Electrical computers and digital processing systems: memory – Storage accessing and control – Access timing
Reexamination Certificate
1999-01-15
2001-07-24
Ellis, Kevin L. (Department: 2180)
Electrical computers and digital processing systems: memory
Storage accessing and control
Access timing
Reexamination Certificate
active
06266750
ABSTRACT:
BACKGROUND TO THE INVENTION
1. Field of the Invention
The present invention relates generally to command processing applications in high bandwidth memory systems.
2. Cross Reference to Other Applications
The following pending application is owned by the assignee of the present application, and its contents are hereby incorporated by reference:
Ser. No. 09/132,158 filed Aug. 10, 1998, invented by Gustavson et. al and entitled, MEMORY SYSTEM HAVING SYNCHRONOUS-LINK DRAM (SLDRAM) DEVICES AND CONTROLLER.
3. Description of the Related Art
The evolution of the dynamic random access memories used in computer systems has been driven by ever-increasing speed requirements mainly dictated by the microprocessor industry. Dynamic random access memories (DRAMs) have generally been the predominant memories used for computers due to their optimized storage capabilities. This large storage capability comes with the price of slower access time and the requirement for more complicated interaction between memories and microprocessors/microcontrollers than in the case of say static random access memories (SRAMS) or non-volatile memories.
In an attempt to address this speed deficiency, DRAM design has implemented various major improvements, all of which are well documented. Most recently, the transition from Fast Page Mode (FPM) DRAM to Extended Data Out (EDO) DRAMs and synchronous DRAMs (SDRAMS) has been predominant. Further speed increases have been achieved with double data rate (DDR) SDRAM, which synchronizes data transfers on both clock edges. New protocol based memory interfaces have recently been developed to further increase the bandwidth and operating frequencies of synchronous memories.
As the complexity of these memories has increased, the associated control systems responsible for internally managing the operation of the memories have also become more complex. These command-driven control systems internally must typically process a stream of commands or instructions that overlap in execution time and have programmable latency (time from receipt of command to first control outputs asserted in response). Programmable latency is desirable in such systems in order to allow the memory controller to schedule the use of shared data, address or control buses for optimum usage. Since the processing of two or more commands may be required to occur simultaneously, many control systems implement multiple functional units operating in parallel. The minimum latency of the control system is therefore limited by the need to (i) decode the command control field(s), (ii) determine the programmed latency associated with the identified command, and (iii) issue the command to a number of parallel functional units before the first control output action can be determined for use by the memory.
A conventional implementation of such a memory system control block comprises a single front end decoding block which decodes external commands and issues internal commands to multiple identical functional elements capable of operating in parallel. The minimum latency therefore cannot be shorter than the time it takes to decode the command in the front end block plus the time required to issue the command to a parallel functional unit, and finally, the time that the functional unit takes to initialize and issue its first control action. The common approach to reducing the minimum latency described above is by replicating the command decoding logic within each parallel functional unit and feeding the command stream to all parallel functional units simultaneously to eliminate the issue and initialization delay. This advantage comes with the cost of a large increase in overall logic complexity, redundant logic, and increased power consumption. As frequency and bandwidth requirements increase, there is a need for a memory system control block which makes optimum use of area and power consumption and which can process commands with a reduced minimum latency than previously achieved in the prior art.
SUMMARY OF THE INVENTION
It is therefore an object of the present invention to provide a command processing system for use in a high bandwidth memory interface which processes commands with a minimum latency.
It is another object of the present invention to provide the command processing system with a minimum increase to the command circuitry.
According to the invention, roughly described, a packet-driven memory control system which implements a variable length pipeline includes a command front end and one or more parallel command sequencers. The command front end decodes an external command packet into an internal command and issues it to a selected one of the command sequencers. The command has associated therewith a desired latency value. A first group of one or more memory control steps for the given command is performed by the command front end if the desired latency value is less than a threshold latency value, or by the selected command sequencer if the desired latency value is greater than or equal to the threshold latency value. The remainder of the memory control steps required for the command are performed by the selected command sequencer. If the first control steps are to be performed by the selected command sequencer, then depending on the desired latency value, the command sequencer further may insert one or more wait states before doing so.
REFERENCES:
patent: 5946260 (1999-08-01), Manning
patent: 6088774 (2000-07-01), Gillingham
patent: 6175905 (2001-01-01), Manning
patent: 6178488 (2001-01-01), Manning
DeMone Paul W.
Gillingham Peter B.
Advanced Memory International, Inc.
Ellis Kevin L.
Fliesler Dubb Meyer & Lovejoy LLP
LandOfFree
Variable length pipeline with parallel functional units does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Variable length pipeline with parallel functional units, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Variable length pipeline with parallel functional units will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2564816