The XCKU035-1FBVA900C is a high-performance Xilinx FPGA from the Kintex® UltraScale™ family, manufactured by AMD (formerly Xilinx). Built on 20nm process technology, it delivers an outstanding balance of DSP processing power, logic density, and high-speed serial connectivity — making it one of the most cost-effective mid-range FPGAs available for demanding embedded applications.

---

What Is the XCKU035-1FBVA900C?

The XCKU035-1FBVA900C is a member of the Kintex UltraScale FPGA family. It features the FBVA900 package — a 900-pin Flip-Chip Ball Grid Array (FCBGA) with a 1.0mm ball pitch — and operates at speed grade -1, the standard commercial performance tier running at a core voltage (VCCINT) of 0.95V.

The part number decodes as follows:

---

XCKU035-1FBVA900C Key Specifications

General Specifications

---

Logic and Compute Resources

---

I/O and Connectivity

---

XCKU035-1FBVA900C Architecture Overview

## Kintex UltraScale FPGA Architecture

The XCKU035-1FBVA900C is built on Xilinx’s UltraScale architecture, which was designed to behave more like an ASIC than a traditional FPGA. Key architectural highlights include:

• ASIC-like clocking: Fine-grained clock gating reduces dynamic power consumption by up to 40% compared to previous-generation 28nm Kintex-7 devices.
• Configurable Logic Blocks (CLBs): Each CLB contains two slices with 6-input LUTs, flip-flops, carry logic, and shift register functionality. LUTs can also be configured as distributed RAM.
• DSP48E2 Slices: Each DSP slice features a 27×18 multiplier, a 30-bit A input, a 96-bit XOR function, pre-adder, and accumulator — ideal for FIR filters, FFTs, and machine learning inference workloads.
• 36Kb Block RAMs: True dual-port memory blocks with built-in FIFO controllers and ECC support, each configurable as two independent 18Kb blocks.
• GTH Transceivers: Next-generation serial transceivers support line rates from 500 Mb/s up to 16.3 Gb/s, enabling protocols such as 10GbE, PCIe Gen3, CPRI, JESD204B, and more.

---

### Clock Management

Each Clock Management Tile (CMT) in the XCKU035-1FBVA900C contains one MMCM (Mixed-Mode Clock Manager) and one PLL (Phase-Locked Loop). With 10 CMTs on-chip, designers have extensive flexibility for clock synthesis, phase alignment, deskewing, and frequency generation across diverse clock domains.

---

### PCIe Hard IP

The XCKU035-1FBVA900C includes a hardened PCIe Gen3 x8 block, eliminating the need to consume fabric logic for PCIe interfaces. This is critical for applications that require high-bandwidth host connectivity, such as FPGA acceleration cards, data acquisition systems, and high-speed data recorders.

---

XCKU035-1FBVA900C Package and Pinout Details

FBVA900 Package Information

The FBVA900 package is footprint-compatible with other Kintex UltraScale and Virtex UltraScale devices that share the same ball pattern, allowing PCB reuse and straightforward design migration between density tiers without a board respin.

---

Speed Grade and Operating Conditions

The -1 speed grade in the XCKU035-1FBVA900C represents the standard commercial performance tier. Kintex UltraScale devices are available in -3, -2, -1, and -1L grades, where -3 offers the highest clock speeds.

The -1 speed grade commercial device operates over a 0°C to +85°C junction temperature range, as indicated by the “C” suffix in the part number. DC and AC characteristics are identical between the commercial and industrial versions at the same speed grade.

---

Typical Applications for the XCKU035-1FBVA900C

The XCKU035-1FBVA900C is well-suited for a wide range of compute- and bandwidth-intensive embedded applications:

#### 100G Networking and Data Center Acceleration

With 20 GTH transceivers at up to 16.3 Gb/s and a hardened PCIe Gen3 x8 interface, the XCKU035-1FBVA900C is a strong fit for 100G packet processing, line cards, and FPGA-based SmartNICs. The high DSP count enables in-line encryption, compression, and protocol offload.

#### Wireless Infrastructure and CPRI/eCPRI

The Kintex UltraScale family was specifically optimized for heterogeneous wireless infrastructure including Remote Radio Heads (RRH) and DFE (Digital Front End) processing. The GTH transceivers natively support CPRI/eCPRI fronthaul links, and the 1,080 DSP slices handle multi-channel DPD (Digital Pre-Distortion), CFR (Crest Factor Reduction), and DUC/DDC operations.

#### Medical Imaging and Ultrasound Systems

DSP-intensive imaging pipelines such as CT reconstruction, MRI post-processing, and ultrasound beamforming benefit from the large block RAM pool and high-throughput DSP slices. The XCKU035 provides a significant density and performance step up from 7-series Kintex devices while maintaining cost competitiveness.

#### High-Resolution Video Processing (8K/4K)

The device’s logic density and I/O bandwidth support 8K/4K video pipeline designs, including frame buffering, color space conversion, scaling, and real-time encode/decode acceleration.

#### Test and Measurement Equipment

High pin count, GTH transceivers, and precise clock management make the XCKU035-1FBVA900C a popular choice for BERT systems, protocol analyzers, mixed-signal test instruments, and high-speed data acquisition cards.

#### FPGA-based Machine Learning Inference

With 1,080 DSP48E2 slices, the device can implement compact neural network inference accelerators for edge AI applications, particularly when paired with external HBM or DDR4 memory over the 468 HP I/Os.

---

Development Tools and Software Support

#### Vivado Design Suite

The XCKU035-1FBVA900C is fully supported by AMD’s Vivado Design Suite, which provides an integrated design environment for synthesis, implementation, simulation, and bitstream generation. Vivado’s co-optimization with the UltraScale architecture enables efficient design closure and timing convergence.

---

Ordering Information

Note: The XCKU035-1FBVA900C is classified as Non-Cancellable and Non-Returnable (NCNR) by most distributors due to market demand. Lead times may vary. Confirm availability with your authorized distributor before committing to a design.

---

XCKU035-1FBVA900C vs. Related Variants

All variants share the same silicon die; differences relate to package, I/O count, speed grade, and operating temperature. The FBVA900 package provides 468 user I/Os, which is optimal for applications requiring moderate I/O density without the cost of larger packages.

---

Frequently Asked Questions

What is the difference between XCKU035-1FBVA900C and XCKU035-1FBVA900I?

The only difference is the temperature grade: the “C” suffix indicates a commercial-grade device (0°C to +85°C junction temperature), while the “I” suffix indicates an industrial-grade device (-40°C to +100°C). Both devices have identical logic resources, speed grade, and silicon. The industrial variant is recommended for applications in harsh or outdoor environments.

Is the XCKU035-1FBVA900C supported by Vivado WebPACK?

No. Kintex UltraScale devices require the Vivado Design Edition or higher. The free WebPACK edition does not support the XCKU035 family. Evaluation boards such as the KCU105 come with a device-locked Vivado license for development purposes.

What protocols do the GTH transceivers support?

The 20 GTH transceivers in the XCKU035-1FBVA900C support a wide range of industry-standard protocols, including 10GbE, 25GbE (single-lane), PCIe Gen3, CPRI/eCPRI, JESD204B, DisplayPort, SATA/SAS, and custom serial interfaces up to 16.3 Gb/s.

Can I migrate from XCKU035 to Virtex UltraScale using the same PCB?

Yes, in many cases. Packages with the same last letter and number sequence (e.g., FBVA900) are footprint compatible across UltraScale architecture families. This allows upgrading to a higher-density or higher-performance device without a full PCB redesign, subject to I/O standard and power supply verification.