Benjie Chen and Robert Morris. Flexible control of parallelism in a multiprocessor pc router. In Proceedings of the 2001. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
A Modeling Framework for Network Processor Systems - Crowley, Baer (2002) (7 citations) (Correct)
....to Linux Figure 1. Sample Ethernet IP router click configuration. From [9] for shared resources like free buffer lists, since there is only one packet active at a time and packets are never pre empted within an element. However, Click also includes support for shared memory multiprocessors [2] and thread safe versions of elements with potentially shared state. Synchronized access and contention for shared resources will be of concern for SMP based systems, as will be seen later. 2.2. Application Model The application is modeled by profiling the Click configuration and simulating its ....
....Figure 3. Framework code declaring traffic, application and system models for the uniprocessor example. bus characterization; PCI reads have a burst size of 2, while writes have a burst size of 4. 3.2. SMP PC Click s operation on SMP based systems was described and measured in [2]. As was the case for the uniprocessorbased PC, the system modeled here will be based on the system used in the study; model estimates will be compared to the observed results from the real system. The system organization of the SMP based PC is very similar to that of the single processor system; ....
[Article contains additional citation context not shown here]
B. Chen and R. Morris. Flexible control of parallelism in a multiprocessor pc router. In Proceedings of the 2001.
A Hybrid Framework for Network Processor System Analysis - Crowley, Baer (Correct)
....system, there is no need for synchronization between elements or for shared resources like free buffer lists, since there is only one packet active at a time and packets are never pre empted within an element. However, Click also includes support for shared memory multiprocessors [2] and thread safe versions of elements with potentially shared state. Synchronized access and contention for shared resources will be of concern for shared memory multiprocessor based (SMP) systems, as will be seen later. 2.2. Application Model The application is modeled by profiling the Click ....
....is not the source and thus must use un bursted memory read transactions. This fact is reflected in the model s PCI bus characterization; PCI reads have a burst size of 2, while writes have a burst size of 4. 12 3.2. SMP PC Click s operation on SMP based systems was described and measured in [2]. The system organization is similar to that of the single processor system; Figure 3 is still an accurate depiction. The only significant resource differences are: additional processors on the host bus, slower clock rates on those processors (500 MHz vs. 700 MHz) a wider PCI bus (64b vs. 32b) ....
[Article contains additional citation context not shown here]
B. Chen and R. Morris. Flexible control of parallelism in a multiprocessor pc router. In Proceedings of the 2001.
High noon at the OKE Corral: Code Organisation and.. - Bos, Samwel (Correct)
....separation between control and data flow. In the OKE Corral highspeed packet processing can be conveniently managed by slow speed control code. The essence can be summed up as follows: 1) We borrow the LEGO like software organisation model that was advocated by the Click router project [CM01] both to build fast data paths and to implement paths for control traffic. 2) One of the components on the control path is an AN runtime (if required, this path can also be used for data packets, but because the control path is fairly slow, this is much less efficient) 3) The configuration and ....
.... and this concerns primarily the nature of the code execution: in kernel native code, versus interpreted code (often executing in userspace) Readers interested in the performance in terms of packets per second that potentially can be achieved with a channel based system should refer to [CM01] Instead, we evaluate the OKE Corral by measuring the performance of the data path components and by comparing the results with alternative implementations. All measurements described in this section were taken on a PIII 1GHz PC running a Linux 2.4.18 kernel. The overhead of a push from entry ....
[Article contains additional citation context not shown here]
Benjie Chen and Robert Morris. Flexible control of parallelism in a multiprocessor pc router. In Proc. of USENIX Annual Technical Conference (USENIX '01), pages 333-- 346, Boston, Massachusetts, June 2001.
An Architecture for Highly Concurrent, Well-Conditioned Internet.. - Welsh (Correct)
....composed using direct function calls or queues. One of the most critical aspects of application design is determining how components should be composed. Click is targeted at a specific application (packet routing) and uses a single thread to process all queues; a multiprocessor extension to Click [26] uses a thread for each processor and performs load balancing across threads. A primary goal in Click is to optimize per packet routing latency, so it is often desirable to have a single thread call directly through multiple packet processing components to avoid queueing and scheduling delays. ....
B. Chen and R. Morris. Flexible control of parallelism in a multiprocessor PC router. In Proceedings of the 2001.
KNITS: Switch-based Connection Hand-off - Papathanasiou, Van Hensbergen (2002) (1 citation) (Correct)
....[21] Among other things this would remove copies from the forwarding process and improve performance through intelligent event notification and streamlined packet handling. Recent versions of Click have shown good scalability on SMP systems with performance approaching 500,000 packets second [22]. Further improvements could be made by utilizing intelligent gigabit NICs with payload cache firmware [23] Despite the potential improvements to the soft switch architecture, it is our belief that the only truly scalable solution will be achieved using hardware acceleration features as found in ....
B. Chen, R. Morris, Flexible Control of Parallelism in a MultiprocessorPC Router, Published in the proceedings of the USENIX annual technical conference, June 2001.
Modular Components for Network Address Translation - Eddie Kohler Robert (2000) Self-citation (Morris) (Correct)
No context found.
Benjie Chen and Robert Morris. Flexible control of parallelism in a multiprocessor PC router. In Proc. 2001.
Shared Memory Multiprocessor Architectures for Software IP.. - Luo, Bhuyan, Chen (2003) Self-citation (Chen) (Correct)
No context found.
B. Chen and R. Morris, Flexible Control of Parallelism in a Multiprocessor PC Router, USENIX'01
Gigabit Routing on a Software-exposed Tiled-Microprocessor - Saif, Anderson, Digangi.. (2005) (Correct)
No context found.
Benjie Chen and Robert Morris. Flexible control of parallelism in a multiprocessor pc router. In Proceedings of the 2001.
The OKE Corral: Code Organisation and Reconfiguration at.. - Bos, Samwel (2002) (4 citations) (Correct)
No context found.
Benjie Chen and Robert Morris. Flexible control of parallelism in a multiprocessor pc router. In Proc. of USENIX Annual Technical Conference (USENIX '01), pages 333-346, Boston, Massachusetts, June 2001.
RR-TCP: A Reordering-Robust - Tcp With Dsack (Correct)
No context found.
B. Chen and R. Morris. Flexible control of parallelism in a multiprocessor PC router. Proceedings of the 2001.
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