Framing Efficiency - 12.0 English

Aurora 64B/66B LogiCORE IP Product Guide (PG074)

Document ID
PG074
Release Date
2023-11-17
Version
12.0 English

There are two factors that affect framing efficiency in the Aurora 64B/66B core:

  1. The size of the frame
  2. A data invalid request from the gearbox occurs after every 32 user_clk(txusrclk2) cycles

The gearbox in GTX and GTH transceivers requires a periodic pause to account for the clock divider ratio and 64B/66B encoding. This appears as back pressure in the AXI4-Stream interface and the user data needs to be stopped for one cycle after every 32 cycles (as shown in the following figure). The s_axi_tx_tready signal in the user interface from the Aurora 64B/66B core is deasserted for one cycle, once after every 32 cycles. The pause cycle is used to compensate the gearbox for the 64B/66B encoding.

Figure 1. Framing Efficiency

For more information on gearbox pause in GTX/GTH/GTY transceivers, see the 7 Series FPGAs GTX/GTH Transceivers User Guide (UG476) or the UltraScale Architecture GTH Transceivers User Guide (UG576), or the UltraScale Architecture GTY Transceivers User Guide (UG578), when applicable.

The Aurora 64B/66B core implements the Strict Aligned option of the Aurora 64B/66B protocol. No data blocks are placed after idle blocks or SEP blocks on a given cycle. The following table is an example calculated after including overhead for clock compensation (CC sequence consisting of a maximum of eight CC characters is sent every 4,992 user_clk cycles) and shows the efficiency for a single-lane channel while illustrating that the efficiency increases as frame length increases.

Table 1. Framing Efficiency Example
User Data Bytes Percent Framing Efficiency
100 96.12
1,000 99.18
10,000 99.89

The following table shows the overhead in a single-lane channel when transmitting 256 bytes of frame data. The resulting data unit is 264 bytes long due to the end-of-frame SEP block. This results in a 3.03% transmitter overhead. Also, the clock compensation blocks must be transmitted for at least three cycles every 10,000 cycles resulting in an additional 0.03% overhead in the transmitter.

Table 2. Typical Overhead for Transmitting 256 Data Bytes
Lane Clock Function
[D0:D7] 1 Channel frame data
[D8:D15] 2 Channel frame data

.

.

.

[D248:D255] 32 Channel frame data
Control block 33 SEP0 block