|
|
Scientific Computing Template Library |
|
|
|
|
|
SCTL build upon the APTL framework by
adding high-performance programming components specifically engineered
for computationally demanding applications.
Using finely tuned inline assembly coding within specialized templated
classes, SCTL provides high-performance processing speed combined with
general ease of use.
SCTL adds the following components to the APTL framework:
|
|
|
|
|
- High performance versions of the APTL Segment, Array,
and StaticVector classes for single and double-precision
floating-point values.
- SparseVector and SparseMatrix classes using a novel
auto-sorting hybrid list/tree structure to provide
dynamic element addition/removal during runtime.
- BlockMatrix class framework for large, dense matrix
operations to provide maximum performance
- Extensive use of expression templates to eliminate
temporary objects for maximum performance
- Automatic vectorization for arrays, vectors, &
matrices (optimized version only)
- SIMD (SSE/SSE2) assembly code expansion for built-in
data-types (optimized version only)
|
|
|
|
|
|
|
See this sample code for a simple example of
how SCTL can boost your applications performance while improving the
general readability.
Our distribution policy is very simple: maintain a freely downloadable
non-optimized version for testing purposes or general usage and charge
for the optimized versions. SCTL is uniquely
configured in that if the user requires the additional performance of
the processor-tuned optimized library, the speed boost is achieved by a
drop-in replacement library.
|
|
|
|
SCTL is currently only supported for use with the
GNU g++ compiler targeted for the
x86 architecture under Linux. The Win32 target is provided through the
use of a Linux-hosted cross-compiler with the
MinGW framework and is currently
included for alpha testing purposes only.
The library component of SCTL is available for following architectures:
- AMD™ Athlon™ XP/FX/64 (both 32-bit & 64-bit)
- Intel® Pentium® (3/4)
- Generic Pentium®-class processors
Any technical comments/inquiries are welcome to
send us an email.
|
|
top of page
|
|
