Hardware-efficient matrix inversion algorithm for complex adaptive systems

Design and implementation of high performance matrix inversion based on reconfigurable processor

“…In [9], a FPGA based efficient matrix inversion procedure is described using QR decomposition. The computation time of a 23 Â 23 matrix inversion on Xilinx Spartan3 XC3S1000 is 253 us.…”

“…The complexity of the direct analytic method increases exponentially with the size of matrix, so matrix decom-position becomes the most common method applied in matrix inversion with large dimensions, such as LU decomposition (with partial pivoting) [4,5,6,7], QR decomposition [8,9,10], Cholesky decomposition [11], etc. By analyzing these algorithms [12], LU decomposition (with partial pivoting) has better generality than Cholesky decomposition which only applies to symmetric positive definite matrices, and lower computational complexity than QR decomposition.…”

Matrix computing coprocessor for an embedded system

“…Although MCC consumes limits the usage of ASICs, and when ASICs are integrated to embedded computing systems to improve system performance, much time is consumed by the interaction between ASIC and other parts of the system; thus, performance improvement is limited. Moreover, the data types of the hardware in [4,7,12,19,8,15] are fixed-point or short floating-point, which indicate short computing times and minimal hardware resources. It can be seen from Table 12 that our widely used processor is more suitable for embedded systems.…”

“…It can be seen from Table 12 that our widely used processor is more suitable for embedded systems. MCC was also compared with the hardware introduced in [4,7,10,12,19,8,15]. To ensure objectivity, the power consumption of the processors in [12,15] was normalized to 0.18 lm.…”

“…A matrix inversion core generator tool, GUSTO, has been developed to ease the design space exploration across different matrix inversion architectures [7]. A matrix inversion architecture for embedded systems was proposed in [8]. For example, a 10 Â 10 matrix inversion was implemented in a Spartan3 device.…”