Harald J Smit, 584411 Spicewood Springs Rd, Austin, TX 78759

6389485, May 14, 2002

Oct 2, 1998

09/165084

Iliese Claire Chelstowski - Austin TX
Thomas Winters Fox - Georgetown TX
Bimal Poddar - El Dorado Hills CA
Harald Jean Smit - Austin TX

International Business Machines Corporation - Armonk NY

G06T 1550

709328, 345426

A graphics adapter capable of supporting lighting models from multiple Application Programming, Interfaces (APIs) within a graphics system is disclosed. The graphics adapter includes a dot product machine and a control means. The dot product machine can perform geometry computations regardless of the API format in which the graphics data is expressed. The control means is utilized to multiplex between a set of graphics data having a first API format and a set of graphics data having a second API format in order to determine the inputs of the dot product machine. As a result, lighting models from multiple APIs can be supported by a single graphics adapter.

6559856, May 6, 2003

Sep 9, 1999

09/392835

Gordon Clyde Fossum - Austin TX
Thomas Winters Fox - Georgetown TX
Bimal Poddar - El Dorado Hills CA
Harald Jean Smit - Austin TX

International Business Machines Corporation - Armonk NY

G09G 502

345600, 345426, 345503, 708204, 708208

Lighting parameters are received as floating-point numbers from a software application via an application programming interface (API). The floating-point numbers are converted to a fixed-point representation having a preselected number of bits. The number of bits is selected in accordance with a predetermined number of bits required by a frame buffer, which thus establishes the number of color values supported by the graphics display system. In order to preserve accuracy to within the number of bits in each value in the frame buffer, the representation in the fixed-point engine includes additional bits relative to the number of bits in the color values sent to the frame buffer. Floating-point values received via the graphics API are converted to fixed-point representations by first prescaling the floating-point values. The fixed-point operations generate a lighting signal value in accordance with a lighting model, and the number of bits therein is reduced to correspond to the number of bits in each data value required by the frame buffer, and the lighting data values are sent as color values to the frame buffer.

6677954, Jan 13, 2004

Nov 8, 2001

10/010469

Allen Jensen - Austin TX
Dale Kirkland - Madison AL
Harald Smit - Austin TX

3Dlabs, Inc., Ltd - Hamilton

G09G 539

345531, 345536

A method for caching graphics-related data in one or more graphics request buffers wherein duplicative graphics-related data is not written to the graphics request buffers. In the preferred method the graphics-related data is sent in frames, and each frame contains frame setup data and graphical model data, and the model data is compared between the stored frame and the new frame to determine if there is new model data to be written to the graphics request buffers.

5818469, Oct 6, 1998

Apr 10, 1997

8/827740

John Joseph Lawless - Round Rock TX
Bimal Poddar - Austin TX
Alice Elizabeth Putney - Round Rock TX
Harald Jean Smit - Austin TX

International Business Machines Corporation - Armonk NY

G06F 1500

345522

A method and implementing multiprocessor computer system 200 in which graphics applications 101 are executed in conjunction with a graphics interface 103 to graphics hardware 115. The methodology is also applicable to an implementing distributed network system. A master thread 105, or master node in a distributed network system, receives commands from a graphics application 101 and assembles 313 the commands into workgroups with an associated workgroup control block 315 and a synchronization tag 317. For each workgroup, the master thread flags changes in the associated workgroup control block. At the end of each workgroup, the master thread copies the changed attributes into the associated workgroup control block 319. The workgroup control blocks are scanned 403 by the rendering threads, or rendering node in a distributed network system, and unprocessed workgroups are locked 406, and the rendering threads attribute state is updated 413 from the previous workgroup control blocks. Once the rendering thread has updated its attributes, it has the necessary state to independently process the workgroup, thus allowing parallel execution.