.png){kind=icon}
.png){kind=icon}
Return
Paravirtualization effect on single- and multi-threaded memory-intensive linear algebra software
.png){kind=icon}
PRE
AI
DOI：10.1007/s10586-009-0080-4.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Previous studies have revealed that paravirtualization imposes minimal per;
mance overhead on High Per;
mance Computing (HPC) workloads, while exposing numerous benefits;
this field. In this study, we are investigating the impact of paravirtualization on the per;
mance of automatically-tuned software systems. We compare peak per;
mance, per;
mance degradation in constrained memory situations, per;
mance degradation in multi-threaded applications, and inter-VM shared memory per;
mance. For comparison purposes, we examine the proficiency of ATLAS, a quintessential example of an autotuning software system, in tuning the BLAS library routines;
paravirtualized systems. Our results show that the combination of ATLAS and Xen paravirtualization delivers native execution per;
mance and nearly identical memory hierarchy per;
mance profiles in both single and multi-threaded scenarios. Furthermore, we show that it is possible to achieve memory sharing among OS instances at native speeds. These results expose new benefits to memory-intensive applications arising from the ability to slim down the guest OS without influencing the system per;
mance. In addition, our findings support a novel and very attractive deployment scenario;
computational science and engineering codes on virtual clusters and computational clouds.
Keywords:
Virtual machine monitors
Paravirtualization
AutoTuning
BLAS
High per
mance
Linear algebra
Cloud computing
Journal
C
IF:
4.1
Papers:
5.0K
Citations:
7.5K
