Re: [lmi] Problem of the week: testing a testing tool

[Greg Chicares]

On 2006-12-24 12:19 UTC, Greg Chicares wrote:
> 
> 0. What's obviously wrong here on the face of it?

Here's a different problem. Inline documentation says...

/// the operation is repeated for (0.1 * 'max_seconds', 'max_seconds']
/// (more or less, to the extent that the initial calibration trial's
/// speed was atypical).

It looks like a more sophisticated initial calibration would
be helpful. Here, 'max_seconds' is ten, but the first operation
timed takes only eight nanoseconds: twenty-five machine cycles.

Running vector_test:
  Speed tests: array length 1
  C:  [8.206e-009] 1000000 iterations took 8 milliseconds
  ET: [1.514e-008] 10000000 iterations took 151 milliseconds
  va: [1.661e-008] 10000000 iterations took 166 milliseconds

  Speed tests: array length 10
  C:  [1.921e-008] 10000000 iterations took 192 milliseconds
  ET: [2.727e-008] 10000000 iterations took 272 milliseconds
  va: [2.678e-008] 1000000 iterations took 26 milliseconds

  Speed tests: array length 100
  C:  [1.743e-007] 10000000 iterations took 1742 milliseconds
  ET: [1.576e-007] 1000000 iterations took 157 milliseconds
  va: [1.499e-007] 1000000 iterations took 149 milliseconds

  Speed tests: array length 1000
  C:  [1.783e-006] 1000000 iterations took 1782 milliseconds
  ET: [1.672e-006] 1000000 iterations took 1671 milliseconds
  va: [1.560e-006] 100000 iterations took 156 milliseconds

  Speed tests: array length 10000
  C:  [1.767e-005] 10000 iterations took 176 milliseconds
  ET: [1.630e-005] 100000 iterations took 1629 milliseconds
  va: [1.674e-005] 10000 iterations took 167 milliseconds

  Speed tests: array length 100000
  C:  [9.007e-004] 10000 iterations took 9006 milliseconds
  ET: [9.087e-004] 10000 iterations took 9086 milliseconds
  va: [8.892e-004] 10000 iterations took 8892 milliseconds