Hardware Reference
In-Depth Information
FIGURE 3.35
The speedup from using multithreading on one core on an i7 processor
averages 1.28 for the Java benchmarks and 1.31 for the PARSEC benchmarks (using an
unweighted harmonic mean, which implies a workload where the total time spent ex-
ecuting each benchmark in the single-threaded base set was the same)
. The energy effi-
ciency averages 0.99 and 1.07, respectively (using the harmonic mean). Recall that anything
above 1.0 for energy efficiency indicates that the feature reduces execution time by more than
it increases average power. Two of the Java benchmarks experience little speedup and have
significant negative energy efficiency because of this. Turbo Boost is off in all cases. These
data were collected and analyzed by
Esmaeilzadeh et al. [2011]
using the Oracle (Sun) HotS-
pot build 16.3-b01 Java 1.6.0 Virtual Machine and the gcc v4.4.1 native compiler.
The harmonic mean of the speedup for the Java benchmarks is 1.28, despite the two bench-
marks that see small gains. These two benchmarks, pjbb2005 and tradebeans, while multi-
threaded, have limited parallelism. They are included because they are typical of a multith-
readed benchmark that might be run on an SMT processor with the hope of extracting some
performance, which they find in limited amounts. The PARSEC benchmarks obtain somewhat
better speedups than the full set of Java benchmarks (harmonic mean of 1.31). If tradebeans
and pjbb2005 were omited, the Java workload would actually have signiicantly better spee-
dup (1.39) than the PARSEC benchmarks. (See the discussion of the implication of using har-
monic mean to summarize the results in the caption of
Figure 3.36
.
)
Search WWH ::
Custom Search