
P a g e | 6 Copyright © Guzik Technical Enterprises. All rights reserved.
CPU-based Processing
In addition to FPGA-based and GPU-based computation, customers
have an option to perform signal processing using a computer CPU.
Multi-core processing libraries, such as OpenMP, allow utilizing full
power of modern 12-core CPU computers. Once more powerful
computers with more cores are released, you can upgrade your
computer keeping your existing AXIe ADC 6000 Digitizer Module.
Ultra-fast GPU-based FFT Measurements
ADC 6000 performs frequency domain analysis using the Fast Fourier Transform (FFT)
performed on GPU. Single NVIDIA
®
Tesla GPU card performs FFT calculations at a
2.5 GSPS processing speed. This means, for example, that if you collect data at 10 GSPS
for 100 µs, process in 400 µs, you will get the full signal spectrum up to 5 GHz with
resolution bandwidth 10 kHz – 500,000 spectral lines – in less than 0.5 ms.
Expandability
The ADC 6000 AXIe-based module, features a real-time 40GByte/s link through standard
AXIe 62-pair Local Bus, which allows cascading multiple modules for increased memory,
processing speed, and faster data streaming to computer. This is a special order
configuration.
40 GSPS ADC
64 GBytes
RAM
…
Up to 14
modules in
1 AXIe
chassis
40 GSPS ADC
64 GBytes
RAM
40 GSPS ADC
64 GBytes
RAM
Up to 896 GBytes of memory is available for any channel in 14 slot AXIe chassis
Possible applications of the cascaded arrangement:
Increased processing speed – acquire data by one board, transfer to and process
by multiple boards, utilizing FPGA processors of several boards.
Increased memory – acquire data by one board, transfer to multiple boards, utilizing
memory of several boards. For example, combined memory of 14 boards is close to
1TByte.
Increased sampling rate – acquire data by two or more boards and interleave them.
Real-time data streaming – acquire data by one board, transfer to other boards, and
stream to an array of PCs in real time.
Comentários a estes Manuais