Method of controlling the mode of parallel operation of a multi-mode parallel graphics processing system (MMPGPS) embodied within a host comuting system

• US 20090027402A1
• Filed: 08/30/2007
• Published: 01/29/2009
• Est. Priority Date: 11/19/2003
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Lucid Information Technology, Ltd.
• Inventors: Bakalash, Reuven, Leviathan, Yaniv
• Granted Patent: US 7,961,194 B2
• Time in Patent Office: Days
• US Class Current: 345/505
• CPC Class Codes: G06F 2209/501 Performance criteria
• CPC Class Codes: G06F 9/5044 considering hardware capabilities
• CPC Class Codes: G06T 1/20 Processor architectures; Processor configuration, e.g. pipelining
• CPC Class Codes: G06T 15/005 General purpose rendering architectures
• CPC Class Codes: G06T 2210/52 Parallel processing
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46

Abstract

A method of controlling the mode of parallel operation of a multi-mode parallel graphics processing system (MMPGPS) embodied within a host computing system having (i) host memory space (HMS) for storing one or more graphics-based applications and a graphics library for generating graphics commands and data (GCAD) during the run-time (i.e. execution) of the graphics-based application, (ii) one or more CPUs for executing said graphics-based applications, (iii) a display device for displaying images containing graphics during the execution of said graphics-based applications, and (iv) a multi-mode parallel graphics rendering subsystem supporting multiple modes of parallel operation selected from the group consisting of object division, image division, and time division and having a plurality of graphic processing pipelines (GPPLs) supporting a parallel graphics rendering process that employs one of the object division, image division and/or time division modes of parallel operation.

Claims

1-305. -305. (canceled)

306. A method of controlling the mode of parallel operation of a multi-mode parallel graphics processing system (MMPGPS) embodied within a host computing system having (i) host memory space (HMS) for storing one or more graphics-based applications and a graphics library for generating graphics commands and data (GCAD) during the run-time (i.e. execution) of the graphics-based application, (ii) one or more CPUs for executing said graphics- based applications, (iii) a display device for displaying images containing graphics during the execution of said graphics-based applications, and (iv) a multi-mode parallel graphics rendering subsystem supporting multiple modes of parallel operation selected from the group consisting of object division, image division, and time division and having a plurality of graphic processing pipelines (GPPLs) supporting a parallel graphics rendering process that employs one of said object division, image division and/or time division modes of parallel operation, said method comprising:

(a) during the run-time of at least one said graphics-based application, analyzing the stream of graphic commands and data (GCAD) produced by said graphics-based application, and/or performance data associated with said MMPGPS and said host computing system;

(b) based on the monitoring of said GCAD and/or said performance data during step (a), automatically controlling the mode of parallel operation of said MMPGPS to as to operate in a first one of said supported modes of parallel operation, during the run-time of said graphics- based application;

(c) while said MMPGPS is operating in the first one of said supported modes of operation, said GPPLs processing said GCAD, and rendering a first set of pixel-composited images containing graphics for display on said display device during the run-time of said graphics-based application;

(d) during the run-time of said graphics-based application, continuing to monitor said GCAD and/or said performance data, automatically controlling the mode of parallel operation of said MMPGPS to as to operate in a second one of said supported modes of parallel operation, during the run-time of said graphics application; and

(e) while said MMPGPS is operating in the second one of said supported modes of operation, said GPPLs processing said GCAD, and rendering a second set of pixel-composited images containing graphics for display on said display device during the run-time of said graphics-based application.

307. The method of claim 306, wherein during step (a), said analysis of GCAD data and performance data involves profiling of scenes in said graphics-based application, on an image frame by image frame basis, in real-time during run-time of said graphics-based application.

308. The method of claim 307, wherein said real-time profiling of scenes in said graphics-based application involves (i) collecting and analyzing said performance data during application run-time, (ii) constructing scene profiles for the image frames associated with particular scenes in said particular graphics-based application, and (iii) maintaining said scene profiles in an application/scene profile database that is accessible during run-time, so that during the run-time of said graphics-based application, said scene profiles maintained in said application/scene profile database can be accessed to determine how to dynamically control the modes of parallel operation of said MMPGPS to optimize system performance.

309. The method of claim 307, wherein said real-time profiling of scenes in said graphics-based application involves real-time detection of scene profile indices directly programmed within pre-profiled scenes of said graphics-based application;

wherein said pre-profiled scenes are analyzed prior to run-time, and indexed with said scene profile indices; and

wherein and mode control parameters (MCPs) corresponding to said scene profile indices, are stored within an application/scene profile database accessible during application run- time.

310. The method of claim 307, wherein said real-time profiling of scenes in said graphics-based application involves real-time detection of mode control commands (MCCs) directly programmed within pre-profiled scenes of said graphics-based application;

wherein said pre-profiled scenes are analyzed prior to run-time, and said MCCs are directly programmed within the individual image frames of each scene; and

wherein during run-time, said automatic mode control module automatically detects said MCCs along the stream of GCAD, and uses said MCCs so as to determine how to dynamically control the modes of parallel operation of said MMPGPS to optimize system performance.

311. The method of claim 306, wherein said display unit is a device selected from the group consisting of an flat-type display panel, a projection-type display panel, and other image display devices.

312. The method of claim 306, wherein said host computing system is a machine selected from the group consisting of a PC-level computer, information server, laptop, game console system, portable computing system, and any computational-based machine supporting the real-time generation and display of 3D graphics.

313. The method of claim 306, wherein said GCAD and performance parameters include elements selected from the group consisting of:

(i) texture count;

(ii) screen resolution;

(iii) polygon volume;

(iv) at each GPU with the GPPL, the utilization of (a) geometry engine, (b) pixel engine, and (c) video memory;

(v) Utilization of CPU;

(vi) total pixels rendered;

(vii) total geometric data rendered;

(viii) workload of each GPU; and

(ix) volumes of transferred data.

314. The method of claim 306, wherein steps (b) and (d), further include the performance of one or more tasks selected from the group consisting of:

(1) Recognizing any particular graphics-based application;

(2) Processing trial &

error results;

(3) Utilizing a application/scene profile from an application/scene profile database;

(4) Performing data aggregation in a historical repository;

(5) Analyzing input performance data;

(6) Analyzing based on integration of (a) frame-based “

atomic”

performance data, (b) aggregated data in said historical repository, and (c) data in an application/scene profile database;

(7) Detecting rendering algorithms used by said graphics-based application;

(8) Detecting the use of the frame buffer (FB) in next successive frame as a preventive condition for time division mode;

(9) Recognizing preventive conditions for other parallel modes;

(10) Evaluating the pixel layer depth at the pixel subsystem of each GPU;

(11) Determining the frame/sec count;

(12) Detecting critical events (e.g. frame/sec drop);

(13) Detecting bottlenecks in said GPPLs;

(14) Measuring and balancing loads among said GPPLs;

(15) Updating an application/scene profile database using data stored in a historical depository; and

(16) Selecting an optimal mode of parallel operation.

315. The method of claim 306, wherein during the course of analysis and profiling, a plurality of the following parameters are considered and analyzed with respect to the determination of a state/mode transition decision:

(1) Pixel processing load

(2) Screen resolution

(3) Depth complexity of the scene

(4) Polygon count

(5) Video-memory usage

(6) Frame/second rate

(7) Change of fames/second rate

(8) Tolerance of latency

(9) Use of the same FB in successive frame, and

(10) User-System Interaction during the running of said graphics-based application.

316. The method of claim 306, wherein transition from object division to image division mode follows a combination of one or more of the following conditions:

Increase in pixel processing load

Increase in screen resolution

Increase in scene depth complexity

Decrease in polygon count

317. The method of claim 306, wherein transition from image division to object division mode follows a combination of one or more of the following conditions:

Increase of polygon count

Increase of video memory footprint

Decrease of scene depth complexity

318. The method of claim 306, wherein transition from object division to time division mode follows a combination of one or more of the following conditions:

Demand for higher frame/second rate

Higher latency is tolerated

There is no use of the FB for successive frame

319. The method of claim 306, wherein transition from time division to object division mode follows a combination of one or more of the following conditions:

Latency is not tolerable

FB is used for successive frame

High polygon count

320. The method of claim 306, wherein transition from time division to image division mode follows a combination of one or more of the following conditions:

Latency is not tolerable

FB is used for successive frame

High pixel processing load

321. The method of claim 306, wherein transition from image division to time division mode follows a combination of one or more of the following conditions:

Demand for higher frame/second rate

Latency is tolerable

High polygon count