AMD Kria KV260 Vision AI Starter Kit (K26 SOM) vs NVIDIA DGX Spark

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.

AMDSOM starter kit (adaptive SoC)No licensed photo yet

NVIDIA DGX Spark desktop AI computer, gold metal-foam chassis, oblique view on a wooden table — NVIDIA DGX Spark. Photo: Daniel Lu (User:dllu), CC BY-SA 4.0, via Wikimedia Commons.

NVIDIA DGX Spark is faster in the only like-for-like comparison (largest gap: object detection, 29.47×).

	AMD Kria KV260 Vision AI Starter Kit (K26 SOM)	NVIDIA DGX Spark
Best object detection	276 FPS D Older data (2023)	990 FPS B 2026-09-27 (page last edited)
Power	7.50 W spec	153 W spec
Perf / W	36.76 FPS/W spec	6.47 FPS/W spec
Price	n/a	$6950 list
Workloads with data	4 of 7	2 of 7

Best published object detection result per platform (may be from different like-for-like groups — see data). n/a = not published; never estimated.

Show data — all workloads, sources and like-for-like groups

object detection

✓ Like-for-like: model-size class nano (YOLO-n), batch 1. Model, precision and stack may still differ.

Platform	Result	Model · precision · stack	Grade	Power	Source
NVIDIA DGX Spark	990 FPS inference_time · b=1	YOLO11n FP16 · Ultralytics TensorRT · nano (YOLO-n)	B	153 W spec platform spec envelope (not measured with this result)	Ultralytics documentation (benchmark table) 2026-09-27 (page last edited)
AMD Kria KV260 Vision AI Starter Kit (K26 SOM)	33.60 FPS throughput (DPU only) · b=1	YOLOv8n-Ghost (Hard-Swish, xView3-SAR) INT8 (Vitis AI quantizer, power-of-two) · Vitis AI 3.5 VART, DPU inference only, 2 threads · nano (YOLO-n)	C	7.50 W spec platform spec envelope (not measured with this result)	arXiv 2507.04842v2 (SAR vessel detection on Kria KV260), Table 6 2025-11-17 (published)

instance segmentation

◌

No published instance segmentation results for AMD Kria KV260 Vision AI Starter Kit (K26 SOM) or NVIDIA DGX Spark yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

pose estimation

Only AMD Kria KV260 Vision AI Starter Kit (K26 SOM) has published pose estimation results.

Platform	Result	Model · precision · stack	Grade	Power	Source
AMD Kria KV260 Vision AI Starter Kit (K26 SOM)	352 FPS E2E throughput, multi thread · b=unknown	pt_movenet_coco_192_192_0.5G INT8 · Vitis AI 3.0 (VART) · mobile-class CNN	D	7.50 W spec reported in result source — spec	Vitis AI 3.0 Model Zoo performance table Older data (2023)

NVIDIA DGX Spark:

◌

No published pose estimation results for NVIDIA DGX Spark or AMD Kria KV260 Vision AI Starter Kit (K26 SOM) yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

image classification

Only AMD Kria KV260 Vision AI Starter Kit (K26 SOM) has published image classification results.

Platform	Result	Model · precision · stack	Grade	Power	Source
AMD Kria KV260 Vision AI Starter Kit (K26 SOM)	214 FPS E2E throughput, multi thread · b=unknown	pt_OFA-resnet50_imagenet_160_160_0.88_1.8G INT8 · Vitis AI 3.0 (VART) · mid-size CNN	D	7.50 W spec reported in result source — spec	Vitis AI 3.0 Model Zoo performance table Older data (2023)

NVIDIA DGX Spark:

◌

No published image classification results for NVIDIA DGX Spark or AMD Kria KV260 Vision AI Starter Kit (K26 SOM) yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

semantic segmentation

Only AMD Kria KV260 Vision AI Starter Kit (K26 SOM) has published semantic segmentation results.

Platform	Result	Model · precision · stack	Grade	Power	Source
AMD Kria KV260 Vision AI Starter Kit (K26 SOM)	32.90 FPS E2E throughput, multi thread · b=unknown	pt_SemanticFPN-mobilenetv2_cityscapes_512_1024_10G INT8 · Vitis AI 3.0 (VART) · mobile-class CNN	D	7.50 W spec reported in result source — spec	Vitis AI 3.0 Model Zoo performance table Older data (2023)

NVIDIA DGX Spark:

◌

No published semantic segmentation results for NVIDIA DGX Spark or AMD Kria KV260 Vision AI Starter Kit (K26 SOM) yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

speech-to-text

◌

No published speech-to-text results for AMD Kria KV260 Vision AI Starter Kit (K26 SOM) or NVIDIA DGX Spark yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

LLM generation

Only NVIDIA DGX Spark has published LLM generation results.

Platform	Result	Model · precision · stack	Grade	Power	Source
NVIDIA DGX Spark	60.39 tok/s tg128 · b=1	Llama 2 7B Q4_0 · llama.cpp Vulkan (FA on, commit 657e011) · LLM 5–14B	C	153 W spec platform spec envelope (not measured with this result)	llama.cpp discussion #10879 Vulkan scoreboard (via GitHub API) 2026-10-10 (page last edited)

AMD Kria KV260 Vision AI Starter Kit (K26 SOM):

◌

No published LLM generation results for AMD Kria KV260 Vision AI Starter Kit (K26 SOM) or NVIDIA DGX Spark yet.
We only show cited public figures — nothing is estimated to fill gaps. Requests tell us what to source or test next.

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.