There’s a moment in every embedded design where you realize you need both a processor and an FPGA — and wiring them together on a PCB is going to be a nightmare. The Zynq-7000 was built for exactly that moment.
The AMD XC7Z035-2FBG676E fuses a dual-core ARM® Cortex™-A9 processor running at 800MHz with Kintex™-7 class FPGA fabric containing 275,000 logic cells — all on a single piece of silicon. No bus bottlenecks between processor and logic. No clock domain crossing headaches. No second power supply. Just one chip that does everything.
The Processing System: A Real Computer
This isn’t a soft-core MicroBlaze bolted onto fabric. The Zynq’s Processing System (PS) is a fully hardened ARM subsystem:
- Dual-core ARM Cortex-A9 @ 800MHz — Runs Linux, FreeRTOS, VxWorks, or bare-metal code
- NEON SIMD engine + VFPv3 FPU — Hardware-accelerated floating point and vector math
- 256KB on-chip SRAM + 512KB L2 cache — Low-latency memory for real-time critical paths
- Hardened DDR3/DDR3L controller — Up to 1066Mbps, no fabric resources consumed
- Full peripheral set — USB 2.0 (×2), Gigabit Ethernet (×2), SD/SDIO (×2), UART, CAN, I2C, SPI, GPIO — all hardened, all ready at boot
The PS boots independently from the PL. Your Linux kernel can be up and running in seconds while the FPGA fabric is still being configured. This is production-ready SoC design, not a science project.
The Programmable Logic: Kintex-Class Firepower
The FPGA side of the XC7Z035 is no afterthought. This is Kintex-7 grade fabric:
- 275,000 logic cells — Build entire hardware accelerators, custom peripherals, or complete datapaths
- 17,400 Kbits block RAM — Massive on-chip storage for frame buffers, DMA engines, and packet queues
- 900 DSP48E1 slices — Enough DSP muscle for real-time beamforming, neural network inference, or 4K video processing
- 8 GTX transceivers @ 12.5Gbps — Multi-gigabit serial I/O for 10GbE, PCIe Gen2, SATA III, JESD204B, and custom protocols
The Magic Is in the Middle
What makes Zynq fundamentally different from an “FPGA + processor” board? The AXI interconnect between PS and PL. The ARM cores can read and write to custom hardware blocks in the FPGA with single-cycle latency through dedicated high-performance ports. DMA engines can stream data between DDR memory and FPGA accelerators at multi-gigabyte-per-second throughput.
This means you can:
- Run computer vision algorithms with the ARM handling high-level control while FPGA fabric processes pixel data at wire speed
- Build software-defined radio systems where the ARM runs the protocol stack and the FPGA implements the modem
- Deploy machine learning inference with ARM-based pre/post-processing and FPGA-based tensor acceleration
Where Engineers Deploy the Zynq-7035
- 5G/LTE small cells and base stations — Digital front-end processing with integrated control plane
- Autonomous vehicles — Sensor fusion combining cameras, LiDAR, and radar with real-time processing
- Medical ultrasound — 128+ channel beamforming with ARM-based image reconstruction
- Industrial robotics — EtherCAT/PROFINET master with multi-axis servo control in fabric
- Aerospace & defense — Software-defined radio, electronic warfare, and SIGINT platforms
- High-performance data acquisition — Multi-channel ADC/DAC interfaces with real-time DSP
Development Ecosystem
Vivado Design Suite handles the FPGA logic design. Vitis (or the classic Xilinx SDK) handles the ARM software side. PetaLinux provides a complete Linux BSP with device tree, kernel, and rootfs generation. The ecosystem is mature, well-documented, and backed by a massive community.
In Stock at PCX — Skip the Lead Time
Zynq-7000 parts have been allocation-constrained for years. When you find stock, you buy it. PCX has the XC7Z035-2FBG676E available now — fully authorized, fully traceable, ready to ship.
Also from PCX: Check out our selection of cost-optimized Spartan-6 and Spartan-7 FPGAs for designs that don’t need the full SoC treatment.