Electrical computers and digital processing systems: processing – Dynamic instruction dependency checking – monitoring or... – Reducing an impact of a stall or pipeline bubble
Reexamination Certificate
2001-06-07
2003-10-14
Shin, Christopher B. (Department: 2182)
Electrical computers and digital processing systems: processing
Dynamic instruction dependency checking, monitoring or...
Reducing an impact of a stall or pipeline bubble
C710S029000, C710S054000, C711S169000, C713S500000
Reexamination Certificate
active
06633972
ABSTRACT:
BACKGROUND INFORMATION
The present invention relates to processor data/instruction flow and storage. More specifically, the present invention relates to a system and method for substituting specific dynamic pipelines with static queues in a pipelined processor.
The architecture of many processors in the art is based on a pipelined design. A processor pipeline is analogous to an assembly line, wherein the complete task is partitioned into a sequence of small sub-steps and each sub-step is carried out at a different stage of the pipeline. The pipelined design allows the processor to simultaneously process several instructions, with each pipeline stage processing successive instructions as they pass through that stage.
When a computational task is broken down into sub-steps to be executed in a pipelined fashion, often a partial result generated at a certain stage has to travel several stages down the pipeline unchanged before it is used by another stage downstream. This process is known as ‘staging’.
FIG. 1
provides an illustration of a typical staging pipeline as known in the art. In this diagram, the data, comprising four input bits
100
,
101
,
102
,
103
, is staged down a pipeline of 5 stages
111
,
112
,
113
,
114
,
115
. The number of stages of a pipeline is referred to as the “depth” (‘d’) of the pipeline. The number of inputs is referred to as the “width” (‘w’) of the pipeline.
Typical pipelined processors such as is shown in
FIG. 1
, although transporting data a relatively short distance, consume significant power. The power is consumed by the changes in state of each of the flip-flops
120
, by the line segments
122
connecting each stage to its successor, and most significantly, by the clock grid
124
.
Therefore, there is a need to improve upon the current system and method of staging pipelined processors in order to reduce power consumption and to simplify the process of clock tree design, in addition to other advantages.
REFERENCES:
patent: 5113395 (1992-05-01), Murakami et al.
patent: 5247647 (1993-09-01), Brown et al.
patent: 5325495 (1994-06-01), McLellan
patent: 5475644 (1995-12-01), Chauvel et al.
patent: 5488730 (1996-01-01), Brown et al.
patent: 5790776 (1998-08-01), Sonnier et al.
patent: 5944818 (1999-08-01), Baxter et al.
patent: 5987620 (1999-11-01), Tran
patent: 6065126 (2000-05-01), Tran et al.
patent: 6092128 (2000-07-01), Maas et al.
patent: 6115807 (2000-09-01), Grochowski
patent: 6519709 (2003-02-01), Kawauchi
Intel Corporation
Shin Christopher B.
LandOfFree
Method and apparatus for utilizing static queues in... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for utilizing static queues in..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for utilizing static queues in... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3145673