xref: /freebsd/lib/libpmc/pmc.corei7uc.3 (revision f5463265955b829775bbb32e1fd0bc11dafc36ce)
1.\" Copyright (c) 2010 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.Dd March 24, 2010
25.Dt PMC.COREI7UC 3
26.Os
27.Sh NAME
28.Nm pmc.corei7uc
29.Nd uncore measurement events for
30.Tn Intel
31.Tn Core i7 and Xeon 5500
32family CPUs
33.Sh LIBRARY
34.Lb libpmc
35.Sh SYNOPSIS
36.In pmc.h
37.Sh DESCRIPTION
38.Tn Intel
39.Tn "Core i7"
40CPUs contain PMCs conforming to version 2 of the
41.Tn Intel
42performance measurement architecture.
43These CPUs contain 2 classes of PMCs:
44.Bl -tag -width "Li PMC_CLASS_UCP"
45.It Li PMC_CLASS_UCF
46Fixed-function counters that count only one hardware event per counter.
47.It Li PMC_CLASS_UCP
48Programmable counters that may be configured to count one of a defined
49set of hardware events.
50.El
51.Pp
52The number of PMCs available in each class and their widths need to be
53determined at run time by calling
54.Xr pmc_cpuinfo 3 .
55.Pp
56Intel Core i7 and Xeon 5500 PMCs are documented in
57.Rs
58.%B "Intel(R) 64 and IA-32 Architectures Software Developes Manual"
59.%T "Volume 3B: System Programming Guide, Part 2"
60.%N "Order Number: 253669-033US"
61.%D December 2009
62.%Q "Intel Corporation"
63.Re
64.Ss COREI7 AND XEON 5500 UNCORE FIXED FUNCTION PMCS
65These PMCs and their supported events are documented in
66.Xr pmc.ucf 3 .
67.Ss COREI7 AND XEON 5500 UNCORE PROGRAMMABLE PMCS
68The programmable PMCs support the following capabilities:
69.Bl -column "PMC_CAP_INTERRUPT" "Support"
70.It Em Capability Ta Em Support
71.It PMC_CAP_CASCADE Ta \&No
72.It PMC_CAP_EDGE Ta Yes
73.It PMC_CAP_INTERRUPT Ta \&No
74.It PMC_CAP_INVERT Ta Yes
75.It PMC_CAP_READ Ta Yes
76.It PMC_CAP_PRECISE Ta \&No
77.It PMC_CAP_SYSTEM Ta \&No
78.It PMC_CAP_TAGGING Ta \&No
79.It PMC_CAP_THRESHOLD Ta Yes
80.It PMC_CAP_USER Ta \&No
81.It PMC_CAP_WRITE Ta Yes
82.El
83.Ss Event Qualifiers
84Event specifiers for these PMCs support the following common
85qualifiers:
86.Bl -tag -width indent
87.It Li cmask= Ns Ar value
88Configure the PMC to increment only if the number of configured
89events measured in a cycle is greater than or equal to
90.Ar value .
91.It Li edge
92Configure the PMC to count the number of de-asserted to asserted
93transitions of the conditions expressed by the other qualifiers.
94If specified, the counter will increment only once whenever a
95condition becomes true, irrespective of the number of clocks during
96which the condition remains true.
97.It Li inv
98Invert the sense of comparison when the
99.Dq Li cmask
100qualifier is present, making the counter increment when the number of
101events per cycle is less than the value specified by the
102.Dq Li cmask
103qualifier.
104.El
105.Ss Event Specifiers (Programmable PMCs)
106Core i7 and Xeon 5500 uncore programmable PMCs support the following events:
107.Bl -tag -width indent
108.It Li GQ_CYCLES_FULL.READ_TRACKER
109.Pq Event 00H , Umask 01H
110Uncore cycles Global Queue read tracker is full.
111.It Li GQ_CYCLES_FULL.WRITE_TRACKER
112.Pq Event 00H , Umask 02H
113Uncore cycles Global Queue write tracker is full.
114.It Li GQ_CYCLES_FULL.PEER_PROBE_TRACKER
115.Pq Event 00H , Umask 04H
116Uncore cycles Global Queue peer probe tracker is full.
117The peer probe tracker queue tracks snoops from the IOH and remote sockets.
118.It Li GQ_CYCLES_NOT_EMPTY.READ_TRACKER
119.Pq Event 01H , Umask 01H
120Uncore cycles were Global Queue read tracker has at least one valid entry.
121.It Li GQ_CYCLES_NOT_EMPTY.WRITE_TRACKER
122.Pq Event 01H , Umask 02H
123Uncore cycles were Global Queue write tracker has at least one valid entry.
124.It Li GQ_CYCLES_NOT_EMPTY.PEER_PROBE_TRACKER
125.Pq Event 01H , Umask 04H
126Uncore cycles were Global Queue peer probe tracker has at least one valid entry.
127The peer probe tracker queue tracks IOH and remote socket snoops.
128.It Li GQ_ALLOC.READ_TRACKER
129.Pq Event 03H , Umask 01H
130Counts the number of tread tracker allocate to deallocate entries.
131The GQ read tracker allocate to deallocate occupancy count is divided
132by the count to obtain the average read tracker latency.
133.It Li GQ_ALLOC.RT_L3_MISS
134.Pq Event 03H , Umask 02H
135Counts the number GQ read tracker entries for which a full cache line read
136has missed the L3.
137The GQ read tracker L3 miss to fill occupancy count is divided by this count
138to obtain the average cache line read L3 miss latency.
139The latency represents the time after which the L3 has determined that the
140cache line has missed.
141The time between a GQ read tracker allocation and the L3 determining that the
142cache line has missed is the average L3 hit latency.
143The total L3 cache line read miss latency is the hit latency + L3 miss
144latency.
145.It Li GQ_ALLOC.RT_TO_L3_RESP
146.Pq Event 03H , Umask 04H
147Counts the number of GQ read tracker entries that are allocated in the read
148tracker queue that hit or miss the L3.
149The GQ read tracker L3 hit occupancy count is divided by this count to obtain
150the average L3 hit latency.
151.It Li GQ_ALLOC.RT_TO_RTID_ACQUIRED
152.Pq Event 03H , Umask 08H
153Counts the number of GQ read tracker entries that are allocated in the read
154tracker, have missed in the L3 and have not acquired a Request Transaction ID.
155The GQ read tracker L3 miss to RTID acquired occupancy count is
156divided by this count to obtain the average latency for a read L3 miss to
157acquire an RTID.
158.It Li GQ_ALLOC.WT_TO_RTID_ACQUIRED
159.Pq Event 03H , Umask 10H
160Counts the number of GQ write tracker entries that are allocated in the
161write tracker, have missed in the L3 and have not acquired a Request
162Transaction ID.
163The GQ write tracker L3 miss to RTID occupancy count is divided by this count
164to obtain the average latency for a write L3 miss to acquire an RTID.
165.It Li GQ_ALLOC.WRITE_TRACKER
166.Pq Event 03H , Umask 20H
167Counts the number of GQ write tracker entries that are allocated in the
168write tracker queue that miss the L3.
169The GQ write tracker occupancy count is divided by the this count to obtain the average L3 write miss latency.
170.It Li GQ_ALLOC.PEER_PROBE_TRACKER
171.Pq Event 03H , Umask 40H
172Counts the number of GQ peer probe tracker (snoop) entries that are
173allocated in the peer probe tracker queue that miss the L3.
174The GQ peer probe occupancy count is divided by this count to obtain the average L3 peer
175probe miss latency.
176.It Li GQ_DATA.FROM_QPI
177.Pq Event 04H , Umask 01H
178Cycles Global Queue Quickpath Interface input data port is busy importing
179data from the Quickpath Interface.
180Each cycle the input port can transfer 8 or 16 bytes of data.
181.It Li GQ_DATA.FROM_QMC
182.Pq Event 04H , Umask 02H
183Cycles Global Queue Quickpath Memory Interface input data port is busy
184importing data from the Quickpath Memory Interface.
185Each cycle the input port can transfer 8 or 16 bytes of data.
186.It Li GQ_DATA.FROM_L3
187.Pq Event 04H , Umask 04H
188Cycles GQ L3 input data port is busy importing data from the Last Level Cache.
189Each cycle the input port can transfer 32 bytes of data.
190.It Li GQ_DATA.FROM_CORES_02
191.Pq Event 04H , Umask 08H
192Cycles GQ Core 0 and 2 input data port is busy importing data from processor
193cores 0 and 2.
194Each cycle the input port can transfer 32 bytes of data.
195.It Li GQ_DATA.FROM_CORES_13
196.Pq Event 04H , Umask 10H
197Cycles GQ Core 1 and 3 input data port is busy importing data from processor
198cores 1 and 3.
199Each cycle the input port can transfer 32 bytes of data.
200.It Li GQ_DATA.TO_QPI_QMC
201.Pq Event 05H , Umask 01H
202Cycles GQ QPI and QMC output data port is busy sending data to the Quickpath
203Interface or Quickpath Memory Interface.
204Each cycle the output port can transfer 32 bytes of data.
205.It Li GQ_DATA.TO_L3
206.Pq Event 05H , Umask 02H
207Cycles GQ L3 output data port is busy sending data to the Last Level Cache.
208Each cycle the output port can transfer 32 bytes of data.
209.It Li GQ_DATA.TO_CORES
210.Pq Event 05H , Umask 04H
211Cycles GQ Core output data port is busy sending data to the Cores.
212Each cycle the output port can transfer 32 bytes of data.
213.It Li SNP_RESP_TO_LOCAL_HOME.I_STATE
214.Pq Event 06H , Umask 01H
215Number of snoop responses to the local home that L3 does not have the
216referenced cache line.
217.It Li SNP_RESP_TO_LOCAL_HOME.S_STATE
218.Pq Event 06H , Umask 02H
219Number of snoop responses to the local home that L3 has the referenced line
220cached in the S state.
221.It Li SNP_RESP_TO_LOCAL_HOME.FWD_S_STATE
222.Pq Event 06H , Umask 04H
223Number of responses to code or data read snoops to the local home that the
224L3 has the referenced cache line in the E state.
225The L3 cache line state is changed to the S state and the line is
226forwarded to the local home in the S state.
227.It Li SNP_RESP_TO_LOCAL_HOME.FWD_I_STATE
228.Pq Event 06H , Umask 08H
229Number of responses to read invalidate snoops to the local home that the L3
230has the referenced cache line in the M state.
231The L3 cache line state is invalidated and the line is forwarded to the
232local home in the M state.
233.It Li SNP_RESP_TO_LOCAL_HOME.CONFLICT
234.Pq Event 06H , Umask 10H
235Number of conflict snoop responses sent to the local home.
236.It Li SNP_RESP_TO_LOCAL_HOME.WB
237.Pq Event 06H , Umask 20H
238Number of responses to code or data read snoops to the local home that the
239L3 has the referenced line cached in the M state.
240.It Li SNP_RESP_TO_REMOTE_HOME.I_STATE
241.Pq Event 07H , Umask 01H
242Number of snoop responses to a remote home that L3 does not have the
243referenced cache line.
244.It Li SNP_RESP_TO_REMOTE_HOME.S_STATE
245.Pq Event 07H , Umask 02H
246Number of snoop responses to a remote home that L3 has the referenced line
247cached in the S state.
248.It Li SNP_RESP_TO_REMOTE_HOME.FWD_S_STATE
249.Pq Event 07H , Umask 04H
250Number of responses to code or data read snoops to a remote home that the L3
251has the referenced cache line in the E state.
252The L3 cache line state is changed to the S state and the line is forwarded to
253the remote home in the S state.
254.It Li SNP_RESP_TO_REMOTE_HOME.FWD_I_STATE
255.Pq Event 07H , Umask 08H
256Number of responses to read invalidate snoops to a remote home that the L3
257has the referenced cache line in the M state.
258The L3 cache line state is invalidated and the line is forwarded to the
259remote home in the M state.
260.It Li SNP_RESP_TO_REMOTE_HOME.CONFLICT
261.Pq Event 07H , Umask 10H
262Number of conflict snoop responses sent to the local home.
263.It Li SNP_RESP_TO_REMOTE_HOME.WB
264.Pq Event 07H , Umask 20H
265Number of responses to code or data read snoops to a remote home that the L3
266has the referenced line cached in the M state.
267.It Li SNP_RESP_TO_REMOTE_HOME.HITM
268.Pq Event 07H , Umask 24H
269Number of HITM snoop responses to a remote home
270.It Li L3_HITS.READ
271.Pq Event 08H , Umask 01H
272Number of code read, data read and RFO requests that hit in the L3
273.It Li L3_HITS.WRITE
274.Pq Event 08H , Umask 02H
275Number of writeback requests that hit in the L3.
276Writebacks from the cores will always result in L3 hits due to the inclusive property of the L3.
277.It Li L3_HITS.PROBE
278.Pq Event 08H , Umask 04H
279Number of snoops from IOH or remote sockets that hit in the L3.
280.It Li L3_HITS.ANY
281.Pq Event 08H , Umask 03H
282Number of reads and writes that hit the L3.
283.It Li L3_MISS.READ
284.Pq Event 09H , Umask 01H
285Number of code read, data read and RFO requests that miss the L3.
286.It Li L3_MISS.WRITE
287.Pq Event 09H , Umask 02H
288Number of writeback requests that miss the L3.
289Should always be zero as writebacks from the cores will always result in
290L3 hits due to the inclusive property of the L3.
291.It Li L3_MISS.PROBE
292.Pq Event 09H , Umask 04H
293Number of snoops from IOH or remote sockets that miss the L3.
294.It Li L3_MISS.ANY
295.Pq Event 09H , Umask 03H
296Number of reads and writes that miss the L3.
297.It Li L3_LINES_IN.M_STATE
298.Pq Event 0AH , Umask 01H
299Counts the number of L3 lines allocated in M state.
300The only time a cache line is allocated in the M state is when the line
301was forwarded in M state is forwarded due to a Snoop Read Invalidate Own request.
302.It Li L3_LINES_IN.E_STATE
303.Pq Event 0AH , Umask 02H
304Counts the number of L3 lines allocated in E state.
305.It Li L3_LINES_IN.S_STATE
306.Pq Event 0AH , Umask 04H
307Counts the number of L3 lines allocated in S state.
308.It Li L3_LINES_IN.F_STATE
309.Pq Event 0AH , Umask 08H
310Counts the number of L3 lines allocated in F state.
311.It Li L3_LINES_IN.ANY
312.Pq Event 0AH , Umask 0FH
313Counts the number of L3 lines allocated in any state.
314.It Li L3_LINES_OUT.M_STATE
315.Pq Event 0BH , Umask 01H
316Counts the number of L3 lines victimized that were in the M state.
317When the victim cache line is in M state, the line is written to its home cache agent
318which can be either local or remote.
319.It Li L3_LINES_OUT.E_STATE
320.Pq Event 0BH , Umask 02H
321Counts the number of L3 lines victimized that were in the E state.
322.It Li L3_LINES_OUT.S_STATE
323.Pq Event 0BH , Umask 04H
324Counts the number of L3 lines victimized that were in the S state.
325.It Li L3_LINES_OUT.I_STATE
326.Pq Event 0BH , Umask 08H
327Counts the number of L3 lines victimized that were in the I state.
328.It Li L3_LINES_OUT.F_STATE
329.Pq Event 0BH , Umask 10H
330Counts the number of L3 lines victimized that were in the F state.
331.It Li L3_LINES_OUT.ANY
332.Pq Event 0BH , Umask 1FH
333Counts the number of L3 lines victimized in any state.
334.It Li QHL_REQUESTS.IOH_READS
335.Pq Event 20H , Umask 01H
336Counts number of Quickpath Home Logic read requests from the IOH.
337.It Li QHL_REQUESTS.IOH_WRITES
338.Pq Event 20H , Umask 02H
339Counts number of Quickpath Home Logic write requests from the IOH.
340.It Li QHL_REQUESTS.REMOTE_READS
341.Pq Event 20H , Umask 04H
342Counts number of Quickpath Home Logic read requests from a remote socket.
343.It Li QHL_REQUESTS.REMOTE_WRITES
344.Pq Event 20H , Umask 08H
345Counts number of Quickpath Home Logic write requests from a remote socket.
346.It Li QHL_REQUESTS.LOCAL_READS
347.Pq Event 20H , Umask 10H
348Counts number of Quickpath Home Logic read requests from the local socket.
349.It Li QHL_REQUESTS.LOCAL_WRITES
350.Pq Event 20H , Umask 20H
351Counts number of Quickpath Home Logic write requests from the local socket.
352.It Li QHL_CYCLES_FULL.IOH
353.Pq Event 21H , Umask 01H
354Counts uclk cycles all entries in the Quickpath Home Logic IOH are full.
355.It Li QHL_CYCLES_FULL.REMOTE
356.Pq Event 21H , Umask 02H
357Counts uclk cycles all entries in the Quickpath Home Logic remote tracker
358are full.
359.It Li QHL_CYCLES_FULL.LOCAL
360.Pq Event 21H , Umask 04H
361Counts uclk cycles all entries in the Quickpath Home Logic local tracker are
362full.
363.It Li QHL_CYCLES_NOT_EMPTY.IOH
364.Pq Event 22H , Umask 01H
365Counts uclk cycles all entries in the Quickpath Home Logic IOH is busy.
366.It Li QHL_CYCLES_NOT_EMPTY.REMOTE
367.Pq Event 22H , Umask 02H
368Counts uclk cycles all entries in the Quickpath Home Logic remote tracker is
369busy.
370.It Li QHL_CYCLES_NOT_EMPTY.LOCAL
371.Pq Event 22H , Umask 04H
372Counts uclk cycles all entries in the Quickpath Home Logic local tracker is
373busy.
374.It Li QHL_OCCUPANCY.IOH
375.Pq Event 23H , Umask 01H
376QHL IOH tracker allocate to deallocate read occupancy.
377.It Li QHL_OCCUPANCY.REMOTE
378.Pq Event 23H , Umask 02H
379QHL remote tracker allocate to deallocate read occupancy.
380.It Li QHL_OCCUPANCY.LOCAL
381.Pq Event 23H , Umask 04H
382QHL local tracker allocate to deallocate read occupancy.
383.It Li QHL_ADDRESS_CONFLICTS.2WAY
384.Pq Event 24H , Umask 02H
385Counts number of QHL Active Address Table (AAT) entries that saw a max of 2
386conflicts.
387The AAT is a structure that tracks requests that are in conflict.
388The requests themselves are in the home tracker entries.
389The count is reported when an AAT entry deallocates.
390.It Li QHL_ADDRESS_CONFLICTS.3WAY
391.Pq Event 24H , Umask 04H
392Counts number of QHL Active Address Table (AAT) entries that saw a max of 3
393conflicts.
394The AAT is a structure that tracks requests that are in conflict.
395The requests themselves are in the home tracker entries.
396The count is reported when an AAT entry deallocates.
397.It Li QHL_CONFLICT_CYCLES.IOH
398.Pq Event 25H , Umask 01H
399Counts cycles the Quickpath Home Logic IOH Tracker contains two or more
400requests with an address conflict.
401A max of 3 requests can be in conflict.
402.It Li QHL_CONFLICT_CYCLES.REMOTE
403.Pq Event 25H , Umask 02H
404Counts cycles the Quickpath Home Logic Remote Tracker contains two or more
405requests with an address conflict.
406A max of 3 requests can be in conflict.
407.It Li QHL_CONFLICT_CYCLES.LOCAL
408.Pq Event 25H , Umask 04H
409Counts cycles the Quickpath Home Logic Local Tracker contains two or more
410requests with an address conflict.
411A max of 3 requests can be in conflict.
412.It Li QHL_TO_QMC_BYPASS
413.Pq Event 26H , Umask 01H
414Counts number or requests to the Quickpath Memory Controller that bypass the
415Quickpath Home Logic.
416All local accesses can be bypassed.
417For remote requests, only read requests can be bypassed.
418.It Li QMC_NORMAL_FULL.READ.CH0
419.Pq Event 27H , Umask 01H
420Uncore cycles all the entries in the DRAM channel 0 medium or low priority
421queue are occupied with read requests.
422.It Li QMC_NORMAL_FULL.READ.CH1
423.Pq Event 27H , Umask 02H
424Uncore cycles all the entries in the DRAM channel 1 medium or low priority
425queue are occupied with read requests.
426.It Li QMC_NORMAL_FULL.READ.CH2
427.Pq Event 27H , Umask 04H
428Uncore cycles all the entries in the DRAM channel 2 medium or low priority
429queue are occupied with read requests.
430.It Li QMC_NORMAL_FULL.WRITE.CH0
431.Pq Event 27H , Umask 08H
432Uncore cycles all the entries in the DRAM channel 0 medium or low priority
433queue are occupied with write requests.
434.It Li QMC_NORMAL_FULL.WRITE.CH1
435.Pq Event 27H , Umask 10H
436Counts cycles all the entries in the DRAM channel 1 medium or low priority
437queue are occupied with write requests.
438.It Li QMC_NORMAL_FULL.WRITE.CH2
439.Pq Event 27H , Umask 20H
440Uncore cycles all the entries in the DRAM channel 2 medium or low priority
441queue are occupied with write requests.
442.It Li QMC_ISOC_FULL.READ.CH0
443.Pq Event 28H , Umask 01H
444Counts cycles all the entries in the DRAM channel 0 high priority queue are
445occupied with isochronous read requests.
446.It Li QMC_ISOC_FULL.READ.CH1
447.Pq Event 28H , Umask 02H
448Counts cycles all the entries in the DRAM channel 1 high priority queue are
449occupied with isochronous read requests.
450.It Li QMC_ISOC_FULL.READ.CH2
451.Pq Event 28H , Umask 04H
452Counts cycles all the entries in the DRAM channel 2 high priority queue are
453occupied with isochronous read requests.
454.It Li QMC_ISOC_FULL.WRITE.CH0
455.Pq Event 28H , Umask 08H
456Counts cycles all the entries in the DRAM channel 0 high priority queue are
457occupied with isochronous write requests.
458.It Li QMC_ISOC_FULL.WRITE.CH1
459.Pq Event 28H , Umask 10H
460Counts cycles all the entries in the DRAM channel 1 high priority queue are
461occupied with isochronous write requests.
462.It Li QMC_ISOC_FULL.WRITE.CH2
463.Pq Event 28H , Umask 20H
464Counts cycles all the entries in the DRAM channel 2 high priority queue are
465occupied with isochronous write requests.
466.It Li QMC_BUSY.READ.CH0
467.Pq Event 29H , Umask 01H
468Counts cycles where Quickpath Memory Controller has at least 1 outstanding
469read request to DRAM channel 0.
470.It Li QMC_BUSY.READ.CH1
471.Pq Event 29H , Umask 02H
472Counts cycles where Quickpath Memory Controller has at least 1 outstanding
473read request to DRAM channel 1.
474.It Li QMC_BUSY.READ.CH2
475.Pq Event 29H , Umask 04H
476Counts cycles where Quickpath Memory Controller has at least 1 outstanding
477read request to DRAM channel 2.
478.It Li QMC_BUSY.WRITE.CH0
479.Pq Event 29H , Umask 08H
480Counts cycles where Quickpath Memory Controller has at least 1 outstanding
481write request to DRAM channel 0.
482.It Li QMC_BUSY.WRITE.CH1
483.Pq Event 29H , Umask 10H
484Counts cycles where Quickpath Memory Controller has at least 1 outstanding
485write request to DRAM channel 1.
486.It Li QMC_BUSY.WRITE.CH2
487.Pq Event 29H , Umask 20H
488Counts cycles where Quickpath Memory Controller has at least 1 outstanding
489write request to DRAM channel 2.
490.It Li QMC_OCCUPANCY.CH0
491.Pq Event 2AH , Umask 01H
492IMC channel 0 normal read request occupancy.
493.It Li QMC_OCCUPANCY.CH1
494.Pq Event 2AH , Umask 02H
495IMC channel 1 normal read request occupancy.
496.It Li QMC_OCCUPANCY.CH2
497.Pq Event 2AH , Umask 04H
498IMC channel 2 normal read request occupancy.
499.It Li QMC_ISSOC_OCCUPANCY.CH0
500.Pq Event 2BH , Umask 01H
501IMC channel 0 issoc read request occupancy.
502.It Li QMC_ISSOC_OCCUPANCY.CH1
503.Pq Event 2BH , Umask 02H
504IMC channel 1 issoc read request occupancy.
505.It Li QMC_ISSOC_OCCUPANCY.CH2
506.Pq Event 2BH , Umask 04H
507IMC channel 2 issoc read request occupancy.
508.It Li QMC_ISSOC_READS.ANY
509.Pq Event 2BH , Umask 07H
510IMC issoc read request occupancy.
511.It Li QMC_NORMAL_READS.CH0
512.Pq Event 2CH , Umask 01H
513Counts the number of Quickpath Memory Controller channel 0 medium and low
514priority read requests.
515The QMC channel 0 normal read occupancy divided by this count provides the
516average QMC channel 0 read latency.
517.It Li QMC_NORMAL_READS.CH1
518.Pq Event 2CH , Umask 02H
519Counts the number of Quickpath Memory Controller channel 1 medium and low
520priority read requests.
521The QMC channel 1 normal read occupancy divided by this count provides the
522average QMC channel 1 read latency.
523.It Li QMC_NORMAL_READS.CH2
524.Pq Event 2CH , Umask 04H
525Counts the number of Quickpath Memory Controller channel 2 medium and low
526priority read requests.
527The QMC channel 2 normal read occupancy divided by this count provides the
528average QMC channel 2 read latency.
529.It Li QMC_NORMAL_READS.ANY
530.Pq Event 2CH , Umask 07H
531Counts the number of Quickpath Memory Controller medium and low priority
532read requests.
533The QMC normal read occupancy divided by this count provides the average
534QMC read latency.
535.It Li QMC_HIGH_PRIORITY_READS.CH0
536.Pq Event 2DH , Umask 01H
537Counts the number of Quickpath Memory Controller channel 0 high priority
538isochronous read requests.
539.It Li QMC_HIGH_PRIORITY_READS.CH1
540.Pq Event 2DH , Umask 02H
541Counts the number of Quickpath Memory Controller channel 1 high priority
542isochronous read requests.
543.It Li QMC_HIGH_PRIORITY_READS.CH2
544.Pq Event 2DH , Umask 04H
545Counts the number of Quickpath Memory Controller channel 2 high priority
546isochronous read requests.
547.It Li QMC_HIGH_PRIORITY_READS.ANY
548.Pq Event 2DH , Umask 07H
549Counts the number of Quickpath Memory Controller high priority isochronous
550read requests.
551.It Li QMC_CRITICAL_PRIORITY_READS.CH0
552.Pq Event 2EH , Umask 01H
553Counts the number of Quickpath Memory Controller channel 0 critical priority
554isochronous read requests.
555.It Li QMC_CRITICAL_PRIORITY_READS.CH1
556.Pq Event 2EH , Umask 02H
557Counts the number of Quickpath Memory Controller channel 1 critical priority
558isochronous read requests.
559.It Li QMC_CRITICAL_PRIORITY_READS.CH2
560.Pq Event 2EH , Umask 04H
561Counts the number of Quickpath Memory Controller channel 2 critical priority
562isochronous read requests.
563.It Li QMC_CRITICAL_PRIORITY_READS.ANY
564.Pq Event 2EH , Umask 07H
565Counts the number of Quickpath Memory Controller critical priority
566isochronous read requests.
567.It Li QMC_WRITES.FULL.CH0
568.Pq Event 2FH , Umask 01H
569Counts number of full cache line writes to DRAM channel 0.
570.It Li QMC_WRITES.FULL.CH1
571.Pq Event 2FH , Umask 02H
572Counts number of full cache line writes to DRAM channel 1.
573.It Li QMC_WRITES.FULL.CH2
574.Pq Event 2FH , Umask 04H
575Counts number of full cache line writes to DRAM channel 2.
576.It Li QMC_WRITES.FULL.ANY
577.Pq Event 2FH , Umask 07H
578Counts number of full cache line writes to DRAM.
579.It Li QMC_WRITES.PARTIAL.CH0
580.Pq Event 2FH , Umask 08H
581Counts number of partial cache line writes to DRAM channel 0.
582.It Li QMC_WRITES.PARTIAL.CH1
583.Pq Event 2FH , Umask 10H
584Counts number of partial cache line writes to DRAM channel 1.
585.It Li QMC_WRITES.PARTIAL.CH2
586.Pq Event 2FH , Umask 20H
587Counts number of partial cache line writes to DRAM channel 2.
588.It Li QMC_WRITES.PARTIAL.ANY
589.Pq Event 2FH , Umask 38H
590Counts number of partial cache line writes to DRAM.
591.It Li QMC_CANCEL.CH0
592.Pq Event 30H , Umask 01H
593Counts number of DRAM channel 0 cancel requests.
594.It Li QMC_CANCEL.CH1
595.Pq Event 30H , Umask 02H
596Counts number of DRAM channel 1 cancel requests.
597.It Li QMC_CANCEL.CH2
598.Pq Event 30H , Umask 04H
599Counts number of DRAM channel 2 cancel requests.
600.It Li QMC_CANCEL.ANY
601.Pq Event 30H , Umask 07H
602Counts number of DRAM cancel requests.
603.It Li QMC_PRIORITY_UPDATES.CH0
604.Pq Event 31H , Umask 01H
605Counts number of DRAM channel 0 priority updates.
606A priority update occurs when an ISOC high or critical request
607is received by the QHL and there is a matching request with normal priority
608that has already been issued to the QMC.
609In this instance, the QHL will send a priority update to QMC to expedite the request.
610.It Li QMC_PRIORITY_UPDATES.CH1
611.Pq Event 31H , Umask 02H
612Counts number of DRAM channel 1 priority updates.
613A priority update occurs when an ISOC high or critical request is received by
614the QHL and there is a matching request with normal priority that has
615already been issued to the QMC.
616In this instance, the QHL will send a priority update to QMC to expedite the request.
617.It Li QMC_PRIORITY_UPDATES.CH2
618.Pq Event 31H , Umask 04H
619Counts number of DRAM channel 2 priority updates.
620A priority update occurs when an ISOC high or critical request is received by
621the QHL and there is a matching request with normal priority that has
622already been issued to the QMC.
623In this instance, the QHL will send a priority update to QMC to expedite the request.
624.It Li QMC_PRIORITY_UPDATES.ANY
625.Pq Event 31H , Umask 07H
626Counts number of DRAM priority updates.
627A priority update occurs when an ISOC high or critical request is received by
628the QHL and there is a matching request with normal priority that has
629already been issued to the QMC.
630In this instance, the QHL will send a priority update to QMC to expedite the request.
631.It Li QHL_FRC_ACK_CNFLTS.LOCAL
632.Pq Event 33H , Umask 04H
633Counts number of Force Acknowledge Conflict messages sent by the Quickpath
634Home Logic to the local home.
635.It Li QPI_TX_STALLED_SINGLE_FLIT.HOME.LINK_0
636.Pq Event 40H , Umask 01H
637Counts cycles the Quickpath outbound link 0 HOME virtual channel is stalled
638due to lack of a VNA and VN0 credit.
639Note that this event does not filter out when a flit would not have been
640selected for arbitration because another virtual channel is getting arbitrated.
641.It Li QPI_TX_STALLED_SINGLE_FLIT.SNOOP.LINK_0
642.Pq Event 40H , Umask 02H
643Counts cycles the Quickpath outbound link 0 SNOOP virtual channel is stalled
644due to lack of a VNA and VN0 credit.
645Note that this event does not filter out when a flit would not have been
646selected for arbitration because another virtual channel is getting arbitrated.
647.It Li QPI_TX_STALLED_SINGLE_FLIT.NDR.LINK_0
648.Pq Event 40H , Umask 04H
649Counts cycles the Quickpath outbound link 0 non-data response virtual
650channel is stalled due to lack of a VNA and VN0 credit.
651Note that this event does not filter out when a flit would not have been
652selected for arbitration because another virtual channel is getting arbitrated.
653.It Li QPI_TX_STALLED_SINGLE_FLIT.HOME.LINK_1
654.Pq Event 40H , Umask 08H
655Counts cycles the Quickpath outbound link 1 HOME virtual channel is stalled
656due to lack of a VNA and VN0 credit.
657Note that this event does not filter out when a flit would not have been
658selected for arbitration because another virtual channel is getting arbitrated.
659.It Li QPI_TX_STALLED_SINGLE_FLIT.SNOOP.LINK_1
660.Pq Event 40H , Umask 10H
661Counts cycles the Quickpath outbound link 1 SNOOP virtual channel is stalled
662due to lack of a VNA and VN0 credit.
663Note that this event does not filter out when a flit would not have been
664selected for arbitration because another virtual channel is getting arbitrated.
665.It Li QPI_TX_STALLED_SINGLE_FLIT.NDR.LINK_1
666.Pq Event 40H , Umask 20H
667Counts cycles the Quickpath outbound link 1 non-data response virtual
668channel is stalled due to lack of a VNA and VN0 credit.
669Note that this event does not filter out when a flit would not have been
670selected for arbitration because another virtual channel is getting arbitrated.
671.It Li QPI_TX_STALLED_SINGLE_FLIT.LINK_0
672.Pq Event 40H , Umask 07H
673Counts cycles the Quickpath outbound link 0 virtual channels are stalled due
674to lack of a VNA and VN0 credit.
675Note that this event does not filter out when a flit would not have been
676selected for arbitration because another virtual channel is getting arbitrated.
677.It Li QPI_TX_STALLED_SINGLE_FLIT.LINK_1
678.Pq Event 40H , Umask 38H
679Counts cycles the Quickpath outbound link 1 virtual channels are stalled due
680to lack of a VNA and VN0 credit.
681Note that this event does not filter out when a flit would not have been
682selected for arbitration because another virtual channel is getting arbitrated.
683.It Li QPI_TX_STALLED_MULTI_FLIT.DRS.LINK_0
684.Pq Event 41H , Umask 01H
685Counts cycles the Quickpath outbound link 0 Data ResponSe virtual channel is
686stalled due to lack of VNA and VN0 credits.
687Note that this event does not filter out when a flit would not have been
688selected for arbitration because another virtual channel is getting arbitrated.
689.It Li QPI_TX_STALLED_MULTI_FLIT.NCB.LINK_0
690.Pq Event 41H , Umask 02H
691Counts cycles the Quickpath outbound link 0 Non-Coherent Bypass virtual
692channel is stalled due to lack of VNA and VN0 credits.
693Note that this event does not filter out when a flit would not have been
694selected for arbitration because another virtual channel is getting arbitrated.
695.It Li QPI_TX_STALLED_MULTI_FLIT.NCS.LINK_0
696.Pq Event 41H , Umask 04H
697Counts cycles the Quickpath outbound link 0 Non-Coherent Standard virtual
698channel is stalled due to lack of VNA and VN0 credits.
699Note that this event does not filter out when a flit would not have been
700selected for arbitration because another virtual channel is getting arbitrated.
701.It Li QPI_TX_STALLED_MULTI_FLIT.DRS.LINK_1
702.Pq Event 41H , Umask 08H
703Counts cycles the Quickpath outbound link 1 Data ResponSe virtual channel is
704stalled due to lack of VNA and VN0 credits.
705Note that this event does not filter out when a flit would not have been
706selected for arbitration because another virtual channel is getting arbitrated.
707.It Li QPI_TX_STALLED_MULTI_FLIT.NCB.LINK_1
708.Pq Event 41H , Umask 10H
709Counts cycles the Quickpath outbound link 1 Non-Coherent Bypass virtual
710channel is stalled due to lack of VNA and VN0 credits.
711Note that this event does not filter out when a flit would not have been
712selected for arbitration because another virtual channel is getting arbitrated.
713.It Li QPI_TX_STALLED_MULTI_FLIT.NCS.LINK_1
714.Pq Event 41H , Umask 20H
715Counts cycles the Quickpath outbound link 1 Non-Coherent Standard virtual
716channel is stalled due to lack of VNA and VN0 credits.
717Note that this event does not filter out when a flit would not have been
718selected for arbitration because another virtual channel is getting arbitrated.
719.It Li QPI_TX_STALLED_MULTI_FLIT.LINK_0
720.Pq Event 41H , Umask 07H
721Counts cycles the Quickpath outbound link 0 virtual channels are stalled due
722to lack of VNA and VN0 credits.
723Note that this event does not filter out when a flit would not have been
724selected for arbitration because another virtual channel is getting arbitrated.
725.It Li QPI_TX_STALLED_MULTI_FLIT.LINK_1
726.Pq Event 41H , Umask 38H
727Counts cycles the Quickpath outbound link 1 virtual channels are stalled due
728to lack of VNA and VN0 credits.
729Note that this event does not filter out when a flit would not have been
730selected for arbitration because another virtual channel is getting arbitrated.
731.It Li QPI_TX_HEADER.BUSY.LINK_0
732.Pq Event 42H , Umask 02H
733Number of cycles that the header buffer in the Quickpath Interface outbound
734link 0 is busy.
735.It Li QPI_TX_HEADER.BUSY.LINK_1
736.Pq Event 42H , Umask 08H
737Number of cycles that the header buffer in the Quickpath Interface outbound
738link 1 is busy.
739.It Li QPI_RX_NO_PPT_CREDIT.STALLS.LINK_0
740.Pq Event 43H , Umask 01H
741Number of cycles that snoop packets incoming to the Quickpath Interface link
7420 are stalled and not sent to the GQ because the GQ Peer Probe Tracker (PPT)
743does not have any available entries.
744.It Li QPI_RX_NO_PPT_CREDIT.STALLS.LINK_1
745.Pq Event 43H , Umask 02H
746Number of cycles that snoop packets incoming to the Quickpath Interface link
7471 are stalled and not sent to the GQ because the GQ Peer Probe Tracker (PPT)
748does not have any available entries.
749.It Li DRAM_OPEN.CH0
750.Pq Event 60H , Umask 01H
751Counts number of DRAM Channel 0 open commands issued either for read or write.
752To read or write data, the referenced DRAM page must first be opened.
753.It Li DRAM_OPEN.CH1
754.Pq Event 60H , Umask 02H
755Counts number of DRAM Channel 1 open commands issued either for read or write.
756To read or write data, the referenced DRAM page must first be opened.
757.It Li DRAM_OPEN.CH2
758.Pq Event 60H , Umask 04H
759Counts number of DRAM Channel 2 open commands issued either for read or write.
760To read or write data, the referenced DRAM page must first be opened.
761.It Li DRAM_PAGE_CLOSE.CH0
762.Pq Event 61H , Umask 01H
763DRAM channel 0 command issued to CLOSE a page due to page idle timer
764expiration.
765Closing a page is done by issuing a precharge.
766.It Li DRAM_PAGE_CLOSE.CH1
767.Pq Event 61H , Umask 02H
768DRAM channel 1 command issued to CLOSE a page due to page idle timer
769expiration.
770Closing a page is done by issuing a precharge.
771.It Li DRAM_PAGE_CLOSE.CH2
772.Pq Event 61H , Umask 04H
773DRAM channel 2 command issued to CLOSE a page due to page idle timer
774expiration.
775Closing a page is done by issuing a precharge.
776.It Li DRAM_PAGE_MISS.CH0
777.Pq Event 62H , Umask 01H
778Counts the number of precharges (PRE) that were issued to DRAM channel 0
779because there was a page miss.
780A page miss refers to a situation in which a page is currently open and
781another page from the same bank needs to be opened.
782The new page experiences a page miss.
783Closing of the old page is done by issuing a precharge.
784.It Li DRAM_PAGE_MISS.CH1
785.Pq Event 62H , Umask 02H
786Counts the number of precharges (PRE) that were issued to DRAM channel 1
787because there was a page miss.
788A page miss refers to a situation in which a page is currently open and
789another page from the same bank needs to be opened.
790The new page experiences a page miss.
791Closing of the old page is done by issuing a precharge.
792.It Li DRAM_PAGE_MISS.CH2
793.Pq Event 62H , Umask 04H
794Counts the number of precharges (PRE) that were issued to DRAM channel 2
795because there was a page miss.
796A page miss refers to a situation in which a page is currently open and
797another page from the same bank needs to be opened.
798The new page experiences a page miss.
799Closing of the old page is done by issuing a precharge.
800.It Li DRAM_READ_CAS.CH0
801.Pq Event 63H , Umask 01H
802Counts the number of times a read CAS command was issued on DRAM channel 0.
803.It Li DRAM_READ_CAS.AUTOPRE_CH0
804.Pq Event 63H , Umask 02H
805Counts the number of times a read CAS command was issued on DRAM channel 0
806where the command issued used the auto-precharge (auto page close) mode.
807.It Li DRAM_READ_CAS.CH1
808.Pq Event 63H , Umask 04H
809Counts the number of times a read CAS command was issued on DRAM channel 1.
810.It Li DRAM_READ_CAS.AUTOPRE_CH1
811.Pq Event 63H , Umask 08H
812Counts the number of times a read CAS command was issued on DRAM channel 1
813where the command issued used the auto-precharge (auto page close) mode.
814.It Li DRAM_READ_CAS.CH2
815.Pq Event 63H , Umask 10H
816Counts the number of times a read CAS command was issued on DRAM channel 2.
817.It Li DRAM_READ_CAS.AUTOPRE_CH2
818.Pq Event 63H , Umask 20H
819Counts the number of times a read CAS command was issued on DRAM channel 2
820where the command issued used the auto-precharge (auto page close) mode.
821.It Li DRAM_WRITE_CAS.CH0
822.Pq Event 64H , Umask 01H
823Counts the number of times a write CAS command was issued on DRAM channel 0.
824.It Li DRAM_WRITE_CAS.AUTOPRE_CH0
825.Pq Event 64H , Umask 02H
826Counts the number of times a write CAS command was issued on DRAM channel 0
827where the command issued used the auto-precharge (auto page close) mode.
828.It Li DRAM_WRITE_CAS.CH1
829.Pq Event 64H , Umask 04H
830Counts the number of times a write CAS command was issued on DRAM channel 1.
831.It Li DRAM_WRITE_CAS.AUTOPRE_CH1
832.Pq Event 64H , Umask 08H
833Counts the number of times a write CAS command was issued on DRAM channel 1
834where the command issued used the auto-precharge (auto page close) mode.
835.It Li DRAM_WRITE_CAS.CH2
836.Pq Event 64H , Umask 10H
837Counts the number of times a write CAS command was issued on DRAM channel 2.
838.It Li DRAM_WRITE_CAS.AUTOPRE_CH2
839.Pq Event 64H , Umask 20H
840Counts the number of times a write CAS command was issued on DRAM channel 2
841where the command issued used the auto-precharge (auto page close) mode.
842.It Li DRAM_REFRESH.CH0
843.Pq Event 65H , Umask 01H
844Counts number of DRAM channel 0 refresh commands.
845DRAM loses data content over time.
846In order to keep correct data content, the data values have to be
847refreshed periodically.
848.It Li DRAM_REFRESH.CH1
849.Pq Event 65H , Umask 02H
850Counts number of DRAM channel 1 refresh commands.
851DRAM loses data content over time.
852In order to keep correct data content, the data values have to be
853refreshed periodically.
854.It Li DRAM_REFRESH.CH2
855.Pq Event 65H , Umask 04H
856Counts number of DRAM channel 2 refresh commands.
857DRAM loses data content over time.
858In order to keep correct data content, the data values have to be
859refreshed periodically.
860.It Li DRAM_PRE_ALL.CH0
861.Pq Event 66H , Umask 01H
862Counts number of DRAM Channel 0 precharge-all (PREALL) commands that close
863all open pages in a rank.
864PREALL is issued when the DRAM needs to be refreshed or needs to go
865into a power down mode.
866.It Li DRAM_PRE_ALL.CH1
867.Pq Event 66H , Umask 02H
868Counts number of DRAM Channel 1 precharge-all (PREALL) commands that close
869all open pages in a rank.
870PREALL is issued when the DRAM needs to be refreshed or needs to go
871into a power down mode.
872.It Li DRAM_PRE_ALL.CH2
873.Pq Event 66H , Umask 04H
874Counts number of DRAM Channel 2 precharge-all (PREALL) commands that close
875all open pages in a rank.
876PREALL is issued when the DRAM needs to be refreshed or needs to go
877into a power down mode.
878.El
879.Sh SEE ALSO
880.Xr pmc 3 ,
881.Xr pmc.amd 3 ,
882.Xr pmc.atom 3 ,
883.Xr pmc.core 3 ,
884.Xr pmc.corei7 3 ,
885.Xr pmc.iaf 3 ,
886.Xr pmc.soft 3 ,
887.Xr pmc.tsc 3 ,
888.Xr pmc.ucf 3 ,
889.Xr pmc.westmere 3 ,
890.Xr pmc.westmereuc 3 ,
891.Xr pmc_cpuinfo 3 ,
892.Xr pmclog 3 ,
893.Xr hwpmc 4
894.Sh HISTORY
895The
896.Nm pmc
897library first appeared in
898.Fx 6.0 .
899.Sh AUTHORS
900The
901.Lb libpmc
902library was written by
903.An Joseph Koshy Aq Mt jkoshy@FreeBSD.org .
904