ProfileMe: hardware support for instruction-level profiling on out-of-order processors
MICRO 30 Proceedings of the 30th annual ACM/IEEE international symposium on Microarchitecture
Using hardware performance monitors to isolate memory bottlenecks
Proceedings of the 2000 ACM/IEEE conference on Supercomputing
A scalable cross-platform infrastructure for application performance tuning using hardware counters
Proceedings of the 2000 ACM/IEEE conference on Supercomputing
A Portable Programming Interface for Performance Evaluation on Modern Processors
International Journal of High Performance Computing Applications
Modeling application performance by convolving machine signatures with application profiles
WWC '01 Proceedings of the Workload Characterization, 2001. WWC-4. 2001 IEEE International Workshop
Producing wrong data without doing anything obviously wrong!
Proceedings of the 14th international conference on Architectural support for programming languages and operating systems
Evaluating Sampling Based Hotspot Detection
ARCS '09 Proceedings of the 22nd International Conference on Architecture of Computing Systems
Detailed performance analysis using coarse grain sampling
Euro-Par'09 Proceedings of the 2009 international conference on Parallel processing
Rapid identification of architectural bottlenecks via precise event counting
Proceedings of the 38th annual international symposium on Computer architecture
Hi-index | 0.00 |
Performance monitoring hardware is available on most modern microprocessors in the form of hardware counters and other registers that record data about processor events. This hardware may be used in counting mode, in which aggregate events counts are accumulated, and/or in sampling mode, in which time-based or event-based sampling is used to collect profiling data. This paper discusses uses of these two modes and considers the issues of efficiency and accuracy raised by each. Implications for the PAPI cross-platform hardware counter interface are also discussed.