From: broehl@sunee.uwaterloo.ca (Bernie Roehl) Subject: Re: HARDWARE: Sega HMD for December? Date: Mon, 17 May 1993 19:51:23 GMT Message-ID: Organization: University of Waterloo In article <1suio8INNap0@news.u.washington.edu> diego@minerva.st.dsi.unimi.it (Diego Montefusco) writes: >> Start-Fox-FX-killer game Silpheed which at some stages in the game >> redraws HALF A MILLION POLIGONS IN REAL TIME, yes, 500,000 poligons >> (Although NOT shaded, but simply flat poligon, still a good number for >> a 200 dollar system!!!). >Second I can't believe the CD-ROM can draw 500,000 polys in real time This again raises the question of performance measurement for VR systems. The most common way of measuring performance is in polygons per second. Of course, it's not always clear what's actually being measured; one common approach (and one with which I strongly disagree) is to take the number of polygons in the database and multiply by the maximum number of frames per second. Thus if the system described above were getting 25 frames per second in a world that has 20,000 polygons, they'd get the half-million polys per second figure that was mentioned above. My objection to this method is that it's not a true reflection of the performance of the system. The vast majority of those 20,000 polygons are not visible at any given instant (offscreen, backfacing, too far away to be seen, etc). And you can get *very* high frame rates just by looking at simple parts of the world (say, empty sky for example). Bernie Roehl University of Waterloo Dept of Electrical and Computer Engineering Mail: broehl@sunee.UWaterloo.ca Voice: (519) 885-1211 x 2607 [work]