SINcore is the Verilog IP core of Matrix singular value and singular vector calculation .
In summary , Sincore is smart data process module to help customers process the data in a smart way .It can be embedded in the FPGA.
The computational complexity of matrix eigenvalues and eigenvectors is very large, so large size matrix are normally calculated on a PC (cluster, in parallel) or a mainframe now. In most cases, the calculation is performed offline. It is larger scale, high cost and it cannot be implanted into a micro-system. There is considerable market demand for matrix calculation on DSP-based embedded device.
SinCore implements single precision floating-point matrix singular values and singular vectors on Altera series FPGA . The resulting function occupies 30% of low end devices EP2C20Q240 for NX16 matrix size，where N is an integer less than or equal to 16.
- Bilateral Householder transformation.
- Bilateral Givens transformations of two pairs of diagonal matrix ,QR iteration
- Establishing structure of cumulative transformation matrix .
- The output of the result of the matrix singular values and left and right singular vectors to memory.
- Support of floating point format of standard IEEE754.
- Support of the single matrix factorization.
- Support of the CORDIC algorithm for solving the square root.
- Verilog HDL based design.
- Accuracy after four decimal points.
- Supports of all devices of Altera FPGA .
- The basic matrix size is M*16（M<=16）.
- Based on the customerís requirement, we can provide M*32（M<=32）,M*64（M<=64）size of matrix calculation cores.
- Smaller and faster speed.
- Itechland used unique smart algorithm to realized it. The two IP cores are convenient to use and portable. It also only occupy small resources which can be used in a small equipment and it can run in a micro-system.
- The main contribution of the module is the use of optimized algorithms in order to avoid matrix multiplication operation. It reduces logic occupation with improved calculation speed. The SinCore can be easily implemented on conventional FPGA devices, this module written in Verilog HDL.
- Reduced the signal to noise ratio. SNR 6 BT. It can detect weak signals.
- It can be used in lots of fields such as BSS control engineering, filtering，image treatment and compression and communication and medical engineering, channel estimation and etc . Matrix eigenvalue and eigenvector are an important and basic operation in DSP algorithms and implementations in many fields, including radar, sonar, and multiple antenna systems for communications, blind identification, automotive control. It forms the basic mathematics module which can be used in lots of different areas.