IEEE Spectrum November, 2013 - 13

AmD

MultiCoRE MEMoRy: More processor cores-
this AMD Opteron has six-means the computer has
a harder time managing how memory moves into and
out of the processor's cache.

ory banks (which can stretch their limits).
Moreover, CPUs have no higher-level knowledge of the kinds of jobs they're doing. This
means a self-contained numerical simulation with complex equations but little need
for memory access would run with exactly
the same cache resources as would a graph
search, a memory-hogging hunt for relationships between stored data.
So Sanchez and his graduate student
Nathan Beckmann thought, Why not let
the OS trim the cache size for pure computation and swell its ranks for graph search?

in with the OS, that the researchers call
Peekahead. This module was adapted from
the Lookahead Cache, developed more than
a decade ago by Beijing computer scientists.
Peekahead computes the best configuration of CPU caches based on the upcoming
jobs it expects the cores to do in the coming clock cycles.
"When you let software be in charge, you
have to be careful of your overhead," Sanchez
says. A poorly designed cache management
system, he says, might trim the cache to its
optimum size and do it again every fraction
of a second. But doing so taxes the CPU. And
what's the point of a CPU efficiency algorithm
that requires extraordinary amounts of CPU
time? "The exact solution is really expensive.
So we have to come up with a quick way of
getting the job done so that the overhead
doesn't negate the gains you get," he says.
Linley Gwennap of the Linley Group, a
semiconductor consulting firm based in
Mountain View, Calif., says he's impressed
with Jigsaw but cautions that it's not quite
ready for chip-fab prime time. "The problem
is generally that a scheme that's effective
on one processor may not be effective on
another processor with a different hardware
design," he says. "Every time the processor
changes, you have to redo your software,
which customers generally don't like."
Sanchez counters that software applications and utilities would remain unaffected
by Jigsaw. "Only the operating system code
needs to be aware of that intimate knowlThe first step, they say, would be to give
perhaps 1 percent of the CPU's footprint to
edge of the hardware, like the topology of
a simple piece of hardware that could moni- the different portions of the cache," he says.
tor in real time the cache activity in each
Jason Mars, an assistant professor of comcore. Hardware cache monitors would give
puter science at the University of Michigan,
Jigsaw the independent oversight it would
says Jigsaw works well as a proof of concept,
need to play air traffic controller with the which he says chipmakers might adapt as
CPU's caches.
they see fit.
Second, Sanchez and Beckmann say, the
"The crisp novelty in this work has to do
OS's kernel needs at most a few thousand with the codesign between hardware and
more lines of code. That's not much of an
software," Mars says. "Much of the prior
addition, considering that Linux's kernel work was biased in one direction. More was
in 2012 weighed in with 15 million lines and
expected to be done in hardware, and there
Apple's and Microsoft's kernels unofficially was a little bit less flexibility. Jigsaw really...
contained tens of millions more than that.
builds a holistic system that spans both the
One of Jigsaw's more prominent fea- hardware and the software."
tures is a software module, to be folded -m a r k a nder son

nEwS

SPECTRUM.IEEE.ORG

nORTh aMERICan

nOV 2013

http://SPECTRUM.IEEE.ORG

Table of Contents for the Digital Edition of IEEE Spectrum November, 2013