TY - GEN
T1 - Using profile information to assist advanced compiler optimization and scheduling
AU - Chen, W.
AU - Bringmann, R.
AU - Mahlke, S.
AU - Anik, S.
AU - Kiyohara, T.
AU - Warter, N.
AU - Lavery, D.
AU - Hwu, W. M.
AU - Hank, R.
AU - Gyllenhaal, J.
N1 - Publisher Copyright:
© Springer-Verlag Berlin Heidelberg 1993.
PY - 1993
Y1 - 1993
N2 - Compilers for superscalar and VLIW processors must expose sufficient instruction-level parallelism in order to achieve high performance. Compile-time code transformations which expose instruction-level parallelism typically take into account the constraints imposed by all execution scenarios in the program. However, there are additional opportunities to increase instruction-level parallelism along the frequent execution scenario at the expense of the less frequent execution sequences. Profile information identifies these important execution sequences in a program. In this paper, two major categories of profile information are studied: control-flow and memory-dependence. Profile-based transformations have been incorporated into the IMPACT compiler. These transformations include global optimization, acyclic global scheduling, and software pipelining. The effectiveness of these profile-based techniques is evaluated for a range of superscalar and VLIW processors.
AB - Compilers for superscalar and VLIW processors must expose sufficient instruction-level parallelism in order to achieve high performance. Compile-time code transformations which expose instruction-level parallelism typically take into account the constraints imposed by all execution scenarios in the program. However, there are additional opportunities to increase instruction-level parallelism along the frequent execution scenario at the expense of the less frequent execution sequences. Profile information identifies these important execution sequences in a program. In this paper, two major categories of profile information are studied: control-flow and memory-dependence. Profile-based transformations have been incorporated into the IMPACT compiler. These transformations include global optimization, acyclic global scheduling, and software pipelining. The effectiveness of these profile-based techniques is evaluated for a range of superscalar and VLIW processors.
UR - http://www.scopus.com/inward/record.url?scp=70449836627&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=70449836627&partnerID=8YFLogxK
U2 - 10.1007/3-540-57502-2_38
DO - 10.1007/3-540-57502-2_38
M3 - Conference contribution
AN - SCOPUS:70449836627
SN - 9783540575023
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 31
EP - 48
BT - Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings
A2 - Banerjee, Utpal
A2 - Gelernter, David
A2 - Nicolau, Alex
A2 - Padua, David
PB - Springer
T2 - IFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017
Y2 - 3 September 2017 through 7 September 2017
ER -