Peter R. Sebastian
Abstract
HEMI is an experimental instrumentation system being developed for use with the CYBER 70 and 170 Series computers in order to ascertain the extent to which an integrated approach to instrumentation is economically and technologically viable for performance measurement and evaluation purposes.
HEMI takes advantage of the distributed CYBER computer architecture. This consists of a pool of Peripheral Processors (PPs) - (mainly dedicated to I/O and system tasks) while the CPU capabilities are reserved mostly for computation; Central Memory constitutes the communications link.
HEMI uses one of the PPs as its major processor. A hardware data acquisition front end is interfaced to one of the I/O channels and driven by the PP. Hardware probes sample events at suitable testpoints, while the PP has software access to Central Memory (Operating System tables and parameters), Status Registers, I/O Channel Flags, etc.
A data reduction package is used to produce a variety of reports from the data collected. A limited on-line data reduction and display capability is also provided.
This paper will describe the current status of the project as well as anticipated applications of HEMI.
- 1 R.A. Aschenbrenner, L. Amiot and N.K. Natarajan, "The Neurotron Monitor System", Proc. AFIPS FJCC, Vol. 39, 1971; pp 31-37.Google Scholar
- 2 T.E. Bell, "Computer Performance Analysis; Measurement Objectives and Tools", Report prepared for N.A.S.A. and U.S.A.F. Project RAND, R-584-NASA/PR, Feb. 71.Google Scholar
- 3 T.E. Bell, "Computer Performance Analysis: Minicomputer Based Hardware Monitoring", Report prepared to U.S.A.F. Project RAND, R-696-PR, Jun. 72.Google Scholar
- 4 D.J. Campbell and W.J. Heffner, "Measurement and Analysis of Large Operating Systems During System Development", Proc. AFIPS FJCC, Vol. 33, 1968; pp 903-914.Google Scholar
- 5 H.N. Cantrell and A.L. Ellison, "Multiprogramming System Performance Measurement and Analysis", Proc. AFIPS SJCC, Vol. 32, 1968; pp 213 - 221.Google Scholar
- 6 Dr. G. Carlson, "A User's View of Hardware Performance Monitors or How to Get More Computer for Your Dollar", Proc. IFIP Congress, 1971, Booklet TA-5; pp 128-132.Google Scholar
- 7 "VS2 Release 2: Built in Monitor and Multiple V.S.?" Computerworld, 20 Mar. 74.Google Scholar
- 8 M.E. Drummond Jr., "Evaluation and Measurement Techniques for Digital Computer Systems", Prentice Hall, 1973.Google Scholar
- 9 S.H. Fuller, R.J. Swan and W.A. Wulf, "The Instrumentation of C. mmp, a Multi-(Mini) Processor", Proc. IEEE, COMPCON 73; pp 173-176.Google Scholar
- 10 J. Hughes and D. Cronshaw, "On Using a Hardware Monitor as an Intelligent Peripheral", ACM (SIGME) Performance Evaluation Review, Vol. 2, No. 4, Dec. 73; pp 3-19. Google ScholarDigital Library
- 11 A.G. Nemeth and P.D. Rovner, "User Program Measurement in a Time-Shared Environment", CACM, Oct. 71, Vol. 14, No. 10; pp 661-666. Google ScholarDigital Library
- 12 J.D. Noe, "Acquiring and Using a Hardware Monitor", Datamation, Apr. 74; pp 89-95.Google Scholar
- 13 J.E. Shemer and J.B. Robertson, "Instrumentation of Time-Shared Systems", IEEE Computer, Jul. - Aug. 72; pp 39-48. Google ScholarDigital Library
- 14 D.F. Stevens, "System Evaluation on the Control Data 6600", Proc. 1968 IFIP Congress; C34 - C38.Google Scholar
Index Terms
- HEMI (Hybrid Events Monitoring Instrument)
Recommendations
HEMI (Hybrid Events Monitoring Instrument)
SIGMETRICS '74: Proceedings of the 1974 ACM SIGMETRICS conference on Measurement and evaluationHEMI is an experimental instrumentation system being developed for use with the CYBER 70 and 170 Series computers in order to ascertain the extent to which an integrated approach to instrumentation is economically and technologically viable for ...
Wrong Path Events: Exploiting Unusual and Illegal Program Behavior for Early Misprediction Detection and Recovery
MICRO 37: Proceedings of the 37th annual IEEE/ACM International Symposium on MicroarchitectureControl and data speculation are widely used to improve processor performance. Correct speculation can reduce execution time, but incorrect speculation can lead to increased execution time and greater energy consumption. This paper proposes a mechanism ...
Hybrid multithreading for VLIW processors
CASES '09: Proceedings of the 2009 international conference on Compilers, architecture, and synthesis for embedded systemsSeveral multithreading techniques have been proposed to reduce resource underutilization in Very Long Instruction Word (VLIW) processors. Simultaneous MultiThreading (SMT) is a popular technique that improves processor performance by issuing multiple ...
Comments