论文信息 - A Memory Profiling Tool for Performance Tuning of OLTP Workloads

A Memory Profiling Tool for Performance Tuning of OLTP Workloads

The performance of CPU has been improved more than 50% per year but the performance improvement of mem- ory in latency is much lower than those of CPU. Therefore the penalty of memory access has been increased every year. Recent works show optimization on scientific workloads is not effective on commercial workloads like DBMSs. We have developed a memory profiling tool to collect performance information for OLTP and evaluated the effectiveness of the tool.

Hirotaka Yoshioka

[1] Chris Sears. The Elements of Cache Programming Style , 2000, Annual Linux Showcase & Conference.

[2] Brinkley Sprunt,et al. Pentium 4 Performance-Monitoring Features , 2002, IEEE Micro.

[3] David J. Sager,et al. The microarchitecture of the Pentium 4 processor , 2001 .

[4] Jeffrey Dean,et al. ProfileMe: hardware support for instruction-level profiling on out-of-order processors , 1997, Proceedings of 30th Annual International Symposium on Microarchitecture.

[5] David A. Patterson,et al. Computer Architecture: A Quantitative Approach , 1969 .

[6] Dionisios N. Pnevmatikatos,et al. Cache performance of the SPEC92 benchmark suite , 1993, IEEE Micro.

[7] David J. DeWitt,et al. DBMSs on a Modern Processor: Where Does Time Go? , 1999, VLDB.