B. Mukherjee, G. Eisenhauer and K. Ghosh, A Machine Independent Interface for Lightweight Threads, Operating Systems Review of the ACM Special Inter-est Group in Operating Systems, pages 33-47, January 1994. 158  Home/Search   Document Details and Download   Summary   Related Articles   Check

....functionality for multiprogramming support with arbitrary forms of parallelism. Although in Unix platforms there is a variety of custom based user level thread packages, this is not the case for the Windows NT platforms. The only one user level thread package that we found is that of Cthreads [8]. However, there is no information given about the implementation and technical details of this porting. The development on Cthreads itself continues only to the extent that it supports ongoing projects. 6 Conclusions and Future Work This paper presented the de elopment of two multithreaded ....

B. Mukherjee, G. Eisenhauer and K. Ghosh, A Machine Independent Interface for Lightweight Threads, Operating Systems Review of the ACM Special Inter-est Group in Operating Systems, pages 33-47, January 1994. 158

....functionality for multiprogramming support with arbitrary forms of parallelism. Although in Unix platforms there is a variety of custom based user level thread packages, this is not the case for the Windows NT platforms. The only one user level thread package that we found is that of Cthreads [8]. However, there is no information given about the implementation and technical details of this porting. The development on Cthreads itself continues only to the extent that it supports ongoing projects. 6 Conclusions and Future Work This paper presented the development of two multithreaded ....

B. Mukherjee, G. Eisenhauer and K. Ghosh, A Machine Independent Interface for Lightweight Threads, Operating Systems Review of the ACM Special Inter-est Group in Operating Systems, pages 33-47, January 1994.

....rather misleadingly called monitors. However, notify and wait operations must be explicitly invoked within such sections, rather than being automatically associated with entry and exit. There are many other thread packages available providing lightweight processes with shared memory communication [40, 41, 87, 155]. Rendezvous. Rendezvous based programming models are distributed memory paradigms using a particular cooperation mechanism. In the rendezvous communication model, an interaction between two processes A and B takes place when A calls an entry of B, and B executes an accept for that 38 entry. An ....

B. Mukherjee, G. Eisenhauer, and K. Ghosh. A machine independent interface for lightweight threads. ACM Operating Systems Review, 28(1):33--47, January 1994.

....The PANDA system provided two abstractions: protection domain and virtual processor. Aegis allowed applications to run user code safely in the kernel. The Arena approach is to provide interface primitives to low level hardware abstractions, but to disallow direct access to the processor hardware. Mukherjee et al. 1994) presented a hardware independent interface for threads with a hw save context( hwrestore context( pair, though their interface was implemented on top of operating systems, and not, as here, on the hardware itself. Upcalls in Arena are associated with threads and are thus asynchronous. ....

Mukherjee, B, G. Eisenhauer and K. Ghosh (1994) A machine independent interface for lightweight threads. Operating Systems Review 28(1), 33-47.

....are commonly available in other thread libraries are used within the engine. These two aspects of the engine allow the substitution of the Solaris multithreading library with another library easily. Aside from the currently used Solaris multithreading library, HPAM Sim was tested using Cthreads [19]. In addition to producing fast simulators, using lightweight threads offers an additional advantage. HPAM Sim can be used on a uniprocessor host as well as any multiprocessor host that supports multithreading. This allows the simulation of small benchmarks on uniprocessor hosts and the simulation ....

Mukherjee, B., Eisenhauer, G., and Ghosh, K. A Machine Independent Interface for Lightweight Threads. Operating Systems Review of the ACM Special Interest Group in Operating Systems, pages 33--47, January 1994.

....program written in FORTRAN or C with message passing calls. This program is first instrumented (i.e code is injected into the assembly code to count the number of cycles) using the augmentation module of HPAM Sim. The program is then executed on a light weight multithreaded interface (Cthreads [Mu et al. 93] to the physical machine. Therefore, HPAM Sim can run on a uniprocessor as well as a multiprocessor machines that support multithreading. This allows the simulation of small and medium scale applications using a uniprocessor machine, and large applications using multiprocessor machines. 5 ....

B. Mukherjee, G. Eisenhauer and K. Ghosh, "A Machine Independent Interface for Lightweight Threads," College of Computing, Georgia Institute of Technology, Tech. Rep. GIT-CC-93-53, 1993.

....a PIONIER grant from the Netherlands Organization for Scientific Research (N.W.O. performance on most platforms, all of these runtime systems employ hand tuned, custom thread packages with a different implementation on most platforms. Portable thread packages, such as Pthreads [10] or Cthreads [16], suffer from two drawbacks that preclude their use in most parallel runtime systems: 1) their portability comes at the expense of raw performance on a given platform, and thus these packages are typically slower than most hand crafted, platformspecific thread packages; and (2) their closed, ....

B. Mukherjee, G. Eisenhauer, and K. Ghosh. A Machine Independent Interface for Lightweight Threads. Technical Report CIT-CC-93/53, College of Computing, Georgia Institute of Technology, Atlanta, Georgia, 1993.

....An extensive discussion of these approaches is beyond the scope of this section. In the absence of architecture support, lightweight threads that support efficient inter thread communication provide an efficient and attractive platform for parallel processing. Further information can be found in [15, 17, 12, 16, 11, 23, 27, 29, 13, 6, 25, 28, 4]. 7 Conclusions This article compares the performance of two data driven multithreaded execution models and their associated synchronization schemes. The blocking model relies on frame based data storage. A frame contains all the data related to a code block. All the thread instances that belong ....

B. Mukherjee, G. Eisenhauer, and K. Ghosh. A machine independent interface for lightweight threads. Technical Report CIT-CC-93/53, College of Computing, Georgia Institutre of Technology, 1993.

....all matrix elements are initialized to zero. The element update appears as shown below: if (ij = kl) TFM[ijkl] 2. 0 (L[lt] kt] L[kt] lt] else TFM[ijkl] L[lt] kt] L[kt] lt] Parallelization of this computation is implemented using a portable multiprocessor threads package, Cthreads [3]. The program spawns m 1 concurrently executing threads, including one master thread and m slave threads. Each thread runs on a distinct processor and executes a set of outer loop iterations, O(h 2 m) The number of iterations per thread determines computation granularity. The array TFM , which ....

....that there are two threads updating an array A in shared memory. All elements of A are initially zero, and exactly two new values will be added to each value in A. Suppose that Time Thread 1 Thread 2 A Case 1 init 0 0 0 0 0 0 A[1] A[2] 0 x x 0 0 1 A[2] 0 x x 0 0 2 A[4] A[4] 0 x x 0 x 3 A[0] A[3] x x x x x 4 A[1] x x x x x 5 A[0] x x x x x 6 A[3] x x x x x Case2 init 0 0 0 0 0 0 A[1] A[2] 0 x x 0 0 1 A[2] 0 x x 0 0 2 A[4] 0 x x 0 x 3 A[0] A[4] x x x 0 x 4 A[1] A[3] x x x x x 5 A[0] x x x x x 6 A[3] x x x x x Table 1. An example of memory accesses for two threads on an array A. x ....

[Article contains additional citation context not shown here]

B. Mukherjee, G. Eisenhauer, and K. Ghosh. A machine independent interface for lightweight threads. Operating Systems Review of the ACM Special Interest Group in Operating Systems, page 33, 1994.

....us to roll back to the middle of an event. Such checkpointing, schedulability analyses, and maintenance of timing information is performed transparently to the application, which is written as a collection of events with execution times and deadlines (timestamps) A lightweight threads library [19] was modified to perform these operations. The slot list can be optimized. From the preceding examples, it is clear that for the schedulability analyses of an arriving event, we need to update the slots corresponding only to those events with deadlines larger than that on the arriving event. The ....

....LP that rejected the original event. 7 These threads may be the same in certain implementations. Keeping them distinct ensures generality, and is useful in rolling back to the middle of an event, as stated later. 8 Context save and restore take about 6 secs. each on the KSR, as reported in [19]. 9 We use the same thread for dispatchingother threads andperformingschedulability analyses. Deadlines: E1: 3 E2: 5 E3: 10 E4: 20 E5: 25 E6: 35 E1 E2 E3 E4 E5 E6 10 20 30 Hash Table Slot List Figure 3: Hashing scheme to bypass the initial part of the slot list an example to explain the ....

B. Mukherjee, G. Eisenhauer, and K. Ghosh. A machine independent interface for lightweight threads. In OS Review of the ACM Special Interest Group in Operating Systems, pages 33--47, January 1994.

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