Using Super Sample Rate (SSR) Blocks in Model Composer - 2022.1 English

Vitis Model Composer User Guide (UG1483)

Document ID
UG1483
Release Date
2022-05-26
Version
2022.1 English

While the Super Sample Rate (SSR) feature introduced in this section is applicable to all Xilinx® devices, this section refers specifically to Xilinx RFSoC devices. The integration of direct RF-sampling data converters with Xilinx’s technology offers the most flexible, smallest footprint, and lowest power solution for a wide range of high performance RF applications such as Wireless communications, cable access, test & measurement, and radar. RFSoC devices provide hardened Digital Up Converters (DUC) and Digital Down Converters (DDC). NCO, Complex Mixers, and Filters are hard Macros, and filter characteristics are optimized for general Commercial applications.

Figure 1. RFSoC Device

Depending on what is needed, RFSoC devices can be used in two ways.

  • Use the available hardened NCO & Complex Mix and Half Band Decimation/interpolation filters.
  • If the sequence of the hardened blocks does not meet the design requirement, you can bypass them as shown in the figure above.

In the latter case, to meet the design requirements, you may need to implement the NCO, Complex Mixers and DDC blocks in the fabric using the HDL Blockset in Model Composer. To do this, bypass the hardened blocks, and let Model Composer IPs run at Programmable Logic (PL) clock frequency. When the sample rate from the ADC is in GSPS, and PL handles only the MSPS range of data, you must accept and compute multiple parallel samples every clock cycle for each data channel. The number of parallel samples is determined by calculating the ratio between the sample frequency and the Programmable Logic clock frequency, which is defined as an SSR parameter.

What is SSR?

SSR is a parameter that determines how many parallel samples to accept for every clock cycle.

How SSR helps users?

  • SSR is beneficial for users who cannot use the hardened RFSoC DUC and DDCs.
  • SSR provides programmatic subsystems for NCO and Complex Mixer among many others. The user input parameters in the block mask and Model Composer programmatically construct the underlying subsystem with multiple DDS blocks.
  • SSR avoids manual and structural modifications to your design, which accelerates the design-cycle.

SSR Library

Model Composer provides a separate set of library blocks for handling SSR. Currently, Model Composer supports 25 vector blocks, which can be accessed from the MATLAB® Library Browser.

Figure 2. SSR Block set in HDL Library

The SSR parameter can be defined for all the blocks present in the SSR block set. When you add a block from the library, the default SSR value is 4, and the maximum SSR valiue is 256.

The SSR block set is defined in Xilinx SSR Blockset.

Figure 3. Default SSR Value

No matter what the SSR rate is, you only need to provide a limited number of signal connections as with a normal IP block. Model Composer automatically takes care of all the parallel path connection internal to the SSR block, according to the SSR parameter value provided.

For example, for a Vector AddSub block, when SSR parameter is modified to 3, the internal connections are done automatically as shown below. This creates 3 parallel paths for computation and results in single output.

Figure 4. Vector AddSub Fabric Example