Getting Dynamic Power Numbers of Cache as Zero

23 views

Skip to first unread message

kira...@umn.edu

unread,

Jul 23, 2016, 12:13:43 AM7/23/16

to ESESC

Forgive me if this is a silly question but I just started using ESESC and ran some sample simulations. For the Power numbers of the cache I'm getting dynamic power as 0 for most cache levels. My output is shown below:

********************************************************************************************************

Cache Occ AvgMemLat MemAccesses MissRate ( RD , WR, BUS) Dyn_Pow (mW) Lkg_Pow (mW)

IL1(0) 0.0 2.1 3853085 0.03% (100.0%, 0.0%, 0.0%) 59 0

ITLB(0) 0.0 2.1 3852809 0.01% (100.0%, 0.0%, 0.0%) 0 0

--------------------------------------------------------------------------------------------------

DL1(0) 0.0 10.7 1600359 4.31% ( 97.4%, 92.0%, 0.0%) 59 0

--------------------------------------------------------------------------------------------------

L2(0) 0.0 86.2 73856 88%.52 ( 10.3%, 0.0%, 0.0%) 0 77

L3(0) 0.0 67.8 71927 69.52% ( 30.5%, 0.0%, 0.0%) 0 355

MemBus(0) 0.0 107.5 21650 0.00% (100.0%, 0.0%, 0.0%) 0 8

PTLB(0) 0.0 11.1 1599881 6.34% ( 93.7%, 0.0%, 0.0%) 0 0

********************************************************************************************************

I tried digging deeper into the code to see the interaction between the various counters and McPat but it's gonna take me some time to fully grasp everything.

I noticed even in their presentation at ISCA, their sample report file reported Dynamic power as zero (as seen over here: https://masc.soe.ucsc.edu/esesc/resources/8-thermal.pdf). Can somebody please shed some insight as to what's causing this? Surely, the dynamic power of a cache with several thousand access can't be zero.