Abstract: A system and method for performing speculative execution of state machine operation in a graphics accelerator. In accordance with one aspect of the invention, the method includes the step of executing steps in a first state machine that is operating on a graphic primitive. As is known, a graphic primitive is defined by a plurality of vertices. In accordance with the invention, the preferred embodiment receives the coordinate parameters for the second to last primitive vertex. Then it evaluates one or more conditions that indicate whether steps in a second state machine need to be executed, based upon parameters of primitive vertices already received. It then branches to and begins executing steps in another state machine, based upon the tentative conditions, and continuing execution of the steps in the transformation state machine in parallel with the continued execution of the steps in the another state machine.

Abstract: Apparatus for electrically verifying a functional unit contained within an integrated circuit comprises a functional unit, a state machine, a number of integrated circuit input pins, and means for alternately providing the functional unit with control data derived from the state machine, and control data derived from the number of integrated circuit input pins. The means for providing control data from alternating sources comprises a multiplexor which receives a first set of inputs from the state machine, and a second set of inputs from a test control block. The test control block monitors various of the integrated circuit input pins for a designated instruction, receives control data via the input pins, and controls the operation of the multiplexor. The test control block comprises a number of test registers which can be configured to receive two or more states of control data.

Abstract: An enhanced test system in a processor having a memory supporting multiple memory schemes. The memory is partitioned into memory blocks and memory sub-blocks. A plurality of uniform data units each comprising a plurality of data fields is written to and read from each successive memory block in a FIFO manner so that a data field within each data unit, having a maximum field width, occupies each of the multiple memory locations at least once during testing. The enhanced test system maximizes the number of adjacent full-width data fields to test vertically and horizontally for field overflow within memory by writing and reading seriatim by data unit or partitioned by data field width, in adjacent memory blocks and sub-blocks, or overlapping memory blocks and overlapping sub-blocks.

Abstract: A computer graphics system includes a plurality of geometry accelerators for processing vertex data representative of graphics primitives and providing rendering data. The system includes a distributor responsive to a stream of vertex data for distributing to the geometry accelerators chunks of the vertex data for processing by the geometry accelerators to provide chunks of rendering data. The distributor generates an end of chunk bit indicative of the end of each of the chunks of vertex data. The system further includes a concentrator for receiving the chunks of rendering data from each of the geometry accelerators and for combining the chunks of rendering data into a stream of rendering data in response to end of chunk bits. The stream of rendering data and the stream of vertex data represent sequences of graphics primitives having the same order. A rasterizer generates pixel data representative of a graphics display in response to the stream of rendering data.

Abstract: A geometry accelerator for a computer graphics system performs transformations on vertex data representative of graphics primitives, decomposes quadrilaterals into triangles, and performs lighting, clipping and plane equation calculations for each primitive. The geometry accelerator incorporates a memory mapping technique that achieves high efficiency transfer of vertex information from the host computer to the geometry accelerator. A double buffered vertex RAM with granularity permits the quantity of data transferred to the geometry accelerator to be reduced. The transformation and decomposition engines of the geometry accelerator employ data management techniques in which calculations for shared vertices and shared edges of primitives are performed only once.

Abstract: A minimal logic multiplexer system using tri-state drivers with one-hot enabling lead, provides high-speed access to processor elements by any one of a plurality of control units. The multiplexer system is implemented in a manner that minimizes the circuit implementation, minimizes gate delay within the circuit implementation, and allows processing instructions to pass from a control unit to the processor elements by way of multiplexed control lines therebetween. The multiplexer system contains control unit gate groups that are enabled and disabled in parallel by a select lead. Each control unit gate group can be implemented internal to the respective control unit or external in a common intermediary multiplexer circuit location.

Abstract: A geometry accelerator for a computer graphics system performs transformations on vertex data representative of graphics primitives, decomposes quadrilaterals into triangles, and performs lighting, clipping and plane equation calculations for each primitive. The geometry accelerator incorporates a memory mapping technique that achieves high efficiency transfer of vertex information from the host computer to the geometry accelerator. A double buffered vertex RAM with granularity permits the quantity of data transferred to the geometry accelerator to be reduced. The transformation and decomposition engines of the geometry accelerator employ data management techniques in which calculations for shared vertices and shared edges of primitives are performed only once.

Abstract: A hardware graphics accelerator accepts lists of polygon vertices from an application environment running application and systems graphics software. After a polygon is rotated and translated as needed, it is checked for trivial accept/reject against the clip limits of the viewing volume, but is not otherwise clipped. Polygons that are not rejected are decomposed into triangles before any other operations on them are performed. After decomposition the triangles are illuminated by light sources, if desired and then clipped by a triangle clipper, rasterized, and the results sent to a frame buffer for display. The triangle clipper incorporates trivial accept/reject operation, and is capable of operating on non-planar quadrilaterals. It avoids ugly artifacts during certain clip operations when the diagonal used to decompose a quadrilateral into triangles intersects a clip plane not parallel to the viewing axis.

Abstract: A geometry accelerator for a computer graphics system performs transformations on vertex data representative of graphics primitives, decomposes quadrilaterals into triangles, and performs lighting, clipping and plane equation calculations for each primitive. The geometry accelerator incorporates a memory mapping technique that achieves high efficiency transfer of vertex information from the host computer to the geometry accelerator. A double buffered vertex RAM with granularity permits the quantity of data transferred to the geometry accelerator to be reduced. The transformation and decomposition engines of the geometry accelerator employ data management techniques in which calculations for shared vertices and shared edges of primitives are performed only once.

Abstract: An enhanced performance queuing system that includes a FIFO queue in an electronic device, where the FIFO queue is controlled by a FIFO queue controller that also provides FIFO queue status relating to space available in the FIFO queue. A first device writes data to the FIFO queue in data chunks or in data item increments within a data chunk. FIFO queue status is requested only to determine if a data chunk sized space is free prior to writing to a data chunk space in the FIFO queue, rather than polling for FIFO queue status prior to each write operation. A second device reads data from the FIFO queue in data chunks or data item increments from within a data chunk. The first device begins writing to the FIFO queue prior to signaling the second device to begin reading, so that the second device can read from the FIFO queue without ever catching up with the first device thereby eliminating the need for requesting FIFO queue status to determine if data is available for reading.

Abstract: An enhanced test system in a processor having a memory supporting multiple memory schemes. The memory is partitioned into memory blocks and memory sub-blocks. A plurality of uniform data units each comprising a plurality of data fields is written to and read from each successive memory block in a FIFO manner so that a data field within each data unit, having a maximum field width, occupies each of the multiple memory locations at least once during testing. The enhanced test system maximizes the number of adjacent full-width data fields to test vertically and horizontally for field overflow within memory by writing and reading seriatim by data unit or partitioned by data field width, in adjacent memory blocks and sub-blocks, or overlapping memory blocks and overlapping sub-blocks.