70 / (dots + 1) = double-buffered fps
The command I used to start the test was:
./sgixdoom -devparm -warp 1 5
There is a degree of subjectivity in guaging the number of dots (they flash on and off), but the results are reasonably consistent. I'm particularly pleased with the XL results which at last show just how good XL is for 2D tasks (the technical report says XL can do over 400M pixels/sec).
For comparison, SGI Doom at 3X scaling blats 19X more pixels to the frame buffer than PC Doom. SGIs have good 2D speed, giving excellent results for Doom. On the main Doom benchmarks page, a R4K/250 HighIMPACT Indigo2 easily beats a PC with a much faster CPU.
Here are the results, in order of 4X scaling. Remember that scaling does not add aditional detail, ie. running Doom at 640x400 on an O2 with the monitor set to 640x480 will look visually identical to running the game at 1280x800 on a MaxIMPACT with the monitor set at 1280x1024. Thus, make use of any available lower resolutions if that allows one to use a smaller scalaing factor to give better frame rate.
Key: 1X = 320x200 normal mode (rarely used; too small) 2X = 640x400 x2 scaling 3X = 960x600 x3 scaling 4X = 1280x800 x4 scaling 1X 2X 3X 4X ^^ ^^ ^^ ^^ Indigo2 R10K/175 MaxIMPACT: 35 35 35 28 Indigo2 R4K/250 XL: 35 35 27 24 Indigo2 R4K/250 HighIMPACT: 35 35 28 23 Indy R5000SC/180 24bit XL: 35 35 27 22 Indy R5000PC/150 24bit XL: 35 32 25 22 Indy R4400SC/200 24bit XL: 35 30 25 22 Indy R4600SC/133 24bit XL: 35 30 23 20 Indigo2 R4K/200 Elan: 35 30 23 18 Indy R4400SC/150 24bit XL: 35 28 21 18 Indy R4600PC/100 24bit XL: 30 24 18 16 Indigo2 R4K/250 Extreme: 35 25 15 12
I will add more results as and when I can, though the data at the moment show that pretty much any SGI can run Doom very playably at 3X scaling.
WARNING: Do NOT use these results to form judgements concerning 3D speed. Doom is not a 3D game. The graphics engine doesn't work that way.
Your feedback is welcome!