R. P. LaRowe Jr. Page Placement for Nonuniform Memory Access Time (NUMA) Shared Memory Multiprocessors. PhD thesis, Duke University, 1991. In preparation. 11  Home/Search   Document Not in Database   Summary   Related Articles   Check

....strategies may be most appropriate. In order to conduct our study of NUMA memory management policies, an implementation approach was deemed the most appropriate primary investigation methodology (though a complementary analytic investigation of the same ideas is also currently underway [6]) The sharing of hardware resources is an important logistical problem associated with implementation studies, however. In this paper, we introduce an operating system kernel, DUnX, that solves the sharing of 10 resources problem by making it possible for OS experimentation and a regular user ....

R. P. LaRowe Jr. Page Placement for Nonuniform Memory Access Time (NUMA) Shared Memory Multiprocessors. PhD thesis, Duke University, 1991. In preparation. 11

....over memory management policy. In this section, we give an overview of the DUnX virtual memory system and the support it provides for our experimental policy evaluation. BBN s nX virtual memory system is described in [3] and a more thorough description of the DUnX mechanisms can be found in [28, 30]. A conventional way for multiple users to share the resources of a GP 1000 machine is to assign each user a dedicated subset of the processor nodes (called a cluster ) DUnX expands upon this by providing users with the ability to specify and dynamically change the memory management policies used ....

R. P. LaRowe Jr. Page Placement for Nonuniform Memory Access Time (NUMA) Shared Memory Multiprocessors. PhD thesis, Duke University, March 1991.

....in a significant additional programming burden. The operating system can play a major role in managing placement through the policies and mechanisms of the virtual memory subsystem (e.g. by migrating and replicating shared pages) OS level NUMA memory management is an area of active research [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 13]. This body of work has demonstrated that dynamic page placement is indeed effective. In fact, the effectiveness question has been the focus of many of the previous studies that took the approach of proposing an algorithm and evaluating its performance (e.g. by comparing it against some static, ....

R. P. LaRowe Jr. Page Placement for Nonuniform Memory Access Time (NUMA) Shared Memory Multiprocessors. PhD thesis, Duke University, 1991.

.... Ramanathan and Ni conducted a study of critical factors in NUMA memory management reported in [25] We have also investigated OS level NUMA memory management through experimentation, both with the USMR programming library for the BBN GP1000 and with our DUnX kernel for the BBN GP1000 and TC2000 [14, 15, 17, 18, 19, 20, 21]. The unique contribution of this current work in relation to previous research is the complementary use of 1) measurements based on a flexible parameterized policy implementation that can explore a wide range of policy behavior and 2) an experimentally validated analytic model. Our focus has been ....

....p r prob. that a memory reference is remote q r prob. of a page fault that results in a replication q m prob. of a page fault that results in a migration q c prob. of a page fault that is a coherency fault Table 2: System Model Input Parameters wait times w l , w r , and w n is discussed in [17]. R = R r R f (1) R r = p l T l p r T r (2) T l = w l t l (3) T r = w n t x ) w r t l ) w n t x ) 4) Consider processor zero in Figure 1 making a reference to local memory, and then a reference to a location in memory module one. In the first case, the only wait required ....

R. P. LaRowe Jr. Page Placement for Nonuniform Memory Access Time (NUMA) Shared Memory Multiprocessors. PhD thesis, Duke University, March 1991.

....the locality issue by migrating or replicating a page of virtual storage to a node that appears to be accessing it frequently. The decision of whether to migrate or replicate a page depends on reference patterns and the type of reference (read or write) being performed. The DUnX operating system [9, 10, 8] is an example of an operating system that supports such dynamic page placement. DUnX is a kernel developed for the BBN Butterfly family of computers. DUnX supports dynamic page placement which means that a page may be migrated or replicated to different nodes in response to observed usage ....

R. P. LaRowe Jr. Page Placement for Nonuniform Memory Access Time (NUMA) Shared Memory Multiprocessors. PhD thesis, Duke University, 1991.

.... of architectural features on the performance of four migration and replication policies in [6] We have also investigated OS level NUMA memory management through experimentation, both with the USMR programming library for the BBN GP1000 and with our DUnX kernel for the BBN GP1000 and TC2000 [9, 10, 11, 12, 13, 14, 15, 18]. For this research, implementation based experimentation has several advantages over more formal approaches. Most importantly, real applications can be used as the workload. Additionally, the likelihood of discovering the subtle issues that may be important to addressing the problem increases. On ....

....write granularity. The ideal policy can be modifed to introduce errors (i.e. poor policy choices) so that their effects can be compared to the ideal performance for different workload assumptions. After validating our model using experimental data obtained with the DUnX operating system kernel [11, 12, 14], we use the ideal policy and its modifications to address several issues. These include the effectiveness of dynamic single copy page placement, the effectiveness of dynamic multiple copy page placement, the cost of using replication coherency fault pairs instead of page migrations, and the cost ....

R. P. LaRowe Jr. Page Placement for Nonuniform Memory Access Time (NUMA) Shared Memory Multiprocessors. PhD thesis, Duke University, March 1991.

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