Exploring FPGA network on chip implementations across various application and network loads

Graham Schelle,D. Grunwald

DOI: https://doi.org/10.1109/FPL.2008.4629905

2008-09-23

Abstract:The network on chip will become a future general purpose interconnect for FPGAs much like todaypsilas standard OPB or PLB bus architectures. However, performance characteristics and reconfigurable logic resource utilization of different network on chip architectures vary greatly relative to bus architectures. Current mainstream FPGA parts only support very small network on chip topologies, due to the high resource utilization of virtual channel based implementations. This observation is reflected in related research where only modest 2times2 or 2times3 networks are demonstrated on FPGAs. Naively it would be assumed that these complex network on chip architectures would perform better than simplified implementations. We show this assumption to be incorrect under light network loading conditions across 3 separate application domains. Using statistical based network loading, a synthetic benchmarking application, a cryptographic accelerator, and a 802.11 transmitter are each demonstrated across network on chip architectures. From these experiments, it can be seen that network on chips with complex routing and switching functionality are still useful under high network loading conditions. Additionally, it is also shown for our network on chip implementations, a simple solution that uses 4-5times less logic resources can provide better network performance under certain conditions.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to design an efficient Imaging Fourier Transform Spectrometer (IFTS) for the Next Generation Space Telescope (NGST) in order to achieve simultaneous deep - field imaging and integral - field spectroscopy. Specifically, the paper mainly solves the following problems:

1. **Improving Observation Efficiency**:
- Traditional spectrometers require slits when performing multi - target spectroscopy, which limits the flexibility and efficiency of observation. However, the IFTS can obtain spectral information for each pixel in the field of view simultaneously, without the need to pre - select targets or be restricted by the slit geometry.
- The IFTS can provide flexible resolution options in a wide - band range (from \( R = 1 \) to \( R = 10^4 \)), and can record the spectral information of adjacent celestial bodies up to the confusion limit.

2. **Enhancing Signal - to - Noise Ratio (SNR)**:
- The IFTS transmits through the full - bandpass to the Focal Plane Array (FPA), thus obtaining the brightest image. This is very advantageous for wide - color - gamut imaging with high signal - to - noise ratio, and is suitable for scientific observations and telescope guiding.
- The design of the four - port Michelson interferometer enables all light to be collected, avoiding the 50% light loss problem in the traditional flat - mirror Michelson interferometer.

3. **Calibration and Stability**:
- The four - port Michelson interferometer has a natural advantage in calibration. By placing a calibration source at the second input end, a complete radiometric calibration can be carried out on the ground, ensuring the long - term stable operation of the instrument in space.
- The calibration source can comprehensively calibrate the offset, spectral response, and nonlinear characteristics of each FPA pixel by changing the temperature or intensity.

4. **Adapting to the Unique Requirements of NGST**:
- The NGST will be deployed at a relatively far distance from the Earth and be shielded to reduce background noise. The design of the IFIRS fully utilizes this environmental advantage and can perform high - sensitivity infrared astronomical observations under low - background - noise conditions.
- The optical layout of the IFIRS includes components such as correction lenses, interferometers, cameras, and detector arrays, ensuring optimal performance in the NGST's L2 orbit.

In summary, this paper aims to make the IFTS an indispensable tool in the NGST mission through design and optimization, thereby achieving important scientific results in multiple astronomical fields.