rjschw@hubcap.UUCP (Robert Schwartz) (07/14/88)
As part of ongoing research, I want to make some measurements on the 80x86 architecture. However, we have some questions some of you out there on the network may be able to answer. What measurements would be useful? Study of some previous research on the VAX-11/780 shows various categories. For example, Clark and Levy (IEEE 1982) measure instruction frequency for a benchmark program, a compiler, and a timesharing workload. Wiecek (ACM 1982) looks at the execution of six different compilers. My plans currently are to look at the following: * Opcode frequency: Static, dynamic, for various programs, compilers etc. * Addressing modes: Same conditions as the opcode measurements. Are other measurements desirable (e.g. instruction pair or triples, opcode successors or predecessors)? Please e-mail responses, I can summarize if there is sufficient interest. Thanks in advance. -- Robert Schwartz UUCP: ...gatech!hubcap!rjschw Clemson University Internet: rjschw@hubcap.clemson.edu "What do ya want for nothin', R-r-r-r-r-ubber biscuit?" --Elwood Blues
andrew@frip.gwd.tek.com (Andrew Klossner) (07/15/88)
[]
"As part of ongoing research, I want to make some measurements
on the 80x86 architecture. However, we have some questions
some of you out there on the network may be able to answer.
What measurements would be useful?"
Why are you taking the measurements in the first place? What
objectives are you trying to achieve? If you have answers to these
questions, pass them along and we can make recommendations.
-=- Andrew Klossner (decvax!tektronix!tekecs!andrew) [UUCP]
(andrew%tekecs.tek.com@relay.cs.net) [ARPA]