file: math_486msdos.timings
qv: math_386msdos.timings
I have tested the Mathematica benchmark on MS-DOS 386(33MHz with 387) and486(25MHz) computers (Not IBM PS/2's). Both of them have 8Mb RAM and VGAgraphics. I found that they are much faster than Mac IIci and similar with68040 NeXTstation and DECStation 3100. There are many other 386's and486's with different performance even with the same clock speed. Some33MHz 486 benches at more than 20 Mips and they will be much faster thanours. I also tested a 386 computer with 4Mb of RAM. It was much slower. I wonder if the Mac IIci tested was equipped with 8Mb RAM. If it had only4Mb, it was not a fair comparison. The results are following. I changed the first sentence to indicated correct machine.
Sang-il Park (spark@sierra.stanford.edu)
=======================================================================
This benchmark were performed on a 25MHz 486 MS-DOS computer
with 8Mbyte RAM and VGA graphics.
The following set of benchmarks are those used by
Wolfram Research, Inc. for timing tests.
Factor[Expand[(1+x)^100]]
{3.18 Second, Null}
Factor[Expand[(1+x)^20]]
{0.22 Second, Null}
1989^1989
{1.6 Second, Null}
Eigenvalues[Table[Random[], {40}, {40} ]]
{1.97 Second, Null}
<13, WorkingPrecision->26]
{0.22 Second, Null}
FindRoot[{f[x, y, z] == 7, g[x, y, z] == -1, h[x, y, z] == 5},
{x, 0}, {y, 0}, {z, 1}, AccuracyGoal->13, WorkingPrecision->26]
{0.28 Second, Null}
Clear[h];
Clear[g];
Clear[f];
f20[x_] := N[Sum[Sin[(2 k - 1) x]/(2 k - 1), {k, 20}]]
Plot[f20[x], {x, 0, Pi/2}]
{23.13 Second, -Graphics-}
NIntegrate[f20[x], {x, 1, 2}, AccuracyGoal->10, WorkingPrecision->20];
{11.87 Second, Null}
N[Integrate[f20[x], {x, 1, 2}], 10];
{1.76 Second, Null}
-FindMinimum[-f20[x], {x, 0, 0, Pi/2}];
{0.83 Second, Null}
Clear[f20];
f[x_, n_Integer] := N[Sum[Sin[(2 k - 1) x]/(2 k - 1), {k, n}]]
Plot[{f[x, 8], f[x, 9], f[x, 10]}, {x, 0, Pi}]
{29.44 Second, -Graphics-}
f[1, 500];
{1.48 Second, Null}
f[1, 10000];
{31.7 Second, Null}
FactorInteger[266382004787];
{0., Null}
Clear[f];
<True]
{0., Null}
f[t_] := Integrate[Exp[x^2], {x, -1, t}]
Plot[f[y], {y, -1, 2}]
{19.83 Second, -Graphics-}
Clear[f];
tmp = Integrate[(1 + x)/(1 + x^2 + x^4 + x^6), x];
{0.38 Second, Null}
tmp = D[tmp, x];
{0.06 Second, Null}
Simplify[ tmp ];
{0.71 Second, Null}
y = Series[f[x], {x, 0, 6}]; y = y /. {f[0] -> 0, f'[0] -> 1,\
> Derivative[n_][f][0] -> a[n]}; eqn = D[y, {x, 2}] + Series[Sin[x], {x, 0,\
> 6}] y == 0; var = Table[a[k], {k, 2, 6}]; Solve[eqn, var];
{0.06 Second, Null}
y =. eqn =. var =.
The following set of benchmarks are those from the
MatLab benchmark suite.
a = Table[Random[], {50}, {50}];
printIt[a.a;]
{0.44 Second, Null}
a = Table[Random[], {50}, {50}];
Inverse[a];
{2.8 Second, Null}
a = Table[Random[], {25}, {25}];
Eigenvalues[a];
{0.27 Second, Null}
a = Table[Random[], {i, 4096}] + I;
Fourier[a];
{1.27 Second, Null}
a = Table[Random[], {100}, {100}];
b = Table[Random[], {100}];
x = LinearSolve[a, b];
{27.02 Second, Null}
a = Table[0, {i, 1000}];
Do[a[[i]] = 1, {i, 1000}];
{0.44 Second, Null}
a = IdentityMatrix[25];
ListPlot3D[a];
{0.06 Second, Null}
This concludes the benchmark tests that are
included in this package.