOPS_RETIRED.ALL
704.Pq Event C2H, Umask 01H
705Counts the number of micro-ops retired.
706Use cmask=1 and invert to count active cycles or stalled cycles.
707.It Li UOPS_RETIRED.RETIRE_SLOTS
708.Pq Event C2H, Umask 02H
709Counts the number of retirement slots used each cycle.
710.It Li MACHINE_CLEARS.MEMORY_ORDERING
711.Pq Event C3H, Umask 02H
712Counts the number of machine clears due to memory order conflicts.
713.It Li MACHINE_CLEARS.SMC
714.Pq Event C3H, Umask 04H
715Counts the number of times that a program writes to a code section.
716.It Li MACHINE_CLEARS.MASKMOV
717.Pq Event C3H, Umask 20H
718Counts the number of executed AVX masked load operations that refer to an
719illegal address range with the mask bits set to 0.
720.It Li BR_INST_RETIRED.ALL_BRANCH
721.Pq Event C4H, Umask 00H
722Branch instructions at retirement.
723.It Li BR_INST_RETIRED.CONDITIONAL
724.Pq Event C4H, Umask 01H
725Counts the number of conditional branch instructions retired.
726.It Li BR_INST_RETIRED.NEAR_CALL
727.Pq Event C4H, Umask 02H
728Direct and indirect near call instructions retired.
729.It Li BR_INST_RETIRED.ALL_BRANCHES
730.Pq Event C4H, Umask 04H
731Counts the number of branch instructions retired.
732.It Li BR_INST_RETIRED.NEAR_RETURN
733.Pq Event C4H, Umask 08H
734Counts the number of near return instructions retired.
735.It Li BR_INST_RETIRED.NOT_TAKEN
736.Pq Event C4H, Umask 10H
737Counts the number of not taken branch instructions retired.
738.It Li BR_INST_RETIRED.NEAR_TAKEN
739.Pq Event C4H, Umask 20H
740Number of near taken branches retired.
741.It Li BR_INST_RETIRED.FAR_BRANCH
742.Pq Event C4H, Umask 40H
743Number of far branches retired.
744.It Li BR_MISP_RETIRED.ALL_BRANCHES
745.Pq Event C5H, Umask 00H
746Mispredicted branch instructions at retirement.
747.It Li BR_MISP_RETIRED.CONDITIONAL
748.Pq Event C5H, Umask 01H
749Mispredicted conditional branch instructions retired.
750.It Li BR_MISP_RETIRED.NEAR_CALL
751.Pq Event C5H, Umask 02H
752Direct and indirect mispredicted near call instructions retired.
753.It Li BR_MISP_RETIRED.ALL_BRANCH
754.Pq Event C5H, Umask 04H
755Mispredicted macro branch instructions retired.
756.It Li BR_MISP_RETIRED.NOT_TAKEN
757.Pq Event C5H, Umask 10H
758Mispredicted not taken branch instructions retired.
759.It Li BR_MISP_RETIRED.TAKEN
760.Pq Event C5H, Umask 20H
761Mispredicted taken branch instructions retired.
762.It Li FP_ASSIST.X87_OUTPUT
763.Pq Event CAH, Umask 02H
764Number of X87 assists due to output value.
765.It Li FP_ASSIST.X87_INPUT
766.Pq Event CAH, Umask 04H
767Number of X87 assists due to input value.
768.It Li FP_ASSIST.SIMD_OUTPUT
769.Pq Event CAH, Umask 08H
770Number of SIMD FP assists due to Output values.
771.It Li FP_ASSIST.SIMD_INPUT
772.Pq Event CAH, Umask 10H
773Number of SIMD FP assists due to input values.
774.It Li FP_ASSIST.ANY
775.Pq Event CAH, Umask 1EH
776Cycles with any input/output SSE* or FP assists.
777.It Li ROB_MISC_EVENTS.LBR_INSERTS
778.Pq Event CCH, Umask 20H
779Count cases of saving new LBR records by hardware.
780.It Li MEM_TRANS_RETIRED.LOAD_LATENCY
781.Pq Event CDH, Umask 01H
782Sample loads with specified latency threshold.
783PMC3 only.
784Specify threshold in MSR 0x3F6.
785.It Li MEM_TRANS_RETIRED.PRECISE_STORE
786.Pq Event CDH, Umask 02H
787Sample stores and collect precise store operation via PEBS record.
788PMC3 only.
789.It Li MEM_UOP_RETIRED.LOADS
790.Pq Event D0H, Umask 01H
791Qualify retired memory uops that are loads.
792Combine with umask 10H, 20H, 40H, 80H.
793.It Li MEM_UOP_RETIRED.STORES
794.Pq Event D0H, Umask 02H
795Qualify retired memory uops that are stores.
796Combine with umask 10H, 20H, 40H, 80H.
797.It Li MEM_UOP_RETIRED.STLB_MISS
798.Pq Event D0H, Umask 10H
799Qualify retired memory uops with STLB miss.
800Must combine with umask 01H, 02H, to produce counts.
801.It Li MEM_UOP_RETIRED.LOCK
802.Pq Event D0H, Umask 20H
803Qualify retired memory uops with lock.
804Must combine with umask 01H, 02H, to produce counts.
805.It Li MEM_UOP_RETIRED.SPLIT
806.Pq Event D0H, Umask 40H
807Qualify retired memory uops with line split.
808Must combine with umask 01H, 02H, to produce counts.
809.It Li MEM_UOP_RETIRED_ALL
810.Pq Event D0H, Umask 80H
811Qualify any retired memory uops.
812Must combine with umask 01H, 02H, to produce counts.
813.It Li MEM_LOAD_UOPS_RETIRED.L1_HIT
814.Pq Event D1H, Umask 01H
815Retired load uops with L1 cache hits as data sources.
816Must combine with umask 01H, 02H, to produce counts.
817.It Li MEM_LOAD_UOPS_RETIRED.L2_HIT
818.Pq Event D1H, Umask 02H
819Retired load uops with L2 cache hits as data sources.
820.It Li MEM_LOAD_UOPS_RETIRED.LLC_HIT
821.Pq Event D1H, Umask 04H
822Retired load uops which data sources were data hits in LLC without snoops
823required.
824.It Li MEM_LOAD_UOPS_RETIRED.HIT_LFB
825.Pq Event D1H, Umask 40H
826Retired load uops which data sources were load uops missed L1 but hit FB due
827to preceding miss to the same cache line with data not ready.
828.It Li MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_MISS
829.Pq Event D2H, Umask 01H
830Retired load uops which data sources were LLC hit and cross-core snoop missed in
831on-pkg core cache.
832.It Li MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_HIT
833.Pq Event D2H, Umask 02H
834Retired load uops which data sources were LLC and cross-core snoop hits in
835on-pkg core cache.
836.It Li MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_HITM
837.Pq Event D2H, Umask 04H
838Retired load uops which data sources were HitM responses from shared LLC.
839.It Li MEM_LOAD_UOPS_LLC_HIT_RETIRED.XSNP_NONE
840.Pq Event D2H, Umask 08H
841Retired load uops which data sources were hits in LLC without snoops required.
842.It Li MEM_LOAD_UOPS_LLC_HIT_RETIRED.LLC_MISS
843.Pq Event D4H, Umask 02H
844Retired load uops with unknown information as data source in cache serviced the load.
845.It Li L2_TRANS.DEMAND_DATA_RD
846.Pq Event F0H, Umask 01H
847Demand Data Read requests that access L2 cache.
848.It Li L2_TRANS.RF0
849.Pq Event F0H, Umask 02H
850RFO requests that access L2 cache.
851.It Li L2_TRANS.CODE_RD
852.Pq Event F0H, Umask 04H
853L2 cache accesses when fetching instructions.
854.It Li L2_TRANS.ALL_PF
855.Pq Event F0H, Umask 08H
856L2 or LLC HW prefetches that access L2 cache.
857.It Li L2_TRANS.L1D_WB
858.Pq Event F0H, Umask 10H
859L1D writebacks that access L2 cache.
860.It Li L2_TRANS.L2_FILL
861.Pq Event F0H, Umask 20H
862L2 fill requests that access L2 cache.
863.It Li L2_TRANS.L2_WB
864.Pq Event F0H, Umask 40H
865L2 writebacks that access L2 cache.
866.It Li L2_TRANS.ALL_REQUESTS
867.Pq Event F0H, Umask 80H
868Transactions accessing L2 pipe.
869.It Li L2_LINES_IN.I
870.Pq Event F1H, Umask 01H
871L2 cache lines in I state filling L2.
872Counting does not cover rejects.
873.It Li L2_LINES_IN.S
874.Pq Event F1H, Umask 02H
875L2 cache lines in S state filling L2.
876Counting does not cover rejects.
877.It Li L2_LINES_IN.E
878.Pq Event F1H, Umask 04H
879L2 cache lines in E state filling L2.
880Counting does not cover rejects.
881.It Li L2_LINES-IN.ALL
882.Pq Event F1H, Umask 07H
883L2 cache lines filling L2.
884Counting does not cover rejects.
885.It Li L2_LINES_OUT.DEMAND_CLEAN
886.Pq Event F2H, Umask 01H
887Clean L2 cache lines evicted by demand.
888.It Li L2_LINES_OUT.DEMAND_DIRTY
889.Pq Event F2H, Umask 02H
890Dirty L2 cache lines evicted by demand.
891.It Li L2_LINES_OUT.PF_CLEAN
892.Pq Event F2H, Umask 04H
893Clean L2 cache lines evicted by L2 prefetch.
894.It Li L2_LINES_OUT.PF_DIRTY
895.Pq Event F2H, Umask 08H
896Dirty L2 cache lines evicted by L2 prefetch.
897.It Li L2_LINES_OUT.DIRTY_ALL
898.Pq Event F2H, Umask 0AH
899Dirty L2 cache lines filling the L2.
900Counting does not cover rejects.
901.It Li SQ_MISC.SPLIT_LOCK
902.Pq Event F4H, Umask 10H
903Split locks in SQ.
904.El
905.Sh SEE ALSO
906.Xr pmc 3 ,
907.Xr pmc.atom 3 ,
908.Xr pmc.core 3 ,
909.Xr pmc.corei7 3 ,
910.Xr pmc.corei7uc 3 ,
911.Xr pmc.iaf 3 ,
912.Xr pmc.ivybridge 3 ,
913.Xr pmc.ivybridgexeon 3 ,
914.Xr pmc.k7 3 ,
915.Xr pmc.k8 3 ,
916.Xr pmc.sandybridgeuc 3 ,
917.Xr pmc.sandybridgexeon 3 ,
918.Xr pmc.soft 3 ,
919.Xr pmc.tsc 3 ,
920.Xr pmc.ucf 3 ,
921.Xr pmc.westmere 3 ,
922.Xr pmc.westmereuc 3 ,
923.Xr pmc_cpuinfo 3 ,
924.Xr pmclog 3 ,
925.Xr hwpmc 4
926.Sh HISTORY
927The
928.Nm pmc
929library first appeared in
930.Fx 6.0 .
931.Sh AUTHORS
932.An -nosplit
933The
934.Lb libpmc
935library was written by
936.An Joseph Koshy Aq Mt jkoshy@FreeBSD.org .
937The support for the Sandy Bridge
938microarchitecture was written by
939.An Davide Italiano Aq Mt davide@FreeBSD.org .
940