|
From: Jiayuan M. <men...@gm...> - 2012-01-15 01:35:37
|
Dear all,
I'm using callgrind to analyze the cache behavior at assembly level. After
generating outputs from callgrind_annotate --auto=yes, I observe lines like:
--------------------------------------------------------------------------------
Ir Dr Dw I1mr D1mr D1mw ILmr DLmr DLmw
404,381 404,381 . . . . . . . lfd 7,.LC6@l
(9)
In this example, the instruction reads the cache for 404381 times, but
D1mr, D1mw are all dots, does it mean there is no misses for this
instruction at all?
Thanks,
Jiayuan
|