Current Employer:

VMware, Inc., December 2008 to present.

Former Employers:

Address:

Steamboat Springs, Colorado, USA. A.k.a. "Ski Town USA" (a pretty good place for mountain biking too!)

Education:

Master's Degree, Computer Science, University of Wisconsin - Madison, 1994
Bachelor's Degree, Computer Science, University of Wisconsin - Oshkosh, 1990
High School, Stratford High School, Stratford, WI, 1986

Email:

brian.e.paul gmail.com

Software I've written or contributed to:

• Mesa - a free OpenGL work-alike library (plus, the new Gallium3D infrastructure)
• VMware Workstation, Fusion, ESX products
• Chromium - contributor to the Chromium project
• DRI - contributor to the Direct Rendering Infrastructure in XFree86
• VNC Proxy / Renderserver - remote 3D graphics over VNC
• Blockbuster - a high-res movie player for scientific visualization applications.
• Glean - contributor to Allen Akin's Glean (OpenGL validation) project
• Togl - an OpenGL widget for Tcl/Tk, with Ben Bederson
• Vis5D - visualization system, with Bill Hibbard
• TR - tile rendering library for OpenGL
• V-Blocks - virtual building blocks (very old, requires IRIS GL)
• Avid Marquee - 3D text, graphics, video animation

Other stuff I've done:

• OpenGL ideas - a random collection of ideas for new OpenGL features.
• OpenGL ARB - I've been involved in the OpenGL ARB for several years now and have contributed to a number of efforts including OpenGL 1.3, 1.4, 1.5, 2.0 and many extensions.
• Course notes for "OpenGL and Window System Integration" course at SIGGRAPH '97

Graphics Computers I've Used:

• Radio Shack TRS-80 Model 3/4 - 1982 to 1986
  48KB RAM, 1MHz Z80, 180KB floppy disk, TRS-DOS, 128x48 monochrome display.
  This was my very first computer in high school. I taught myself BASIC and Pascal on this system.
  One TRS-80 in the school lab had the high-res graphics option (640x240). I wrote a simple drawing program for it in BASIC.
• Apple II+ - 1984 to 1986
  48KB RAM, 1 MHz 6502, two floppy drives, monochrome display.
  I wrote a CNC milling machine simulator for this computer, among other things.
• Atari 800XL - 1983 to 1985
  64KB RAM, 1.7 MHz 6502, cassette tape storage, up to 320x192 pixels monochrome, more colors at lower resolutions. 256 colors max.
  This was my first home computer. Did lots of graphics programming with the Action! language (without floating point).
• Atari 130XE - 1985 to 1987
  128KB RAM, 1.7 MHz 6502, 180KB floppy disk.
  I wrote my first ray-tracer, triangle renderer and 3-D modelling software on this system in BASIC while in college. A 10-frame, 160x192 pixel, ray-traced animation took two days to render!
• Amiga 500 - 1987 to 1991
  1MB RAM, 7.8 MHz 68000, two 800KB floppy disks, up to 640x400 display, 4096-color palette. I wrote various ray tracers and scan-line rasterizers on this system in Modula-2.
• Pixar Image Computer 2 - 1989 to 1990
  1280x1024x36-bit display. Sun 3/150 front-end: 16MHz 68020, 16MB RAM, 80MB disk, monochome display. For my senior computer science project in college I wrote a 3-D modeller for this system in C. It generated scene files which were rendered with ChapReyes, the predecessor to Renderman.
• Sun SPARCstation 2 - 1989 to 1990
  The computer science department got a few of these workstations during my senior year of college. I typed in a ray tracer from a program listing in a magazine and rigged up a display system that converted the 24-bit image files to an 8-bit form suitable for the SPARCstation display.
• Stardent GS2000 (aka Stellar ST2000) - 1991 to 1994
  4-stream CPU, 20 MIPS, 80 MFLOPS vector processor, 64MB RAM, three 60MB hard disks, 1280x1024x24-bit display, 120Ktri/sec, XFDI graphics library (C/Fortran).
  I primarily helped develop Vis5D on this system.
• Amiga 3000 - 1991 to 1995
  3MB RAM, 16 MHz 68030, 10MB disk. The first bits of Mesa were written on this computer. Used HAM mode to approximate a normal RGB (non-indexed) framebuffer.
• SGI 340VGX - 1992 to 1995
  4x33MHz MIPS R3000, 64MB RAM, 1280x1024x24-bit display 1M shaded tri/sec.
  Helped develop Vis5D and VisAD on this system using IRIS GL. A really fantastic system for its day.
  Also, wrote a subset Ada compiler for this system when in graduate school. It generated real MIPS code.
• Assorted workstations - 1992 to 1997
  At the SSEC I programmed various graphics workstations from IBM, Sun, HP and DEC using X, IRIS GL, PEX, Starbase, etc. Nowadays it's all PCs. Back then there was a lot more variety in computers and I liked that. Each new system was a new adventure.
• Pentium PC - 1995 to 1998
  100MHz, 16MB RAM, 1GB disk, 1024x768x8-bit Cirrus Logic graphics card.
  This was the first PC I bought for myself (around $2500 IIRC).
• SGI Onyx Reality Engine 2 - 1995 to 1997
  4x150MHz MIPS R4400, 128MB RAM, 1280x1024x24-bit display 1.8M textured tri/sec
  Another really great SGI system.
• SGI (assorted) - 1994 to 1999
  Indigo, Indigo2, O2, Indy, Crimson, Infinite Reality, etc.
  I did a lot of IRIS GL and OpenGL programming on these systems.
• Generic PCs - 1999 to today
  Various intel CPUs from 100Mhz to multi GHz. 512MB to GBs of RAM. Multi-gigabytes of disk space. 1024x768x8-bit displays or better. Assorted consumer and pro graphics cards.
• More PCs - circa 2011
  8-core Intel CPU, 12GB RAM, 3TB disk, dual 1920x1200 displays, NVIDIA GTX-280 graphics card.

Other Unusual or Old Computers I've Used and Programmed:

• PDP-11 - 1988-1990
  The UW Oshkosh was still operating a PDP-11 (or some variant) back then. It was used for the "Unix" classes. Most of the terminals were VT-220 (text-only) displays but there was one graphics display that supported 1024x768 by 4 (or 16?) colors. For my graphics class I implemented a rasterizer that used Warnock's algorithm for hidden surface removal (no Z buffer) which used that graphics terminal.
• Cromemco - Summer 1989
  While still in high-school I took a college-level summer class in computer programming at the local branch of the UW system. We learned Pascal programming (though I had already used Pascal in high school) on some kind of Cromemco system. It was a basically a minicomputer with about a dozen text terminals.
• Cray 2, Cray X-MP, Cray Y-MP - Early 1990s
  I did some visualization-related work on Cray systems while I worked at the SSEC. The Crays were located at the NCSA and other sites. Distributed Vis5D used a Cray for data set hosting and isosurface construction. Isosurfaces were sent by network to an SGI workstation where they were rendered. The Crays ran the batch-oriented Unicos OS. Despite being the "fastest" computers around, they were pretty pokey when using the command line.
• Thinking Machines CM-5 - Early 1990s
  The University of Wisconsin computer science department had a CM-5 in the early 1990s. I briefly used the machine to do some experiments with distributed visualization, like the Cray systems. To take advantage of the CM-5 vector processors you had to program in Fortran or a variant of C called C* (C-star). It used a weird array prefix notation for parallel/vectorizable arrays. Ex: float x = [i][j]array;
• Intel Touchstone Delta - Early 1990s
  I briefly used this machine during graduate school. I wrote/ported a simple parallel ray tracer to this machine.
• X Terminals - 1991 to 1995
  X Terminals were a step up from text terminals. They simply ran an X server which received X protocol from apps running on other hosts. They were an inexpensive way to let several people share an expensive workstation. I especially liked the 21" 1280x1024 monochrome terminal that I used for a while at the SSEC. The black and white display was great for developing code. I implemented monochrome X visual support in Mesa so that I could display OpenGL on that terminal. The pinnacle of that was running SGI's Performer Town demo in dithered black and white on that screen.
• Sony/IBM Cell processor - 2007 to 2008
  Not rare or old, but not too many people have programmed the Cell directly. I wrote most of the Mesa/gallium "cell" driver. Fragment shaders were translated into SPU machine code. Triangle rasterization, texture sampling, blending and depth testing were also performed with SPU code. Until the llvmpipe driver came about, it was the fastest Mesa/gallium software rasterizer in terms of fragment processing speed.