Description
AXVU13F is a flagship Heterogeneous Computing offering with XCVU13P FPGA, 4x HPC FMC+, 1x SODIMM, 1x 100G QSFP28+, and NVIDIA Jetson Orin NX module enabling high signal processing bandwidth, rich expandability. Ideal for data center AI/HPC acceleration, 100G+ Network, high-end ASIC/SoC prototyping and verification, radar/communication real-time signal processing.
FPGA Specification
| FPGA | XCVU13P-2FHGB2104I |
| Speed Grade | -2 |
| Chip Level | Industrial Grade |
| Working Temperature | -40℃~85℃ |
| System Logic Cells (K) | 3,780 |
| CLB Flip-Flops (K) | 3,456 |
| CLB LUTs (K) | 1,728 |
| Max. Distributed RAM (Mb) | 48.3 |
| Total Block RAM (Mb) | 94.5 |
| UltraRAM (Mb) | 360.0 |
| DSP Slices | 12,288 |
| Peak INT8 DSP (TOP/s) | 38.3 |
| PCIe® Gen3 x16 | 4 |
| 150G Interlaken | 8 |
| 100G Ethernet w/ KR4 RS-FEC | 12 |
| GTY 28.21 Gb/s Transceivers | 76 |
| Max. Single-Ended HP I/Os | 702 |
Board Features
| DDR4 | 1x SODIMM socket supports up to 32GB, data width 72bit. |
| QSPI Flash | 2x 128MB, for FPGA bin file and user data storage |
| Crystal Oscillator | 2x 200MHz Differential Crystal, Reference Clock for the FPGA and DDR4 1x high-precision, low-jitter programmable multi-channel clock source with 14 clock channels, reference for GTY. |
| NVIDIA Jetson Orin Sockets | 1x DDR4 SO-DIMM SOCKET 260 PIN, supports 1x PCIE3.0 x4 |
| FMC+ | 4x FMC+, compatible with FMC+ and FMC |
| Gigabit Ethernet | 1x 10/100/1000M Ethernet with RJ45 for data exchange, RGMII to FPGA |
| QSFP28 | 1x QSFP28 100G |
| SMA | 2x SMA Differential I/O, connected to FPGA HPIO, for customer differential signal I/O testing. |
| USB UART | for Serial Communication with PC/Devices |
| JTAG | 14-pin 2.00mm Standard JTAG port for Program Debug and Download |
| Expansion Port | 1x 14-pin 2.54mm port, connected to FPGA HPIO (6 pin), J18 |
| LED | 8x LEDs: 1 Power, 1 DONE, 4 User, 2 UART (TX/RX) |
| Push-Buttons | 1x: Reset |
Nvidia Jetson ORIN NX Features
| Model | Jetson Orin NX 8GB | Jetson Orin NX 16GB |
| AI Performance | 70 TOPS | 100 TOPS |
| Ampere GPU | 1024 NVIDIA® CUDA® cores | 32 Tensor cores | |
| GPU Maximum Operating Frequency | 765 MHz | 918 MHz |
| Arm Cortex-A78AE CPU | Six-core Cortex A78AE Armv8.2 (64-bit) heterogeneous multi-processing (HMP) CPU | Eight-core Cortex A78AE Armv8.2 (64-bit) heterogeneous multi-processing (HMP) CPU |
| CPU Maximum Operating Frequency | 2 GHz | |
| Deep Learning Accelerator | 1x NVDLA, 20 TOPs (Sparse INT8) | 2x NVDLA, 20 TOPS each (Sparse INT8) |
| DLA Maximum Operating Frequency | 610 MHz | 614 MHz |
| Programmable Vision Accelerator (PVA) | 1x v2 | |
| Memory | 8 GB 128-bit LPDDR5 DRAM | 16 GB 128-bit LPDDR5 DRAM |
| Video Decode | Standards supported: H.265 (HEVC), H.264, VP9, AV1 - 1x8K30 (H.265) - 2x4K60 (H.265) - 4x4K30 (H.265) - 9x1080p60 (H.265) - 18x1080p30 (H.265) |
|
| Video Encode | Standards supported: H.265 (HEVC), H.264, AV1 - 1x4K60 (H.265) - 3x4K30 (H.265) - 6x1080p60 (H.265) - 12x1080p30 (H.265) |
|
| Power | 10W | 15W | 20W | 10W | 15W | 25W |
| M.2 | 1x PCIE (x2) Gen4 M.2, for ORIN storage | |
| M.2 | 1x M.2 Key-E, for WiFi/BT module, interface includes PCIEx1, USB2.0, UART, I2S and I2C signals. | |
| DP | 1x DisplayPort Output, for ORIN module, eDP 1.4, VESA DisplayPort 1.4a HBR3 | Maximum Pixel Clock at 1080 MHz: up to 7680x4320 at 30 Hz | |
| USB 3.0 | 4x, Type-A, supports HOST mode 1x, Type-C, supports HOST, SLAVE, OTG modes |
|
| Gigabit Ethernet | 1x, connects ORIN MDI output directly to the RJ45 | |
Power Supply
Voltage Input: +12 V DC, PCIE 8-pin
Current Input: Max. Current 15 A
What's Included
| FPGA Board | 1 | DC Fan (installed) | 1 |
| Mini USB Cable | 1 | Programming Cable | 1 set |
| 12V Power Adapter | 1 | DDR4 | 1x 16GB |
| SMA cable | 4x | SMA Adapter (Angle) | 6x |
Dimensions
| Dimension | 295.00mm x 230.00mm |
| PCB | 20 Layer, Independent GND and Power Layers |
Resources & Downloads
Full documentation including Schematics, Demo projects (if applicable) will be provided by CodeRobin.