05.08.2014 Views

here - Stefan-Marr.de

here - Stefan-Marr.de

here - Stefan-Marr.de

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

8.4. Ad hoc vs. OMOP Performance<br />

100.00<br />

31.62<br />

10.00<br />

3.16<br />

1.00<br />

0.32<br />

Array Access<br />

Class Var Binding<br />

FloatLoop<br />

Instance Var.<br />

Int Loop<br />

Sends<br />

Local Sends<br />

with 10 arguments<br />

Local Sends<br />

Remote Sends<br />

Remote Sends<br />

with 10 arguments<br />

Runtime, normalized to corresponding Ad Hoc<br />

implementation, lower is better<br />

AmbientTalkST<br />

100.00<br />

31.62<br />

10.00<br />

3.16<br />

1.00<br />

0.32<br />

LRSTM<br />

Array Access<br />

Class Var Binding<br />

FloatLoop<br />

Instance Var.<br />

Int Loop<br />

Sends<br />

Sends<br />

with 10 arguments<br />

AST-OMOP on CogVM<br />

RoarVM+OMOP (opt)<br />

AST-OMOP on CogVM<br />

RoarVM+OMOP (opt)<br />

Figure 8.4.: Ad hoc vs. OMOP Microbenchmarks: Runtime normalized to Ad hoc<br />

implementations, logarithmic scale. The AST-OMOP implementation on top of<br />

the CogVM shows significant slowdowns especially for the AmbientTalkST microbenchmarks<br />

because more operations are reified. However, some benchmarks<br />

show speedups. The overhead for the VM-based implementation is generally<br />

lower and outperforms the ad hoc implementation of the STM on a number of<br />

benchmarks.<br />

217

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!