Hi, Nilay:
Yes, I have read ruby.stats. However, I can not judge wether there are related
information used to characterize the inter-core communication. For example,
part of the contents in ruby.stats are as follows:
--- L1Cache ---
- Event Counts -
Load [11995748 11664684 11137641 11468922 23715557 15161293 28839034 12063893
11480052 11476126 11151023 11471308 11477231 11476754 11151479 11472758 ]
217203503
Ifetch [62188323 60547127 57922309 59499768 133898957 78474900 155365090
62731881 59516853 59531683 57924963 59498146 59515900 59533440 57936121
59515400 ] 1143600861
Store [7944011 7690131 7302353 7555219 13244299 9598066 16295580 7962536
7555451 7556568 7298051 7554709 7558383 7554909 7299073 7555827 ] 137525166
L2_Replacement [0 0 0 0 425405 0 149 0 0 0 0 0 0 0 0 0 ] 425554
L1_to_L2 [248684 198146 174629 138323 3935842 916756 2118553 291181 149824
150348 157394 138352 151663 167045 158001 143512 ] 9238253
Trigger_L2_to_L1D [112472 96552 102309 66353 2425022 492711 587828 120932 77870
78196 85906 66436 79573 95243 86332 71660 ] 4645395
Trigger_L2_to_L1I [111656 85357 70368 70339 824136 409948 1459423 148820 70278
70489 69824 70197 70379 70152 70030 70217 ] 3741613
Complete_L2_to_L1 [224128 181909 172677 136692 3249158 902659 2047251 269752
148148 148685 155730 136633 149952 165395 156362 141877 ] 8387008
Other_GETX [526769 535166 539387 539568 364245 448657 452296 523286 538926
539097 539042 538807 538786 538831 538841 538847 ] 8240551
Other_GETS [1153971 1158570 1166776 1167010 590673 1098579 1065804 1137674
1166772 1167060 1167199 1166882 1167374 1167156 1167075 1166975 ] 17875550
Merged_GETS [1390 613 1146 468 817 1298 872 2622 455 532 582 789 545 462 531
465 ] 13587
Other_GETS_No_Mig [3462 3428 2998 3173 8965 7758 8808 8899 3225 3118 2927 3452
2975 3175 3169 3251 ] 72783
NC_DMA_GETS [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ] 0
Invalidate [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ] 0
Ack [684101 492358 319673 313533 11465886 2490321 2923350 879478 324263 317918
323494 333714 322568 322617 324491 321868 ] 22159633
Shared_Ack [131473 136995 139992 137706 111404 138026 140189 136757 138497
139962 135883 126071 137383 136812 136014 138770 ] 2161934
Data [3019 3002 2762 2897 12937 15390 10410 4290 3030 2992 3059 2970 2851 2976
2914 2964 ] 78463
Shared_Data [35682 36241 28466 28563 38597 53072 51944 33445 29511 29402 28952
28007 28826 29455 29282 29737 ] 539182
Exclusive_Data [20839 9190 4134 4160 727334 120102 153548 34897 4534 4282 4663
5513 4731 4446 4765 4131 ] 1111269
Writeback_Ack [0 0 0 0 425351 0 90 0 0 0 0 0 0 0 0 0 ] 425441
Writeback_Nack [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ] 0
All_acks [35686 36244 28468 28565 38600 53074 51947 33446 29513 29404 28954
28009 28828 29457 29284 29739 ] 539218
All_acks_no_sharers [25045 13351 8076 8254 743611 137677 166871 40847 8694 8445
8880 9643 8858 8561 8822 8202 ] 1213837
Flush_line [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ] 0
Block_Ack [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ] 0
Are the "Shared_Ack" and "Share_Data" reflecting such situation? Thank you very
much!
Song Jin
From: Nilay Vaish
Date: 2012-08-16 23:12
To: jinsah1977; gem5 users mailing list
Subject: Re: [gem5-users] Inter-core communication characterization for Parsec
in Alpha FS mode with ruby enabled
On Thu, 16 Aug 2012, jinsong wrote:
> Hi All, I am running parsec in Alpha FS mode (16-core) with ruby
> enabled. I want to retrieve the data for inter-core communication. For
> example, as Bienia mentioned in PACT'08, shared write can be a symptom
> to indicate the inter-core communication quantity. So does gem5 give
> such information or other information to characterize the inter-core
> communication? Any help really appreciated!
>
Did you try looking at the file ruby.stats?
--
Nilay
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
