System Benchmark Functions
v3SpeedFloat
Syntax
#include "speed.h"
u64 v3SpeedFloat( void );
Description
Benchmark IEEE floating point unit performance using 64 bit math.
The test does a matrix multiply spanning ~96KB of data, and thus should
fit in the CPU cache. This function takes several seconds to run, and
the time to run and results vary slightly by CPU/OS/compiler.
Return Values
Relative floating point performance.
1 is roughly 0.5 MFLOPS + overhead.
Errors
None.
Example
u64 fspeed;
fspeed = v3SpeedFloat();
v3SpeedInt
Syntax
#include "speed.h"
u64 v3SpeedInt( void );
Description
Benchmark integet unit performance using 32 bit integers. The test does
a matrix multiply spanning ~48KB of data, and thus should fit
in the CPU cache. This function takes several seconds to run, and
the time to run and results vary slightly by CPU/OS/compiler.
Return Values
Relative integer performance.
1 is roughly 0.5 MIOPS + overhead.
Errors
None.
Example
u64 ispeed;
ispeed = v3SpeedInt();
© Copyright Mithral Communications & Design Inc.
1995-2003.
All rights reserved.
Mithral® and Cosm® are trademarks of
Mithral Communications & Design Inc.
Document last modified: May 22, 2003
|