Proceedings. 25th Annual International Symposium on Computer Architecture (Cat. No.98CB36235)
DOI: 10.1109/isca.1998.694758
|View full text |Cite
|
Sign up to set email alerts
|

Memory system characterization of commercial workloads

Abstract: Commercial applications such as databases and Web servers constitute the largest and fastest-growing segment of the market for multiprocessor servers. Ongoing innovations in disk subsystems, along with the ever increasing gap between processor and memory speeds, have elevated memory system design as the critical performance factor for such workloads. However, most current server designs have been optimized to perform well on scientific and engineering workloads, potentially leading to design decisions that are… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

12
133
1

Publication Types

Select...
4
3
2

Relationship

0
9

Authors

Journals

citations
Cited by 161 publications
(146 citation statements)
references
References 19 publications
12
133
1
Order By: Relevance
“…For instance, the workload characterization studies that analyze micro-architectural behavior of the OLTP workloads demonstrate that transactions exhibit significant stalls during execution [2,7,22,62]; a result we corroborate in Section 6.2. It has also been shown that sharedeverything systems have frequent shared read-write accesses [9,22], which are difficult to predict [56].…”
Section: Shared-everything Database Deploymentssupporting
confidence: 68%
See 1 more Smart Citation
“…For instance, the workload characterization studies that analyze micro-architectural behavior of the OLTP workloads demonstrate that transactions exhibit significant stalls during execution [2,7,22,62]; a result we corroborate in Section 6.2. It has also been shown that sharedeverything systems have frequent shared read-write accesses [9,22], which are difficult to predict [56].…”
Section: Shared-everything Database Deploymentssupporting
confidence: 68%
“…The absolute difference in performance is much lower than in the case of counter incrementing because executing a transaction has significant start-up and finish costs, and during transaction execution a large fraction of the time is spent on operations other than accessing data. For instance, studies show that around 20% of the total instructions executed during OLTP are data loads or stores (e.g., [7,22]). …”
Section: Variable Communication Latenciesmentioning
confidence: 99%
“…Despite an abundance of research over the years, instruction cache (icache) miss stalls remain a source of performance degradation for many commercial applications [2] [7][10] [11] [15][17] [23]. Due to the relatively larger performance cost of data cache misses in most applications, research and development has focused primarily on the data cache miss problem instead.…”
Section: Introductionmentioning
confidence: 99%
“…More specifically, Keeton et al measure an instruction-related stall component of 41% of the total execution time for Informix running TPC-C on a PentiumPro [KP+98]. When running transactional (TPC-B and TPC-C) and decisionsupport (TPC-H) benchmarks on top of Oracle on Alpha processors, instruction stalls account for 45% and 30% of the execution time, respectively [BGB98] [SBG02]. A recent study of DB2 7.2 running TPC-C on Pentium III [SA04] attributes 22% of the execution time to instruction stalls.…”
Section: Database Workloads On Modern Processorsmentioning
confidence: 99%