Function library speeds DSP programming tasks

A new and powerful library of floating-point DSP vectors and functions offers a broad range of callable functions.

Sundance Digital Processing has released the GDD600, a new and powerful library of floating-point DSP vectors and functions.

Offering a broad range of callable functions, the GDD600 library significantly reduces the development time of many DSP applications targeting Texas Instruments' (TI) TMS320 DSP-based platforms.

In addition to the hand-coded and optimised functions, the GDD600 includes a data conversion unit that facilitates the conversion of fixed-point and integer formats into floating-point units, as well as the conversion of floating-point units into integer formats.

"Encompassing over one hundred different functions and transforms, the GDD600 is remarkably beneficial for DSP applications written in either C-code or variants of C-code, enabling programmers to encode DSP and image-processing algorithms in an efficient and timely manner", said Thomas Brooks, C6000 Product Marketing Manager, TI.

With the ability to interchange between a fixed-point and a floating-point DSP processor, the GDD600 is a very helpful tool in the development of DSP real-time applications based on a variety of TI floating and fixed-point processors such as the TMS320C64xT and TMS320C67xT DSP generations.

"DSP applications are often coded modularly, which means the applications are divided into functions that are then called from the application".

"Most major DSP vendors have spent considerable time developing compilers that streamline these application developments", said Flemming Christensen, General Manager, Sundance.

"By providing application designers with the powerful GDD600 library of pre-optimised DSP functions, Sundance makes it possible for DSP programmers to maximise the benefits offered by compilers while reducing and simplifying application developments".

The GDD600 library comprises over 100 functions and macros that perform common DSP operations like fast Fourier transform, fast Hartley transform, discrete cosine transform, FIR/IIR filters, co-ordinate transformations, vector operations, complex number arithmetic operations, pseudorandom numbers generation and data conditioning (spectral windows) operations.

These operations are executed on the IEEE754 floating-point format numbers, which are then implemented on a larger dynamic range that uniformly distributes relative errors, making it unnecessary for developers to scale accumulators, a common practice for fixed-point arithmetic.

The hand-coded and optimised library includes various features such as interruptibility, which is essential for programming asynchronous systems.

The GDD600 also has the ability for re-entrance, thus enabling the same code to be shared between tasks and threads.

This makes the code useable in multitasking systems.

Fast and accurate floating-point math simulation routines also allow the user to benefit from a faster processing floating-point array and scalar data.

Pricing for the GDD600 starts at $5500.

Back to top