Mark K. Seager (03/05/91)
I have been looking at the XBench X11 benchmark and found the summary calculation for the "xStones" number quite mysterious. The "xstones.awk" file in the scripts directory uses various weights in conjunction with the results for various operations (text, rectangles, blit, etc.) to compute the "lineStones", "fillStones", "blitStones", "arcStones", "textStones", "complexStones" and, most importantly, "xStones". Does anyone have a clue as to how or why these weights were chosen? Is there some corelation between large weights and importance of various X11 operations? Secondly, some vendors are quoting a single "X11Perf" performance number. Here is a sample of such, given to me by a nameless vendor: IBM RS6000/320- 1,220; IBM RS6000/320C-1,220; DEC 5000-2,640; Dec 5000PX-2640; SparcStation 2-1,514; SparcStation 2GX-1,514. Does anyone know how these single "X11Perf" numbers are calculated? By the way, the documentation for X11Perf claims that such a number is meaningless! Regards, Mark
tohanson@gonzo.lerc.nasa.gov (Jeff Hanson) (03/05/91)
Mark asks > Does anyone have a clue as > to how or why these weights were chosen? Is there some corelation between > large weights and importance of various X11 operations? I believe the weights and measures is Xbench are from Sun 3/50 performance. Last year, prior to the creation of this group, there was much discussion in the trade papers about how terrible a benchmark Xbench is and that X11perf should be used instead. > Does anyone know how these single "X11Perf" numbers > are calculated? By the way, the documentation for X11Perf claims that such a > number is meaningless! I have no idea how single numbers are calculated but I recommend that anyone interested in X11perf based benchmarks look at Digital Reviews organization of X11perf. They produce a kiviat graph based on comparison with DecStation 3100 where the four corners are general graphics, window tasks, X specific operations and one I can't remember. I will write to Digital Review and ask them to put their organization of X11perf on expo. -- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* \ / \ / \ / \ / \ / \ / Jeff Hanson \ / \ / \ / \ / \ / \ / * ViSC: Better * tohanson@gonzo.lerc.nasa.gov * * * * * * / \ / \ Science / \ / \ NASA Lewis Research Center / \ / \ Through / \ / \ * * * * * * * Cleveland, Ohio 44135 * * * Pictures * * \ / \ / \ / \ Telephone - (216) 433-2284 Fax - (216) 433-2182 \ / \ / \ / *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*