XCVE2602

AMDVersal AI EdgeSoC

The XCVE2602 is a Versal AI Edge SoC FPGA from AMD, built on a 7 nm process. It provides 375,000 LUT6 look-up tables (820,313 system logic cells), about 600,000 LUT4 equivalents, 750,000 flip-flops, 16.7 Mb of block RAM plus 63 Mb of UltraRAM and 984 DSP Engines. Its processing system includes two Arm Cortex-A72 cores, a dual Arm Cortex-R5F real-time processor. The programmable logic offers up to 530 user I/O pins and 32 GTYP transceivers at up to 32.75 Gb/s, with hard PCI Express Gen4 x16 blocks. It is available in 2 packages, from the NSVH1369 (35 × 35 mm) to the VSVH1760 (40 × 40 mm). Designs are built with AMD Vivado.

600KLUT4-equivalent

375KLUTs (LUT6)

984DSP Engines

63 MbUltraRAM

16.7 MbBlock RAM

530Max. user I/O

32Transceivers (32.75 Gb/s)

2× A72Hard processor

Specifications

Vendor	AMD
Family	Versal AI Edge
Series	Versal
Process	7 nm
Configuration memory	SRAM
LUT4-equivalent capacity	600,000
Look-up tables (LUT6)	375,000
System logic cells	820,313
Flip-flops	750,000
Distributed RAM	11,674 Kb
Block RAM	16.7 Mb
UltraRAM	63 Mb
AI Engines	152 AI Engine-ML tiles
DSP (DSP Engine)	984

Max. user I/O	530
GTYP transceivers	32 × 32.75 Gb/s
PCI Express (PL)	Gen4 x16, 4× Gen4 x8
Analog / monitor	System Monitor
Hard processor	2× Arm Cortex-A72
Real-time processor	Dual Arm Cortex-R5F
Design tool	AMD Vivado

Packages

Package	Size (mm)	Pitch (mm)	User I/O	PS I/O	Transceivers
NSVH1369	35 × 35	0.92	368	78	32
VSVH1760	40 × 40	0.92	530	78	32

Package names are shown as in the selection guide; the full ordering code adds speed grade and temperature (see the data sheet).

Alternatives to the XCVE2602

Closest matches in other families and vendors by logic, DSP, memory, I/O and transceivers. Check pin compatibility and tools before migrating.

Device	LUT4-eq.	Vendor logic	DSP	Block RAM	Max I/O	Transceivers	PCIe
XCVC2602AMD Versal AI Core	600K 375K LUT6	820K SLC	984	17 Mb	530	32 GTYP 32.75 Gb/s	Gen5
XCVM1302AMD Versal Prime	514K 322K LUT6	703K SLC	848	18 Mb	454	24 GTY 32.75 Gb/s	Gen4
XCVC2802AMD Versal AI Core	833K 521K LUT6	1139K SLC	1,312	21 Mb	530	32 GTYP 32.75 Gb/s	Gen5
XCVM2202AMD Versal Prime	833K 521K LUT6	1139K SLC	1,312	21 Mb	368	32 GTYP 32.75 Gb/s	Gen5
XCVC1502AMD Versal AI Core	596K 372K LUT6	815K SLC	1,032	30 Mb	508	32 GTY 32.75 Gb/s	Gen4
SX 650Altera Stratix 10 GX/SX	415K 207K ALMs	612K LE	1,152	49 Mb	392	24 XCVR/XCVR 28.3 Gb/s	Gen3

Compare side by side: XCVE2602 vs XCVC2602, XCVE2602 vs XCVM1302, XCVE2602 vs XCVC2802, XCVE2602 vs XCVC1502.

XCVE2602 quick facts

How many LUTs does the XCVE2602 have?

The XCVE2602 has 375,000 LUT6 look-up tables, about 600,000 LUT4 equivalents (820,313 system logic cells as published by AMD).

How many DSP blocks does the XCVE2602 have?

The XCVE2602 has 984 DSP Engines and 16.7 Mb of block RAM plus 63 Mb of UltraRAM.

Does the XCVE2602 have transceivers?

Yes, 32 GTYP transceivers at up to 32.75 Gb/s.

Does the XCVE2602 have a processor?

Yes. It includes a hard 2× Arm Cortex-A72 processor.

What are the alternatives to the XCVE2602?

Devices of similar size in other families include XCVC2602 (AMD Versal AI Core), XCVM1302 (AMD Versal Prime), XCVC2802 (AMD Versal AI Core).

Which software is used to design with the XCVE2602?

Designs for the XCVE2602 are built with AMD Vivado.

Other Versal AI Edge devices

Device	LUT4-eq.	Vendor logic	DSP	Block RAM	Max I/O	Transceivers	PCIe
XCVE2002	32K 20K LUT6	43.8K SLC	90	0.8 Mb	216	—	—
XCVE2102	58.6K 36.6K LUT6	80.1K SLC	176	1.7 Mb	216	—	—
XCVE2202	168K 105K LUT6	230K SLC	324	3.8 Mb	238	8 GTYP 32.75 Gb/s	Gen4
XCVE2302	240K 150K LUT6	329K SLC	464	5.4 Mb	238	8 GTYP 32.75 Gb/s	Gen4
XCVE2602	600K 375K LUT6	820K SLC	984	16.7 Mb	530	32 GTYP 32.75 Gb/s	Gen4
XCVE1752	718K 449K LUT6	981K SLC	1,312	33.5 Mb	530	44 GTY 32.75 Gb/s	Gen4
XCVE2802	833K 521K LUT6	1139K SLC	1,312	21.1 Mb	530	32 GTYP 32.75 Gb/s	Gen4

Source: AMD Versal AI Edge Series Product Selection Guide, XMP454. Data is extracted automatically from the vendor's selection guide; verify critical values in the device data sheet.