U200, U250 Brief Datasheet by Xilinx Inc.

View All Related Products | Download PDF Datasheet
V XILINX. A ALVEON
Adaptable. Intelligent.
Adaptable Accelerator Cards
for Data Center Workloads
Up to 90X higher performance than CPUs1 on key
workloads at 1/3 the cost2
Over 4X higher inference throughput3 and 3X latency
advantage over GPU-based solutions4
Machine learning inference to video processing to
any workload using the same accelerator card
As workloads algorithms evolve, use reconfigurable hardware
to adapt faster than fixed-function accelerator card product cycles
Deploy solutions in the cloud or on-premises interchangeably,
scalable to application requirements
Applications available for common workloads,
or build your own with the application developer tool
Xilinx® Alveo™ Data Center accelerator cards are designed to
meet the constantly changing needs of the modern Data Center,
providing up to 90X performance increase over CPUs for common
workloads, including machine learning inference, video transcoding,
and database search and analytics.
With complex algorithms evolving faster than silicon design cycles,
it’s impossible for fixed function GPU and ASIC devices to keep
pace. Built on Xilinx 16nm UltraScale™ architecture, Alveo U200 and
U250 accelerator cards provide reconfigurable acceleration that
can adapt to continual algorithm optimizations, supporting any
workload type while reducing overall cost of ownership.
Enabling Alveo accelerator cards is a growing ecosystem of Xilinx
and partner applications for common Data Center workloads.
For custom solutions, Xilinx’s Application Developer tool
(SDAccel™ tool) and Machine Learning Suite provide the tools
for developers to bring differentiated applications to market.
OVERVIEW
HIGHLIGHTS
Fast – Highest Performance
Adaptable – Accelerate Any Workload
Accessible – Cloud On-Premises Mobility
Database Search & Analytics
Financial Computing
Machine Learning
Storage Compression
Video Processing/Transcoding
Genomics
ADAPTABLE TO ANY WORKLOAD
1: BlackLynx Elasticsearch on Alveo versus EC2 c4.8xlarge
2: Based on CapEx & OpEx savings for DNN inference on Alveo vs dual-socket
Intel Xeon Platinum servers
3: Accelerating DNNs with Alveo Accelerator Cards White Paper
4: Measured on CNN+BLSTM Speech-to-Text ML inference against Nvidia P4
Breathe New Life Into
Your Data Center
U200 | U250
5000 A000 5000 2000 Images/s 1000 mm Xaun 200mm X XILINX. \n‘e-‘Ama‘fl w: Wm: vvuu xmnx 0200 xmnx 0250 CPU‘GPU — CPU‘AWGO - low" Value is Ben" 05‘0‘520253035LDA550 cummsw Sneuch-m-Yut Lainey (ms)
© Copyright 2018 Xilinx, Inc. Xilinx, the Xilinx logo, Artix, ISE, Kintex, Spartan, Virtex, Vivado, Zynq, and other designated brands included herein are trademarks of Xilinx in the United States
and other countries. AMBA, AMBA Designer, ARM, ARM1176JZ-S, CoreSight, Cortex, and PrimeCell are trademarks of ARM in the EU and other countries. PCIe, and PCI Express are trademarks
of PCI-SIG and used under license. All other trademarks are the property of their respective owners.
Printed in the U.S.A. AC092018
Corporate Headquarters
Xilinx, Inc.
2100 Logic Drive
San Jose, CA 95124
USA
Tel: 408-559-7778
www.xilinx.com
Xilinx Europe
One Logic Drive
Citywest Business Campus
Saggart, County Dublin
Ireland
Tel: +353-1-464-0311
www.xilinx.com
Japan
Xilinx K.K.
Art Village Osaki Central Tower 4F
1-2-2 Osaki, Shinagawa-ku
Tokyo 141-0032 Japan
Tel: +81-3-6744-7777
japan.xilinx.com
Asia Pacific Pte. Ltd.
Xilinx, Asia Pacific
5 Changi Business Park
Singapore 486040
Tel: +65-6407-3000
www.xilinx.com
India
Meenakshi Tech Park
Block A, B, C, 8th & 13th floors,
Meenakshi Tech Park, Survey No. 39
Gachibowli(V), Seri Lingampally (M),
Hyderabad -500 084
Tel: +91-40-6721-4747
www.xilinx.com
Product Brief Xilinx Alveo U200 & U250
Visit www.xilinx.com/alveo for more information. To get started now, visit Nimbix cloud services to test drive Data Center workloads
or purchase Alveo U200 or U250 Data Center accelerator cards for on-premises deployment.
BENCHMARKS
Adapt and Accelerate Any Workload
Increase Real-Time Machine Learning* Throughput by 20X Reduce ML Inference Latency by 3X
TAKE THE NEXT STEP
FEATURES ALVEO U200
ACCELERATOR CARDS
ALVEO U250
ACCELERATOR CARDS
Peak INT8 TOPs 18.6 33.3
DDR Memory Bandwidth 77GB/s 77GB/s
Internal SRAM Bandwidth 31TB/s 38TB/s
Look-up Tables (LUTs) 892,000 1,341,000
Thermal Options Passive or Active Passive or Active
AREA PARTNER WORKLOAD ALVEO ACCELERATION VS CPU
Database Search and Analytics BlackLynx Unstructured Data Elasticsearch 90X
Financial Computing Maxeler Value-at-Risk (VAR) Calculation 89X
Machine Learning Xilinx Real-Time Machine Learning Inference 20X
Video Processing / Transcoding NGCodec HEVC Video Encoding 12X
Genomics Falcon Computing Genome Sequencing 10X
CPU Comparisons: Xeon c4.8xlarge AWS | Xeon E5-2643 v4 3.4GHz | Xeon Platinum c5.18xlarge AWS | Dual Socket E5-2680 v3 2.5GHz | Xeon f1.16xlarge
*GoogLeNet V1 : Accelerating DNNs with Xilinx Alveo Accelerator Cards White Paper CPU+GPU: Nvidia P4 + Xeon CPU E5-2690 v4 @2.60GHz (56 Cores)
CPU+Alveo: Alveo U200 or U250 + Xeon CPU E5-2686 v4 @2.3GHz (8 Cores)

Products related to this Datasheet

BOARD DCAB SERVER U200 ACTIVE
Available Quantity: 27
Unit Price: 7798.72
BOARD DCAB SERVER U200 PASSIVE
Available Quantity: 1
Unit Price: 7798.72
BOARD DCAB SERVER U250 PASSIVE
Available Quantity: 0
Unit Price: 12135.71
BOARD DCAB SERVER U250 ACTIVE
Available Quantity: 0
Unit Price: 12135.71