Hewlett-packard Company

Roger W Swanson

Robert H Miller, Jr

Anthony C Barkans

A technique of processing pipeline commands in parallel so as to minimize pipeline stalls. This is accomplished in accordance with the invention without need for the complex resource allocation techniques of the prior art by arbitrating access to critical pipeline resources on the phase of the system clock. For example, one control process may access the critical pipeline resource only during an even phase of the system clock, while a second control process may access the critical pipeline resource only during the odd phase of the clock. These processes may run at the same time if the pipelined instructions being executed by each process have no data dependencies since structural hazards are effectively eliminated by time-sharing the data buses on the respective phases of the system clock. The benefits of dynamically scheduled pipelined systems may thus be obtained without the complex scoreboarding and other scheduling algorithms used in the prior art to prevent pipeline hazards.

Allocation of resources of a pipelined processor by clock phase for

A pipelined processing system in which context switching for each of the pipelined processing circuits within the pipeline may be accomplished without flushing the data from the pipeline. This is accomplished by sending the pipeline commands and data together through the pipeline and differentiating the commands from the data using a flag added to the commands and data which specifies whether the associated data word is a command or data. During operation of the pipeline, when the input data is received by one of the pipelined processing circuits in the pipeline, the flag is checked to see if the associated data word includes a command. If the associated data word includes data to be processed, it is processed in accordance with the current configuration of the pipeline. However, if the associated data word includes a command for setup and control and the like, each pipelined processing circuit within the pipeline compares its identification value with a tag field in the command to determine whether it is to be reconfigured by that command. If it is to be reconfigured by that command, the appropriate context switching and the like takes place. However, if the current pipelined processing circuit is not to be reconfigured by that command, that command is passed through the current pipelined processing circuit unprocessed so that a similar determination may be made by the next pipelined processing circuit in the pipeline. As a result, setup and control commands for the pipelined processing circuits may be passed through the data processing pipeline along with the data in the desired processing order such that a pipeline data flush is not necessary between reconfigurations of the pipelined processing circuits. Since the pipeline need not be flushed when processes are changed, processing efficiency and throughput are substantially improved.

Processing commands and data in a common pipeline path in a high-speed

A method and apparatus for performing the majority of texture map gradient calculations once per polygon so as to increase processing speed in a graphics system. Texture values are identified for each vertex of an input polygon and are interpolated over the polygon in perspective space in order to find the corresponding values at a given pixel within the polygon. The perspective values of the vertices are linearly interpolated across the polygon to determine the value at the given pixel. The texture gradients are then calculated by defining vectors perpendicular and parallel to the horizon of the plane containing the input polygon so that the resulting components may be calculated. The resulting value is the texture gradient, which may then be used to address a texture map to determine the pre-filtered texture value for that point. A hardware implementation performs the necessary calculations for each pixel in the input polygon. The invention so arranged removes artifacts in the texture mapped image at a low cost and at a high speed.

Gradient calculation for texture mapping

A method and apparatus for interpolating pixels to be displayed on a display screen so as to account for the nonlinearity of distance changes in the perspective projection of a 3-D object onto the display for corresponding linear distance changes in 3-D world space. Each pixel of an input polygon to be displayed on the display screen is given a perspective value in world coordinates for each display point, and this value is passed through the graphics processor along with the shading parameters associated with the each display point. The respective shading parameters for each display point are then scaled by the perspective value for each display point to account for the effects of perspective foreshortening of the displayed object on the display screen. Since no translation to world coordinates is required for the perspective scaling, fast hardware circuitry may be used. The rendered image of the invention has much improved color accuracy since the shading more closely reflects the effects of changing perspective in world space.

Pixel interpolation in perspective space

A pipelined interpolator precomputes both integer and fractional portions of the slope of the function to be interpolated. The fractional portions of the starting value of the dependent variable to be incremented and the fractional value of the slope by which the increment occurs are each scaled to allow integer arithmetic. The scaled fractional portion of the starting value of the dependent variable is offset by a constant to allow carry-outs to be detected with the comparison 'greater than or equal to zero.' The occurrence of a carry-out causes the (unscaled) integer portion of the dependent variable to be incremented by the (unscaled) integer portion of the slope. A selected one of two pairs of simultaneous integer additions on the sets of the integer portion and the scaled fractional portion of the dependent variable will be performed. One pair is selected when the carry-out from the scaled fractional portion did not occur. The other pair is selected when it did. The values produced by the selected pair of additions are fed to the next stage in the pipeline and allowed to actually update the dependent variable's integer and scaled fractional portions.

Growing community of inventors

Fort Collins, CO, United States of America

Roger W Swanson