Electrical computers and digital data processing systems: input/ – Input/output data processing – Input/output access regulation
Reexamination Certificate
2000-11-15
2004-01-20
Gaffin, Jeffrey (Department: 2182)
Electrical computers and digital data processing systems: input/
Input/output data processing
Input/output access regulation
C710S033000, C710S052000, C709S241000, C709S241000
Reexamination Certificate
active
06681270
ABSTRACT:
TECHNICAL FIELD OF THE INVENTION
The technical field of this invention is digital signal processing and more particularly control of data transfers within a digital signal processing system.
BACKGROUND OF THE INVENTION
Digital signal processing (DSP) differs significantly from general purpose processing performed by micro-controllers and microprocessors. One key difference is the strict requirement for real time data processing. For example, in a modem application, it is absolutely required that every sample be processed. Even losing a single data point might cause a digital signal processor application to fail. While processing data samples may still take on the model of tasking and block processing common to general purpose processing, the actual data movement within a digital signal processor system must adhere to the strict real-time requirements of the system.
As a consequence, digital signal processor systems are highly reliant on an integrated and efficient direct memory access (DMA) engine. The direct memory access controller is responsible for processing transfer requests from peripherals and the digital signal processor itself in real time. All data movement by the direct memory access must be capable of occurring without central processing unit (CPU) intervention in order to meet the real time requirements of the system. That is, because the CPU may operate in a software tasking model where scheduling of a task is not as tightly controlled as the data streams require, the direct memory access engine must sustain the burden of meeting all real time data stream requirements in the system.
The early direct memory access has evolved into several successive versions of centralized transfer controllers and more recently into the transfer controller with hub and ports architecture. The transfer controller with hub and ports architecture is described in U.S. Pat. No. 6,496,740 claiming priority from U.K. Patent Application No. 9909196.9 filed Apr. 10, 1999 entitled “TRANSFER CONTROLLER WITH HUB AND PORTS ARCHITECTURE”.
A first transfer controller module was developed for the TMS330C80 digital signal processor from Texas Instruments. The transfer controller consolidated the direct memory access function of a conventional controller along with the address generation logic required for servicing cache and long distance data transfer, also called direct external access, from four digital signal processors and a single RISC (reduced instruction set computer) processor.
The transfer controller architecture of the TMS330C80 is fundamentally different from a direct memory access in that only a single set of address generation and parameter registers is required. Prior direct memory access units required multiple sets for multiple channels. The single set of registers, however, can be utilized by all direct memory access requesters. Direct memory access requests are posted to the transfer controller via set of encoded inputs at the periphery of the device. Additionally, each of the digital signal processors can submit requests to the transfer controller. The external encoded inputs are called “externally initiated packet transfers” (XPTs). The digital signal processor initiated transfers are referred to as “packet transfers” (PTs). The RISC processor could also submit packet transfer requests to the transfer controller.
The transfer controller with hub and ports introduced several new ideas concepts. The first was uniform pipelining. New digital signal processor devices containing a transfer controller with hub and ports architecture have multiple external ports, all of which look identical to the hub. Thus peripherals and memory may be freely interchanged without affecting the hub. The second new idea is the concept of concurrent execution of transfers. That is, up to N transfers may occur in parallel on the multiple ports of the device, where N is the number of channels in the transfer controller with hub and ports core. Each channel in the transfer controller with hub and ports core is functionally just a set of registers. This set of registers tracks the current source and destination addresses, the word counts and other parameters for the transfer. Each channel is identical, and thus the number of channels supported by the transfer controller with hub and ports is highly scalable.
Finally the transfer controller with hub and ports includes a mechanism for queuing transfers up in a dedicated queue memory. The TMS320C80 transfer controller permitted only was one transfer outstanding per processor at a time. Through the queue memory provided by the transfer controller with hub and ports, processors may issue numerous transfer requests up to the queue memory size before stalling the digital signal processor.
SUMMARY OF THE INVENTION
The transfer controller with hub and ports of this invention is an improvement over that described in U.S. Pat. No. 6,496,740 claiming priority from U.K. Patent Application No. 9909196.9 filed Apr. 10, 1999 entitled “TRANSFER CONTROLLER WITH HUB AND PORTS ARCHITECTURE”. The improvement of this invention is effective channel priority processing.
The effective channel priority processing of this invention is the solution to the task of monitoring and arbitration of conflicting transfers that could cause major performance degradation if only the simple channel priority assigned determined which transfer should proceed first. The technique detailed here acts to raise the effective channel priority of a low priority transfer task which is blocking the completion of a higher priority transfer task.
This conflict comes about when both the lower priority channel and the higher priority channel are accessing data from the same port, but the lower priority task was initiated on an earlier processor cycle and hence is at the front of the queue for that port. Effective channel priority processing promotes the first task in the queue of a given port to the highest priority level represented by tasks on that queue.
REFERENCES:
patent: 5764895 (1998-06-01), Chung
patent: 5923654 (1999-07-01), Schnell
patent: 6011793 (2000-01-01), Smith
patent: 6044061 (2000-03-01), Aybay et al.
patent: 6088744 (2000-07-01), Hill
patent: 6249520 (2001-06-01), Steely et al.
patent: 6304578 (2001-10-01), Fluss
patent: 6493347 (2002-12-01), Sindhu et al.
patent: 6516442 (2003-02-01), Wang et al.
patent: 6535510 (2003-03-01), Kalkunte et al.
patent: 6546420 (2003-04-01), Lemier et al.
Agarwala Sanjive
Comisky David A.
Robertson Iain
Brady III W. James
Gaffin Jeffrey
Marshall, Jr. Robert D.
Sorrell Eron
Telecky , Jr. Frederick J.
LandOfFree
Effective channel priority processing for transfer... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Effective channel priority processing for transfer..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Effective channel priority processing for transfer... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3229973