A.R. Alameldeen et al., "Evaluating Nondeterministic Multithreaded Commercial Workloads, Proc. Fifth Workshop Computer Architecture Evaluation Using Commercial Workloads, Int'l Symp. High-Performance Computer Architecture, 2002; www. hpcaconf. org/hpca8/caecw02.pdf. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
ReVive: Cost-Effective Architectural Support for Rollback.. - Prvulovic (2002) (5 citations) (Correct)
....instructions in our applications ranges from 0.06 in Water Sp to 6.4 in Ocean and 9.3 in Radix. This range covers typical miss rates in OLTP and other commercial applications. As one example, several web server and OLTP applications have been reported to have around 3 misses per 1,000 instructions [2]. Consequently, ReVive overheads with commercial workloads should not be higher than those we report here. 6 Evaluation To evaluate ReVive, we examine three issues: overhead in error free execution, storage requirements, and recovery overhead. 6.1 Overhead in Error Free Execution To evaluate ....
A. R. Alameldeen et al. Evaluating Non-deterministic Multithreaded Commercial Workloads. In 5th Workshop on computer Architecture Evaluation using Commercial Workloads, pages 30--38, Feb. 2002.
Temporally Silent Stores - Lepak, Lipasti (2002) (6 citations) (Correct)
.... 5) and exploiting this new program characteristic nonspeculatively with changes to an existing MESI coherence protocol (Section 6) We do not report direct performance benefit due to the fact that the commercial workloads we study are nondeterministic and hence difficult to measure precisely [2]. Rather, we show dramatic reductions in communication misses and argue that substantial performance benefit will result. This argument is consistent with prior results that show the substantial performance impact of communication misses [3,18] Detailed evaluation of performance benefit will be ....
A. Alameldeen, C. Mauer, M. Xu, P. Harper, M. Martin, D. Sorin, M. Hill, and D. Wood. Evaluating non-deterministic multi-threaded commercial workloads. In Proceedings of Computer Architecture Evaluation using Commercial Workloads (CAECW-02), February 2002.
Transactional Lock-Free Execution of Lock-Based Programs - Rajwar, Goodman (2002) (11 citations) (Correct)
....functional simulator works in its own memory and register space and can validate total store ordering (TSO) implementations. Care must be taken in evaluating multithreaded workloads because of non deterministic executions. Random latency perturbations are introduced in the simulator (similar to [1]) 6 Results In Section 6.1 we provide an intuition behind why TLR may improve performance. We present microbenchmark results in Section 6.2 and application results in Section 6.3. 6.1 Performance intuition With TLR, processors request data without acquiring locks and the data request is ....
A. R. Alameldeen, C. J. Mauer, M. Xu, P. J. Harper, M. M. Martin, D. J. Sorin, M. D. Hill, and D. A. Wood. Evaluating non-deterministic multi-threaded commercial workloads. In Fifth Workshop on Computer Architecture Evaluation Using Commercial Workloads, pages 30--38, Feb. 2002.
Using Lightweight Checkpoint/Recovery to Improve the Availability.. - Sorin (2002) Self-citation (Sorin) (Correct)
No context found.
Alaa R. Alameldeen, Carl J. Mauer, Min Xu, Pacia J. Harper, Milo M.K. Martin, Daniel J. Sorin, Mark D. Hill, and David A. Wood. Evaluating Non-deterministic Multi-threaded Commercial Workloads. In Proceedings of the Fifth Workshop on Computer Architecture Evaluation Using Commercial Workloads, pages 30--38, February 2002.
Permission to Make Digital Or Hard Copies of All Or Part.. - Personal Or Classroom Self-citation (Sorin) (Correct)
No context found.
Alaa R. Alameldeen, Carl J. Mauer, Min Xu, Pacia J. Harper, Milo M.K. Martin, Daniel J. Sorin, Mark D. Hill, and David A. Wood. Evaluating Non-deterministic Multithreaded Commercial Workloads. In Proceedings of the Fifth Workshop on Computer Architecture Evaluation Using Commercial Workloads, pages 30--38, February 2002.
SafetyNet: Improving the Availability of Shared Memory . . . - Sorin (2002) (8 citations) Self-citation (Martin Sorin Hill Wood) (Correct)
....up to 400,000 cycles (0. 4 msec at 1GHz) To exercise the protocol implementation, we drove it for billions of cycles with a random tester that injected faults and stressed corner cases by exploiting false sharing and reordering messages [47] Using a methodology described by Alameldeen et al. [2], we simulate each design point multiple times with small, pseudo random perturbations of memory latencies to cause Table 2. Target System Parameters L1 Cache (I and D) 128 KB, 4 way set associative L2 Cache 4 MB, 4 way set associative Memory 2 GB, 64 byte blocks Miss From Memory 180 ns ....
....standard deviation in each direction. Workloads. Commercial applications are an important workload for high availability systems. As such, we evaluate SafetyNet with four commercial applications and one scientific application, described briefly in Table 3 and in more detail by Alameldeen et al. [2]. 4.2 Experiments We perform three experiments to evaluate SafetyNet performance and show the results in Figure 5. For each workload, we plot five bars: two bars for systems unprotected by SafetyNet and three bars for systems with SafetyNet. Experiment 1: Fault Free Performance. In this ....
A. R. Alameldeen, C. J. Mauer, M. Xu, P. J. Harper, M. M. Martin, D. J. Sorin, M. D. Hill, and D. A. Wood. Evaluating Non-deterministic Multi-threaded Commercial Workloads. In Proceedings of the Fifth Workshop on Computer Architecture Evaluation Using Commercial Workloads, pages 30--38, Feb. 2002.
Variability in Architectural Simulations of Multi-threaded.. - Alameldeen, Wood (2003) (8 citations) Self-citation (Alameldeen Wood) (Correct)
....commonly used to evaluate uniprocessor and multiprocessor systems, using both measurement of current systems and simulation of future ones. Execution driven evaluation of these workloads requires full system simulation, since they spend a significant portion of their time in the operating system [1, 2, 26]. This paper identifies performance variability as a potentially major challenge for architectural simulation studies using multi threaded workloads. Variability refers to the differences between performance estimates obtained from multiple runs of the same workload. Time variability occurs when ....
....to facilitate comparisons across experiments. Other commercial benchmarks we study are: Apache (static web content serving) SPECjbb (a Java server benchmark, SPECjbb2000 [34] Slashcode (dynamic web content serving, used by slashdot.org) and ECPerf (a 3 tier Java workload) Alameldeen et al. [1] and Karlsson et al. 15] describe these workloads in more detail. We also study Barnes Hut with 16K bodies and Ocean with a 514x514 grid from the SPLASH 2 benchmark suite [38] as examples of scientific applications. OLTP: DB2 with a TPC C like workload. Our OLTP workload is based on the TPC C ....
[Article contains additional citation context not shown here]
Alaa R. Alameldeen, Carl J. Mauer, Min Xu, Pacia J. Harper, Milo M.K. Martin, Daniel J. Sorin, Mark D. Hill, and David A. Wood. Evaluating Non-deterministic Multi-threaded Commercial Workloads. In Proceedings of the Fifth Workshop on Computer Architecture Evaluation Using Commercial Workloads, pp. 30--38, February 2002.
Full-System Timing-First Simulation - Mauer, al. (2002) (3 citations) Self-citation (Mauer Hill Wood) (Correct)
....performance by verifying bundles of instructions. Finally, we present an absolute performance comparison between TFsim and RSIM [15] a previously published out of order simulator. Table 3 presents the benchmarks we used to evaluate TFsim s performance accuracy and overhead. Alameldeen et al. [1] contains a detailed description of these workloads, including their setup and configuration options. We simulate a total of 200 million instructions in all runs except the OS boot (which is run for 1.2 billion instructions) For example, each processor in a four processor system would execute ....
A. R. Alameldeen, C. J. Mauer, M. Xu, P. J. Harper, M. M. Martin, D. J. Sorin, M. D. Hill, and D. A. Wood. Evaluating Non-deterministic Multi-threaded Commercial Workloads. In Proceedings of the Fifth Workshop on Computer pages 30--38, Feb. 2002.
IEEE 2 Computer - Simulating Commercial Server (Correct)
No context found.
A.R. Alameldeen et al., "Evaluating Nondeterministic Multithreaded Commercial Workloads, Proc. Fifth Workshop Computer Architecture Evaluation Using Commercial Workloads, Int'l Symp. High-Performance Computer Architecture, 2002; www. hpcaconf. org/hpca8/caecw02.pdf.
ReVive: Cost-Effective Architectural Support for Rollback.. - Prvulovic, al. (2002) (5 citations) (Correct)
No context found.
A. R. Alameldeen et al. Evaluating Non-deterministic Multithreaded Commercial Workloads. In 5th Workshop on computer Architecture Evaluation using Commercial Workloads, pages 30--38, Feb. 2002.
Redeeming IPC as a Performance Metric for Multithreaded - Programs Kevin Lepak (2003) (Correct)
No context found.
Alaa R. Alameldeen, Carl J. Mauer, Min Xu, Pacia J. Harper, Milo M.K. Martin, Daniel J. Sorin, Mark D. Hill, and David A. Wood. Evaluating non-deterministic multi-threaded commercial workloads. Workshop On Computer Architecture Evaluation using Commercial Workloads, February 2002.
IEEE 50 Computer - Simulating Commercial Server (Correct)
No context found.
A.R. Alameldeen et al., "Evaluating Nondeterministic Multithreaded Commercial Workloads, Proc. 5th Workshop Computer Architecture Evaluation Using Commercial Workloads, Int'l Symp. High-Performance Computer Architecture, 2002; www.hpcaconf. org/hpca8/caecw02.pdf.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC