MPFS160T vs XCZU3TEG
A comparison of the Microchip PolarFire SoC MPFS160T and the AMD Zynq UltraScale+ MPSoC XCZU3TEG using the data from both vendors' selection guides.
Side-by-side specifications
| MPFS160T | XCZU3TEG | |
|---|---|---|
| Vendor / family | Microchip PolarFire SoC | AMD Zynq UltraScale+ MPSoC |
| Logic (LUT4-equivalent) | 161,000 | 115,200 |
| LUTs as published | 161,000 LUT4 | 72,000 LUT6 |
| Vendor logic figure | 161,000 LE | 157,000 SLC |
| DSP | 498 18×18 math blocks (MACC) | 576 DSP slices |
| Block RAM | 10.2 Mb | 5.1 Mb |
| Large RAM | — | 14 Mb |
| Max. user I/O | 312 | 124 |
| Transceivers | 8 × SERDES 12.5 Gb/s | 8 × GTH 16.3 Gb/s |
| Hard PCI Express | Gen2 x4 | Gen3 x8 |
| Hard processor | 4× RISC-V (RV64GC U54) | 4× Arm Cortex-A53 |
| Process | 28 nm | 16 nm |
| Configuration memory | Non-volatile (SONOS flash) | SRAM |
| Packages | 3 | 2 |
| Design tool | Microchip Libero SoC | AMD Vivado |
Highlighted values are higher. Logic is compared in LUT4 equivalents; see how FPGA vendors count logic.
Key differences
The MPFS160T has about 40% more logic capacity (161K vs 115K LUT4 equivalents).
They use different logic architectures: the MPFS160T is built from LUT4 and the XCZU3TEG from LUT6, so the raw LUT counts (161K vs 72K) are not directly comparable.
For arithmetic, the XCZU3TEG has 576 DSP slices against 498 18×18 math blocks (MACC) on the MPFS160T.
The XCZU3TEG has more on-chip memory (19.1 Mb vs 10.2 Mb).
Both have transceivers: 8 at up to 12.5 Gb/s on the MPFS160T and 8 at up to 16.3 Gb/s on the XCZU3TEG.
Configuration memory: non-volatile (sonos flash) on the MPFS160T, sram on the XCZU3TEG.
The MPFS160T is designed with Microchip Libero SoC and the XCZU3TEG with AMD Vivado.
Full specifications
See the complete spec sheets with packages and grades: MPFS160T (PolarFire SoC) and XCZU3TEG (Zynq UltraScale+ MPSoC).