Specifications

Fabric Density

Up to 138K LUTs (GW5A(S)(T)-138); up to 23K LUTs (GW5A-25)

Family Variants

GW5A (traditional FPGA), GW5AT (transceivers), GW5AR (integrated RAM), GW5AS (embedded MCU), GW5AST (RISC-V AE350), GW5ART (transceivers), GW5A/GW5AT automotive grades

Block SRAM

Single-port, semi-dual-port, true dual-port and semi-dual-port modes with byte write enable

ECC Support

Error detection and correction in block SRAM plus configuration memory soft error recovery (CMSER)

DSP Blocks

12x12, 27x36 and 27x18-bit multiplier modes with a 48-bit accumulator

DSP Features

DSP cascading, embedded pipeline and bypass registers, pre-addition for filter functions, internal feedback loop and barrel shifter

Clocking

Up to 16 global clocks, up to 6/12 high-performance PLLs, up to 16/24 high-speed edge clocks

GPIO Drive Strength

Selectable 4 mA, 8 mA, 12 mA, 16 mA and 24 mA drive

GPIO Features

Bus keeper, pull-up/pull-down, open drain, hot socket, input hysteresis and slew-rate control

On-Chip ADC

60 dB SNR with 1 kHz signal bandwidth, flexible X-channel oversampling, no external voltage source required

SerDes (GW5AT/GW5AST)

270 Mbps to 12.5 Gbps operation with CDR, dedicated RX and TX channels, integrated 8b/10b encoder/decoder

PCIe Hardcore

PCIe 2.0 hardcore with x1, x2, x4 and x8 lane configurations, root complex and endpoint support

Hard Processor

GW5AST devices integrate a hardcore RISC-V AE350_SOC processor

MIPI D-PHY RX Hard PHY

20 Gbps D-PHY RX hard PHY with 8 data lanes plus 2 clock lanes at 2.5 Gbps per lane (GW5A(S)(T)-138)

MIPI C-PHY

MIPI C-PHY RX/TX hardcore, one quad supporting up to three three-wire data lanes at up to 2.5 Gsps per lane

MIPI Link Layers

Built-in MIPI DSI and MIPI CSI-2 RX link layer IPs with SoT HS-Sync, word and lane alignment

Memory Interface

DDR3 memory interfacing built into the Arora V fabric

Configuration

JTAG, SSPI, MSPI, SERIAL and CPU configuration with background programming, SPI flash programming/boot and multi-boot

Security

Bitstream encryption and security with OTP and a unique 64-bit DNA identifier per device

Development Tools

GOWIN EDA — synthesis, placement and routing, bitstream generation and download, power analysis and in-device logic analyzer

Overview

The GOWIN Arora V series is a SRAM-based FPGA line that increases logic resources, interface bandwidth and performance while keeping power consumption low. Devices span from the 23K-LUT GW5A-25 up to the 138K-LUT GW5A(S)(T)-138, with DDR3 interfacing, 12.5 Gbps CDR-based SerDes supporting multiple protocols and flexible packaging options.

Because the family shares a common EDA flow and IP ecosystem, designs can be scaled from low-cost sensor-bridge roles up to server and communications line cards that combine 12.5 Gbps serial links, PCIe 2.0 hardcore endpoints and MIPI D-PHY/C-PHY camera interfaces on one device. The GW5AST variant adds a hardened RISC-V AE350_SOC for control-plane work alongside the programmable fabric.

For edge AI and machine-vision designs the Arora V family pairs high DSP-to-logic ratios with a hard MIPI receive PHY and PCIe endpoint, which is what makes it suitable as a real-time sensor-fusion, video-preprocessing or accelerator-glue device in front of a host SoC or a discrete AI accelerator.

Key Benefits

High performance per watt with SRAM-based fabric, dynamic power management and a 12.5 Gbps SerDes that keeps interface bandwidth high without a separate PHY. Integration of PCIe 2.0 hardcore, MIPI D-PHY/C-PHY hard PHYs, ADC and DDR3 reduces external component count. Design portability across a single EDA flow from 23K to 138K LUTs protects the investment as requirements grow, and security features such as bitstream encryption, OTP and a unique 64-bit DNA identifier support anti-cloning and device authentication.

Applications

Communications line cards and small cells, server and data-centre bridging and interface expansion, machine-vision and industrial imaging pipelines, automotive camera and display aggregation, medical imaging front-ends, LED display control, and real-time AI edge preprocessing where deterministic latency matters more than raw TOPS.

Request a Quote — GOWIN ARORA V FPGA (GW5A / GW5AT / GW5AR) — 138K-LUT SRAM FPGA WITH 12.5 GBPS SERDES

QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.

Get Your Quote →

Related Products

GOWIN LittleBee GW2A-18 Lattice ECP5 LFE5U-45F Microchip PolarFire SoC MPFS250T Efinix Titanium Edge FPGA AMD Versal AI Core XCVC1902 AMD Xilinx Artix-7 XC7A100T Intel Cyclone 10 LP 10CL025