AMD Alveo V70 (Versal AI Core, XDNA AIE-ML): edge AI benchmarks

Name: AMD Alveo V70 (Versal AI Core, XDNA AIE-ML)
Brand: AMD

FPGA / Versal coverage note: 1,243 rows come from the AMD Vitis AI 3.0/3.5 Model Zoo tables; 733 (59%) are image classification. Detection, segmentation and other workloads are thinner, and there is no LLM or speech data for these boards. All are vendor-published (grade D). FPGA / Versal overview.

AMDPCIe accelerator cardNo licensed photo yet

Vendor AMDForm factor PCIe accelerator cardCompute DPU (AIE-ML)Indicative price unknownPower envelope 75 W
spec: card TDP 75 W (AMD product brief); host not includedPublished rows 90

Key results

Best published result per workload for AMD Alveo V70 (Versal AI Core, XDNA AIE-ML)
Workload	Best result	Model · precision	Power · perf/W	Grade · date	Source
object detection	6100 FPS batch unknown	tf_ssdmobilenetv1_3.5 INT8 · Vitis AI 3.5 (VART)	75.00 W (spec) 81.34 FPS/W	D Older data (2023)	Vitis AI 3.5 Model Zoo performance table
pose estimation	8327 FPS batch unknown	pt_movenet_3.5 INT8 · Vitis AI 3.5 (VART)	75.00 W (spec) 111 FPS/W	D Older data (2023)	Vitis AI 3.5 Model Zoo performance table
image classification	63109 FPS batch unknown	tf_mobilenetv1-0.25_3.5 INT8 · Vitis AI 3.5 (VART)	75.00 W (spec) 841 FPS/W	D Older data (2023)	Vitis AI 3.5 Model Zoo performance table
semantic segmentation	1177 FPS batch unknown	tf2_2D-UNET_3.5 INT8 · Vitis AI 3.5 (VART)	75.00 W (spec) 15.70 FPS/W	D Older data (2023)	Vitis AI 3.5 Model Zoo performance table
super-resolution	298 FPS batch unknown	pt_SESR-S_3.5 INT8 · Vitis AI 3.5 (VART)	75.00 W (spec) 3.98 FPS/W	D Older data (2023)	Vitis AI 3.5 Model Zoo performance table

Best published result per workload, across model sizes. For like-for-like groups, open the per-workload tables below.

Show data — per-workload tables

object detection

Like-for-like group	Best result	Model · precision · stack	Grade	Power	Source · date	Rank in group
small (YOLO-s) · b=14	85.28 FPS	yolov5s6_pt INT8 · Vitis AI 3.5 Library (VART), batch 14 DPU	D	75.00 W spec	AMD UG1354 Vitis AI Library performance tables undated	2 of 2
medium (YOLO-m) · b=14	161 FPS	yolov8m_pt INT8 · Vitis AI 3.5 Library (VART), batch 14 DPU	D	75.00 W spec	AMD UG1354 Vitis AI Library performance tables undated	2 of 2
medium (YOLO-m) · b=unknown	280 FPS	pt_yolov6m_3.5 INT8 · Vitis AI 3.5 (VART)	D	75.00 W spec	Vitis AI 3.5 Model Zoo performance table Older data (2023)	1 of 6
mid-size CNN · b=unknown	70.16 FPS	tf_mlperf_resnet34_3.5 INT8 · Vitis AI 3.5 (VART)	D	75.00 W spec	Vitis AI 3.5 Model Zoo performance table Older data (2023)	9 of 34
mobile-class CNN · b=unknown	6100 FPS	tf_ssdmobilenetv1_3.5 INT8 · Vitis AI 3.5 (VART)	D	75.00 W spec	Vitis AI 3.5 Model Zoo performance table Older data (2023)	1 of 6
unknown · b=14	540 FPS	yolov5_nano_pt INT8 · Vitis AI 3.5 Library (VART), batch 14 DPU	D	75.00 W spec	AMD UG1354 Vitis AI Library performance tables undated	1 of 2
unknown · b=unknown	2174 FPS	pt_face-mask-detection_3.5 INT8 · Vitis AI 3.5 (VART)	D	75.00 W spec	Vitis AI 3.5 Model Zoo performance table Older data (2023)	5 of 37

instance segmentation

◌

No published instance segmentation results for AMD Alveo V70 (Versal AI Core, XDNA AIE-ML) yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

pose estimation

Like-for-like group	Best result	Model · precision · stack	Grade	Power	Source · date	Rank in group
mobile-class CNN · b=unknown	8327 FPS	pt_movenet_3.5 INT8 · Vitis AI 3.5 (VART)	D	75.00 W spec	Vitis AI 3.5 Model Zoo performance table Older data (2023)	1 of 6

image classification

Like-for-like group	Best result	Model · precision · stack	Grade	Power	Source · date	Rank in group
mid-size CNN · b=unknown	7780 FPS	pt_OFA-resnet50_0.88_3.5 INT8 · Vitis AI 3.5 (VART)	D	75.00 W spec	Vitis AI 3.5 Model Zoo performance table Older data (2023)	8 of 41
mobile-class CNN · b=unknown	63109 FPS	tf_mobilenetv1-0.25_3.5 INT8 · Vitis AI 3.5 (VART)	D	75.00 W spec	Vitis AI 3.5 Model Zoo performance table Older data (2023)	1 of 31
unknown · b=unknown	12926 FPS	pt_OFA-depthwise-res50_3.5 INT8 · Vitis AI 3.5 (VART)	D	75.00 W spec	Vitis AI 3.5 Model Zoo performance table Older data (2023)	1 of 6

semantic segmentation

Like-for-like group	Best result	Model · precision · stack	Grade	Power	Source · date	Rank in group
mid-size CNN · b=unknown	1177 FPS	tf2_2D-UNET_3.5 INT8 · Vitis AI 3.5 (VART)	D	75.00 W spec	Vitis AI 3.5 Model Zoo performance table Older data (2023)	3 of 6
unknown · b=unknown	187 FPS	pt_pointpillars_3.5 INT8 · Vitis AI 3.5 (VART)	D	75.00 W spec	Vitis AI 3.5 Model Zoo performance table Older data (2023)	2 of 6

speech-to-text

◌

No published speech-to-text results for AMD Alveo V70 (Versal AI Core, XDNA AIE-ML) yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

LLM generation

◌

No published LLM generation results for AMD Alveo V70 (Versal AI Core, XDNA AIE-ML) yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

super-resolution

Like-for-like group	Best result	Model · precision · stack	Grade	Power	Source · date	Rank in group
unknown · b=unknown	298 FPS	pt_SESR-S_3.5 INT8 · Vitis AI 3.5 (VART)	D	75.00 W spec	Vitis AI 3.5 Model Zoo performance table Older data (2023)	3 of 6

Compare with another platform (107)

8xMI355X_2xEPYC_9575F (8x AMD Instinct MI355X 288GB HBM3e) · AMD Kria KV260 Vision AI Starter Kit (K26 SOM) · AMD Ryzen AI Max+ (Strix Halo) iGPU · AMD Ryzen AI Max+ 395 / PRO 395 (Strix Halo, Radeon 8060S) · AMD Ryzen AI processor (SKU not stated) · AMD Versal AI Core VCK190 (VC1902) · AMD Versal AI Edge Series Gen 2 VEK385 · AMD Versal AI Edge VEK280 (VE2802) · AMD Zynq UltraScale+ ZCU102 (ZU9EG) · AMD Zynq UltraScale+ ZCU104 (ZU7EV) · Altera Agilex 5 FPGA E-Series 065B Modular Development Kit (FPGA AI Suite, OFS PCIe) · Altera Agilex 7 FPGA I-Series Development Kit (FPGA AI Suite) · Apple M4 (10-core GPU, 120 GB/s) · Apple Mac mini M4 Pro 64GB · Axelera Metis Edge 130p PCIe card · Axelera Metis Edge 130p PCIe card (16GB) · Axelera Metis Embedded 113m M.2 card · Cisco UCS C880A M8 (8x NVIDIA B300-SXM-270GB) · Coral Dev Board CPU (quad Cortex-A53 @1.5 GHz, CPU only) · CoreWeave GB200 NVL72 (4x NVIDIA GB200) · CoreWeave GB300 NVL72 (4x NVIDIA GB300) · Dell Pro Max with GB10 · G8825Z5 (8x AMD Instinct MI350X 288GB HBM3e) · G894-AD3-AAX7 (8x NVIDIA B300-SXM-270GB) · G894-SD3-AAX7 (8x NVIDIA B300-SXM-270GB) · G894-ZD3-AAX7 (8x NVIDIA B300-SXM-270GB) · Google Coral Dev Board (Edge TPU) · Google Coral USB Accelerator (Edge TPU) on Xeon Gold 6154 host, USB 3.0 · HPE ProLiant Compute DL380a Gen12 (10x NVIDIA RTX PRO 6000 Blackwell Server Edition) · HPE ProLiant Compute DL380a Gen12 (8x NVIDIA RTX PRO 6000 Blackwell Server Edition) · HPE ProLiant Compute XD685 (8x AMD Instinct MI355X 288GB HBM3e) · HPE ProLiant DL385 Gen11 (4x NVIDIA RTX PRO 6000 Blackwell Server Edition) · Hailo-10H (M.2/PCIe, x86 host) · Hailo-8 (M.2/PCIe, x86 host) · Hailo-8L (M.2/PCIe, x86 host) · Intel Arc A-Series Graphics · Intel Arc B-Series Graphics · Intel Atom X6425E · Intel Atom x7425E · Intel Celeron 6305E · Intel Core 5 330 · Intel Core Ultra 7 155H · Intel Core Ultra 7 155H (Meteor Lake) · Intel Core Ultra 7 255H · Intel Core Ultra 7 258V (Lunar Lake) · Intel Core Ultra 7 265H · Intel Core Ultra 7 368H · Intel Core Ultra 9 288V · Intel Core Ultra X7 358H (Panther Lake) · Intel Core i5-1235U · Intel Core i5-1335U · Intel Core i5-13600K · Intel Core i7-1185G7 · Intel Core i7-1185GRE · Intel Core i7-12700H · Intel Core i7-1355U · Intel Data Center GPU Flex 170 · Intel N100 · Intel Xeon Gold 5218T · Intel Xeon Gold 6154 desktop (CPU only, TFLite) · Intel Xeon Gold 6338N · Intel Xeon Platinum 6972P · Intel Xeon Platinum 8280 · Intel Xeon Platinum 8380 · Intel Xeon Platinum 8480+ · Intel Xeon Platinum 8580 · Lattice Avant-E Evaluation Board (sensAI) · Lattice CertusPro-NX Voice and Vision ML Board (sensAI) · Lattice CrossLink-NX-33 Voice and Vision ML Board (sensAI) · Lattice iCE40 UltraPlus Mobile Development Platform (sensAI) · MemryX MX3 M.2 module · Microchip PolarFire SoC Video Kit (VectorBlox CoreVectorBlox) · NDP9120 (N/A) · NUCLEO-L4R5ZI (STM32L4R5ZIT6U) · NUCLEO-U575ZI-Q (STM32U575ZIT6Q) · NVIDIA DGX B200 (8x NVIDIA B200-SXM-180GB) · NVIDIA DGX B300 (8x NVIDIA B300-SXM-270GB) · NVIDIA DGX Spark · NVIDIA GB200 NVL72 (4x NVIDIA GB200) · NVIDIA GB300 NVL72 (4x NVIDIA GB300) · NVIDIA Jetson AGX Orin 64GB · NVIDIA Jetson AGX Thor · NVIDIA Jetson Orin NX 16GB · NVIDIA Jetson Orin Nano Super · NVIDIA Jetson Thor T4000 · Nebius B300 n1 (8x NVIDIA B300-SXM-270GB) · Nebius GB300 NVL72 (4x NVIDIA GB300) · Qualcomm Dragonwing IQ-9075 EVK · Qualcomm Dragonwing RB3 Gen 2 (QCS6490) · Qualcomm Snapdragon X Elite (CRD) · Qualcomm Snapdragon X2 Elite (CRD) · QuantaGrid D75H-10U (8x NVIDIA B300-SXM-270GB) · Raspberry Pi 5 (CPU only) · Raspberry Pi 5 + AI HAT+ (Hailo-8) · Rockchip RK1808 (NPU) · Rockchip RK3562 (NPU) · Rockchip RK3566/RK3568 (NPU) · Rockchip RK3576 (NPU, single core of 2) · Rockchip RK3588 (NPU, single core of 3) · Rockchip RV1109 (NPU) · Rockchip RV1126 (NPU) · Rockchip RV1126B (NPU) · TI AM68A (8 TOPS) · TRIGKEY Mini PC (AMD Ryzen 5 5560U, CPU) · Terasic DE10-Agilex Development Board (Agilex 7, FPGA AI Suite) · XA NB3I-E12 (8x NVIDIA B300-SXM-270GB) · XE7745_8xMI350P_2xEPYC_9455 (8x AMD Instinct MI350P)

How to read these numbers

Every number is labelled. Confidence: A MLPerf closed, audited · B documented third-party method · C community / review · D vendor-published, not independently verified · E estimate. Power basis: measured with the result · spec vendor spec/envelope · power n/a not published.

Machine-readable: /find.json (see llms.txt); full tables are in “Show data”. Confidence grades: A audited MLPerf closed · B documented third-party · C community · D vendor-published · E estimate. Unknown values are shown as unknown, never estimated. Generated 2026-10-16 from public sources.