Incremental Monitoring on Programmable Network Interface Cards
نویسندگان
چکیده
Monitoring of program behavior is a basic necessity for performance tuning and debugging. A major problem is the overhead associated with the monitoring tasks, which influences the program’s behavior and affects the observation results. At the same time, a certain amount of data is needed for the program analysis activities. For this reason, monitoring approaches have to balance the amount of overhead with the amount of extracted information. The proposed incremental monitoring approach provides a solution implemented to run on programmable network interface cards (Myrinet), which perturbs the program’s execution as little as possible1. The resulting data is then used to control the program during subsequent executions, while at the same time incrementally increasing the amount of monitoring. With this multistep approach, any amount of trace data can be extracted without substantial intrusions onto the program’s execution.
منابع مشابه
Simulation - Driven Design of High - Performance Programmable Network Interface Cards by Paul Willmann A
Simulation-Driven Design of High-Performance Programmable Network Interface Cards by Paul Willmann As network link speeds race to 10 Gigabit/sec and beyond, Internet servers will rely on programmable network interface cards (NICs) to relieve the ever increasing frame processing burdens. To meet that need, this work introduces a scalable, programmable NIC architecture that saturates a full-duple...
متن کاملFast NIC-Based Barrier over Myrinet/GM
An efficient barrier implementation is desirable on parallel systems to obtain good parallel speedup and to support finer-grained computation. Some modern Network Interface Cards (NICs) have programmable processors which can be used to provide support for collective communications such as barrier. In this paper, we utilize such a programmable NIC to provide an efficient barrier synchronization ...
متن کاملOffloading MPI Parallel Prefix Scan (MPI_Scan) with the NetFPGA
Parallel programs written using the standard Message Passing Interface (MPI) frequently depend upon the ability to efficiently execute collective operations. MPI_Scan is a collective operation defined in MPI that implements parallel prefix scan which is very useful primitive operation in several parallel applications. This operation can be very time consuming. In this paper, we explore the use ...
متن کاملMonitoring tools of COMPASS experiment at CERN
This paper briefly introduces the data acquisition system of the COMPASS experiment and is mainly focused on the part that is responsible for the monitoring of the nodes in the whole newly developed data acquisition system of this experiment. The COMPASS is a high energy particle experiment with a fixed target located at the SPS of the CERN laboratory in Geneva, Switzerland. The hardware of the...
متن کاملAchieving Gigabit Performance on Programmable Ethernet Network Interface Cards
The shift from Fast Ethernet (100Mbit/s) to Gigabit Ethernet (1000Mbit/s) did not result in the expected tenfold increase in bandwidth at the application level. In this dissertation we make use of programmable Ethernet network cards running at gigabit speeds to identify present bottlenecks and also propose two new techniques aimed at boosting application performance to near-gigabit levels whils...
متن کامل