Performance

Aerial SDK 22-4

The following performance achievement is possible with both BFP9 and BFP14, using either

  • 1+6 x86 cores with hyperthreading

  • 1+6 x86 cores without hyperthreading

4 4T4R peak / 8 4T4R avg cells respecting the following configuration:

Parameters

Peak Cell

Average Cell

Carrier BW (MHz) 100 100
PDSCH layers 4 2
PUSCH layers 4 2
PDSCH throughput (Gbps) 1.732 0.829
PUSCH throughput (Gbps) 0.241 0.123
PDSCH #UEG 6 6
PUSCH #UEG 6 6
PDSCH / PUSCH #PRB 252 / 258 132/132
UCI on PUSCH (bits) 4+6+5 4+6+5
PUCCH format 1 1
PDCCH #DCI 12 12
PUCCH payload (bits) 1 1
PUCCH #UCI 24 24
SSB slot 0 0
SSB instances per slot 1 1
PRACH format B4 B4
PRACH occasions 1 1
PRACH slots 5, 15 5, 15
PDSCH precoding Yes Yes
Previous Changelog
Next Supported Test Vector Configurations
© Copyright 2022, NVIDIA.. Last updated on Apr 20, 2024.