Mathematical Acceleration Subsystem (MASS) consists of libraries of mathematical functions tuned specifically for optimum performance on a variety of IBM architectures.
Benefits
- Scalar and vector libraries are provided.
- Single instruction multiple data (SIMD) libraries are provided for SIMD architectures.
- These libraries and functions:
Are thread safe.
Are tuned specifically for optimum performance on the target architecture, offering improved performance over the corresponding standard system library routines.
Are callable from C, C++, and Fortran applications.
Are intended for applications where slight differences in accuracy or handling of exceptional values can be tolerated.
Features and benefits
| Features | Details | Benefits |
|---|---|---|
| Scalar libraries | The MASS scalar libraries contain an accelerated set of frequently-used single- and double-precision math intrinsic functions. | Provide speedups up to 8x over the standard system library. |
| Vector libraries | The MASS vector libraries provide increased performance by operating on a vector of arguments. They include single-precision, double-precision, and integer functions. | Provide speedups up to over 35x compared to the standard system library. |
| SIMD libraries | The MASS Single Instruction Multiple Data (SIMD) libraries are provided for SIMD architectures. They operate on SIMD vector arguments, and include single-precision, double-precision, and integer functions. | Provide speedups up to 80% compared to the standard system library. |
| Compatible processors | POWER, POWER2, POWER3, POWER4, POWER5, POWER6, POWER7, Blue Gene/L, Blue Gene/P, Cell Broadband Engine (PPU and SPU). | |
| Supported operating systems | AIX, Linux | |
| 32/64-bit mode support | 32-bit mode applications are supported, as well as 64-bit mode applications on processors supporting them. |