| Re: Q: IDL benchmarks [message #5896 is a reply to message #5827] |
Mon, 26 February 1996 00:00   |
Robert Moss
Messages: 74
Registered: February 1996
|
Member |
|
|
I just wanted to mention that before everyone runs out and buys PVWave
on an SGI they should look carefully at these numbers: they are for
much fewer iterations that the corresponding tests in IDL, so direct
comparison to most of the other recent postings without some scaling
is not valid.
Robert M. Moss, Ph.D.
Texaco Inc.
mossrm@texaco.com
Watson R J wrote:
>
> Running PV-WAVE CL Version 5.00 (sgi IRIX mipseb) on an SG Indy
> with 100MHz R4400 and 144Mb time_test gives this ...
>
> 1 0.156311 Empty For loop, 10,000 times
> 2 0.467805 Call empty procedure (1 param) 10000 times
> 3 0.142651 Add 10,000 integer scalars and store
> 4 0.241405 5000 scalar arithmetic loops
> 5 0.0721331 Mult 512 by 512 byte by constant and store
> 6 0.0324579 Shift 512 by 512 byte and store
> 7 0.0448910 Add a constant to a 512 by 512 byte array and store
> 8 0.0431681 Add two 512 by 512 byte images and store
> 9 0.318135 Mult 512 by 512 floating by constant and store
> 10 0.0612410 Shift 512 by 512 floating and store
> 11 0.0830290 Add a constant to a 512 by 512 floating array and store
> 12 0.326699 Add two 512 by 512 floating images and store
> 13 0.0674850 Invert a 50 by 50 random matrix
> 14 1.18785 Transpose 256 x 256 byte, FOR loops
> 15 0.133863 Transpose 256 x 256 byte, row and column ops
> 16 0.0236490 Transpose 256 x 256 byte, transpose
> 17 0.0209789 Log of 10000 numbers, vector ops
> 18 0.321045 Log of 10000 numbers, FOR loop
> 19 0.00391400 Add two 10000 element floating vectors, vector op
> 20 0.267292 Add two 10000 element floating vectors, FOR loop
> 21 0.0432030 8192 point real to complex FFT
> 22 0.372203 Smooth 512 by 512 byte array, 5x5 boxcar
> 23 0.256296 Smooth 512 by 512 floating array, 5x5 boxcar
> 24 0.186448 Write and read 10 512 by 512 byte arrays
> 4.87416 Total Time
> 0.10921606 Geometric mean, 24 tests.
>
> posted out of interest ...
> --
> ------------------------------------------------------------ -------------------
> Robert J Watson. "When in a hole... stop digging ..."
> University of Essex, Electronic Systems Engineering Dept.
> Wivenhoe Park, Internet: watsr@essex.ac.uk
> Colchester, Tel : +44 1206 873029 (direct)
> CO4 3SQ. Fax : +44 1206 872900
> ------------------------------------------------------------ -------------------
************************************************************ ***********
This is not necessarily the opinion of Texaco Inc.
************************************************************ ***********
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
