projects / profile / ivi / kernel-x86-ivi.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 28935ab) | patch
bcache: Delete some slower inline asm
authorKent Overstreet <kmo@daterainc.com>
Thu, 22 Aug 2013 01:36:45 +0000 (18:36 -0700)
committerKent Overstreet <kmo@daterainc.com>
Mon, 11 Nov 2013 05:56:42 +0000 (21:56 -0800)
commit098fb25498214069e6bbf908515f2952dd7654d0
treef170596e7abb95f78a6b89044945aaea81919214tree | snapshot
parent28935ab5163c49ca5c199d67335e5e3c72c50853commit | diff
bcache: Delete some slower inline asm

Never saw a profile of bset_search_tree() where it wasn't bottlenecked
on memory until I got my new Haswell machine, but when I tried it there
it was suddenly burning 20% of the cpu in the inner loop on shrd...

Turns out, the version of shrd that takes 64 bit operands has a 9 cycle
latency. hah.

Signed-off-by: Kent Overstreet <kmo@daterainc.com>
drivers/md/bcache/bset.c diff | blob | history
Domain: System / Uncategorized;