m5-dev => gem5-dev, and regardless you should use gem5-users (like you are).

Gabe

On 07/30/11 04:54, Korey Sewell wrote:
> Unfortunately, it seems that since you asked for a patch and no one
> responded, then no one has the patch for what you're looking for.
> You'll have to dig into the code and add those stats yourself. Similar
> for the bug-fixing you describe later.
>
> If you do get into that, then m5-dev can help you debug as long as you
> provide details of what you are doing along the way (command lines,
> what you tried, what you theorize the problem to be, etc.)
>
> On Fri, Jul 29, 2011 at 4:15 AM, Paolo Grani <[email protected]
> <mailto:[email protected]>> wrote:
>
>     No one....
>
>     Paolo
>
>
>     Il 26/07/2011 11:58, Paolo Grani ha scritto:
>
>         Hi all,
>
>         i need some clarification about the ruby statistics.
>         I read in a previous post that the miss counts for the ruby
>         cache only
>         works with the MOESI_hammer coherence protocol.
>         I would like to know if there is some patch to get the
>         MOESI_CMP_directory protocol working with this information.
>         In fact when a try to run a simulation whit this protocol i
>         get this
>         output for any CacheMemory in Ruby:
>
>          "Cache Stats: system.l1_cntrl0.L1DcacheMemory
>          system.l1_cntrl0.L1DcacheMemory_total_misses: 0
>          system.l1_cntrl0.L1DcacheMemory_total_demand_misses: 0
>          system.l1_cntrl0.L1DcacheMemory_total_prefetches: 0
>          system.l1_cntrl0.L1DcacheMemory_total_sw_prefetches: 0
>          system.l1_cntrl0.L1DcacheMemory_total_hw_prefetches: 0"
>
>
>         In addition i'm also interested in latency informations on
>         memory accesses.
>         Running a simulation always with MOESI_CMP_directory i get this:
>
>         "miss_latency: [binsize: 8 max: 825 count: 19954053 average:
>         3.10522 |
>         standard deviation: 3.91114 | 19923362 3729 6244 6396 3738
>         1081 119 79
>         121 88 87 31 31 27 34 30 36 31 35 371 2435 2496 957 663 874
>         529 110 29
>         27 19 21 17 13 19 16 12 15 16 11 7 9 4 5 5 7 3 9 7 1 5 2 3 0 4
>         2 3 3 0 1
>         1 1 1 3 1 1 1 0 1 0 0 0 1 2 0 1 0 2 1 2 0 1 1 1 0 0 0 0 0 0 0
>         0 1 0 0 0
>         0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
>         0 0 0 0 0
>         0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
>         0 0 0 0 0
>         0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
>         0 0 ]"
>
>
>         For what i can understand this information says  "how many
>         requests took
>         1 cycle, 2 cycles, 3 cycles... to be accomplished" also
>         reporting some
>         aggregate information (count,average....).
>         But the the coherence protocol models a 2-level cache
>         architecture so i
>         would like to know if this informations are for the L2 shared
>         cache or
>         for the L1 level.
>         If the answer is L1 level this information is an aggregate of
>         all the
>         private L1 caches (i'm running a 8 cores simulation so  there
>         are 8
>         private L1 caches) or is specific for some L1 cache?
>
>
>         Finally i found a patch that tries to convert the Ruby's stats
>         to the
>         M5-style Stat class.
>         I can correctly compile the code  but when i try to run it a
>         memory leak
>         occurs.
>         Someboy have tried to fix it?
>
>
>         Thank's in advance,
>
>         Paolo
>
>
>
>     -- 
>     ====================================================
>     The devil is in the details
>     ====================================================
>
>     Ing. Paolo Grani
>     Dipartimento di Ingegneria dell'Informazione
>     via Roma 56, 53100 Siena, Italia
>     Laboratorio di Architettura dei Calcolatori ---> int. 223
>
>     mailto:[email protected] <mailto:[email protected]>
>     http://www.dii.unisi.it
>     ====================================================
>     _______________________________________________
>     gem5-users mailing list
>     [email protected] <mailto:[email protected]>
>     http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
>
>
>
>
> -- 
> - Korey
>
>
> _______________________________________________
> gem5-users mailing list
> [email protected]
> http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to