プログラムを並列実行する際に並列度を決定する要因となるのは, 主に命令間のデータ依存と制御依 存である.本論文では3 つの制御依存制約の緩和手法(分岐予測に基づいた投機的実行, 分岐命令が及 ぼす制御依存関係の正確な解析, 複数の制御フローの同時実行) の利用により高い並列度を得ることが 可能であるスレッドレベル投機的並列処理(TLSP: Thread-Level Speculative Parallel processing) アーキテクチャにおいて3 つのデータ依存制約の緩和手法(メモリ・アドレスの一致/不一致解析, リ ネーミング, 値予測) を利用することによる効果の測定を行った. その結果, メモリ・アドレスの一致 /不一致解析がリネーミングの効果に与える影響の大きさ, および値予測の効果の高さが判明した.Two fundamental restrictions that limit the amount of instruction-level parallelism extracted from sequential programs are control flow and data flow. TLSP (Thread-Level Speculative Parallel processing) architecture gains high parallelism using three techniques (speculation with branch prediction, control dependence analysis, executing multiple flows of control) which relax constraint...
Control and data flow speculation can improve processor performance through increased ILP. First it ...
With the advent of chip-multiprocessors (CMPs), Thread-Level Speculation (TLS) remains a promising t...
Speculative thread-level parallelism has been recently proposed as a source of parallelism to improv...
CPSY2000-49プログラムを並列実行する際に並列度を決定する要因となるのは主に命令間のデータ依存と制御依存である.従来は依存解析範囲の問題から命令間のデータ依存の解析が重視されていたが, LSI...
International Workshop on Informationons and Electrical Engineering (IWIE2002)Two fundamental restri...
The basic idea under speculative parallelization (also called thread-level spec-ulation) [2, 6, 7] i...
Thread-level speculation (TLS) has proven to be a promising method of extracting parallelism from bo...
Thread-level speculation (TLS) has proven to be a promising method of extracting parallelism from bo...
While architects understand how to build cost-effective parallel machines across a wide spectrum of ...
While architects understandhow to build cost-effective parallel machines across a wide spectrum of m...
The traditional single-core processors are being replaced by chip multiprocessors (CMPs) where sever...
许多程序模型是以数学或逻辑为基础设计并分析的,但程序模型的实现首先是个物理系统,它依物理规律,而非数学或逻辑规则活动.但人们习惯于用状态序列对并列程序模型的语义作数学处理.一旦偏序的状态空间用交叉的方...
This paper focuses on the problem of how to find and effectively exploit speculative thread-level pa...
Speculative thread-level parallelization is a promising way to speed up codes that compilers fail to...
TPC-C, subepochs Thread level speculation (TLS) has proven to be a promising method of extracting pa...
Control and data flow speculation can improve processor performance through increased ILP. First it ...
With the advent of chip-multiprocessors (CMPs), Thread-Level Speculation (TLS) remains a promising t...
Speculative thread-level parallelism has been recently proposed as a source of parallelism to improv...
CPSY2000-49プログラムを並列実行する際に並列度を決定する要因となるのは主に命令間のデータ依存と制御依存である.従来は依存解析範囲の問題から命令間のデータ依存の解析が重視されていたが, LSI...
International Workshop on Informationons and Electrical Engineering (IWIE2002)Two fundamental restri...
The basic idea under speculative parallelization (also called thread-level spec-ulation) [2, 6, 7] i...
Thread-level speculation (TLS) has proven to be a promising method of extracting parallelism from bo...
Thread-level speculation (TLS) has proven to be a promising method of extracting parallelism from bo...
While architects understand how to build cost-effective parallel machines across a wide spectrum of ...
While architects understandhow to build cost-effective parallel machines across a wide spectrum of m...
The traditional single-core processors are being replaced by chip multiprocessors (CMPs) where sever...
许多程序模型是以数学或逻辑为基础设计并分析的,但程序模型的实现首先是个物理系统,它依物理规律,而非数学或逻辑规则活动.但人们习惯于用状态序列对并列程序模型的语义作数学处理.一旦偏序的状态空间用交叉的方...
This paper focuses on the problem of how to find and effectively exploit speculative thread-level pa...
Speculative thread-level parallelization is a promising way to speed up codes that compilers fail to...
TPC-C, subepochs Thread level speculation (TLS) has proven to be a promising method of extracting pa...
Control and data flow speculation can improve processor performance through increased ILP. First it ...
With the advent of chip-multiprocessors (CMPs), Thread-Level Speculation (TLS) remains a promising t...
Speculative thread-level parallelism has been recently proposed as a source of parallelism to improv...