A System and Method for generating Cyclic Redundancy Check (CRC) values in a system adapted simultaneously handling a plurality of blocks in parallel is described. Included is a memory or other storage device for storing data blocks, wherein the memory or storage device is adapted to output a plurality of data blocks in parallel. A data bus provides a data path wide enough to accommodate the parallel data blocks and is further coupled to a plurality of CRC cores coupled to the data bus, wherein CRC values are calculated for every combination of data blocks on the data bus. A multiplexer coupled to the CRC cores selects the output of one of the CRC cores based on the number of valid data blocks on the data bus. Once the correct CRC value has been calculated, it is appended to a data segment, comprised of a group of data blocks, for transmission to another device.

System and method for generating 128-bit cyclic redundancy check values with 32-bit granularity

The increasing bandwidth of networks and storage devices in recent years has placed greater emphasis on the performance of low level operating system services. Data must be delivered between hardware devices and user applications in an efficient matter. Motivated by the need for low overhead operating system services, the Raven kernel utilizes user level implementation techniques to reduce kernel intervention for many common services. In particular, our user level send/receive/reply communication implementation generates no kernel interactions per iteration in the best case, and two kernel interactions in the worst case. In more general cases, we observe approximately one kernel interaction for every two send/receive/reply iterations. Device driver support is also done entirely at the user level reducing copy costs and context switching.

User level IPC and device management in the Raven kernel

Emerging high-speed networks and powerful end-systems give rise to a new class of applications such as video-on-demand and teleconferencing. Such applications are very demanding on Quality of Service (QoS) because of the isochronous nature of media they are using. To provide QoS support on an end-to-end basis, the need for the integration of network, transport, and operating services arises. Thus to support the new emerging services, an end-to-end QoS management is required. This paper deals with QoS management for multimedia applications by taking remote access to multimedia database as a case study. The example application is introduced and the entities involved in QoS provision are identified. QoS management activities are defined and a basic QoS management architecture for multimedia applications is presented. A general framework for QoS (re)negotiation is defined and an instantiation of this framework in the context of the example application is presented.

/pdf/an-approach-to-quality-of-service-management-for-distributed-uxv6rbhlvc.pdf

An Approach to Quality of Service Management for Distributed Multimedia Applications

The use of high performance networking technologies, e.g., ATM networks, demands much from both operating systems and processors. During high network loads, user threads may be starved because the processor spends all its time handling interrupts.

To alleviate this problem, we propose using a two level network interface servicing scheme that uses interrupts during low network loads to provide low latency, and polling threads during high network loads to avoid user thread starvation.

In this paper, we examine the use of such a scheme on dual-processor workstations running Windows NT connected by an ATM network. Performance evaluation of our multiprocessor prototype implementation shows that using our two level scheme can improve performance when used carefully.

A scheduling scheme for network saturated NT multiprocessors

The new quality in the relation between networks and end-systems requires a harmonization between these worlds. Based on the requirements and possibilities of new applications, particularly in the field of multimedia, the architecture of personal end-system has to be reconsidered completely.

/pdf/micronet-machines-new-architectural-approaches-for-4e489vq9ar.pdf

Micronet Machines - New Architectural Approaches for Multimedia End-Systems

The design of a parallel protocol processing system for a host system connected to an asynchronous transfer mode (ATM) network is described. Interesting aspects of the design are that the AAL5 adaptation layer is realized in hardware directly on the network adapter, and that the network adapter semetric interfaces to a multiprocessor using the processor's memory bus. Support also exists for implementing parallel protocols up to and including the application layer. Parallel implementations of TCP/IP, ISO Session, and presentation protocols are outlined. >

M. Goldberg

Papers

Parallel host interface for an ATM network