Join us for an exciting and informative preview of the broadest range of next-generation systems optimized for tomorrow’s data center workloads, Powered by 4th Gen Intel® Xeon® Scalable Processors (formerly codenamed Sapphire Rapids). Experts from Supermicro and Intel will discuss how the upcoming Supermicro X13 systems will enable new performance levels utilizing state-of-the-art technology, including DDR5, PCIe 5.0, Compute Express Link™ 1.1, and Intel® Advanced Matrix Extensions (Intel AMX).

1. X13 Pre-Release Update
Featuring 4th Gen Intel® Xeon® Scalable
processors (Formerly Codenamed
Sapphire Rapids)
Bill Chen, GM, System Product Management, Supermicro
Gary Brown, Sr. Product Marketing PM, Xeon & Memory Group, DCAI, Intel
November 17, 2022

2. Bill Chen
GM, System Product
Management, Supermicro
Gary Brown
Sr. Product Marketing PM,
Xeon & Memory Group,
DCAI, Intel
Michael Schulman
Sr. Manager, Content,
Supermicro

3. ©2022 Supermicro
Supermicro X13 Pre-Release Update
Next-Generation Industry
Technologies
• High Bandwidth Memory
• CXL 1.1
• PCIe 5.0
• DDR5
Acceleration Engines in 4th
Gen Intel® Xeon® Scalable
Processors
• Intel Advanced Matrix
Extensions
• Intel Data Streaming
Accelerator
• Intel QuickAssist
Technology
• Intel vRAN Boost
Supermicro X13 Systems
• BigTwin®
• Hyper
• SuperBlade®
• GPU-accelerated
• Edge

4. ©2022 Supermicro
The Supermicro X13 Advantage
Resource-Saving
Architecture
Rack-Scale
Plug and Play
Higher Ambient
Operating Temperatures
Reduces Power
Consumption Per Server
Ready-to-Use
Deployment in Less
Time
Lowers PUE by
Reducing Data Center
AC Requirements

5. ©2022 Supermicro
The Industry’s Broadest Range of Servers
Powered by 4th Gen Intel® Xeon® Scalable processors
SuperBlade®
Highest Density Multi-Node
Architecture for HPC Applications
Hyper-E and Hyper
Best-in-Class High Performance,
Scalable Systems
BigTwin®
Highly Modular Multi-Node
Systems with Tool-Less Design
GrandTwin™
Highly Modular Multi-Node
Systems with Tool-Less Design
FatTwin®
Multi-Node 4U Advanced Twin
Architecture with 8 or 4 Nodes
High Density Systems for Double-
Width, Full Length PCIe GPUs
PCI-E GPU Servers
SuperEdge
High-Density Computing and
Flexibility at the Intelligent Edge
CloudDC
All-in-one Rackmount Platforms
for Cloud Data Centers
WIO (UP)
Single Socket Optimized for
I/O workloads
Compact Form Factors for
5G and Edge computing
Edge
Highest Performance and Flexibility
for Enterprise Applications
Quad Socket
Multi-Processor
1U Petascale
Revolutionary Petascale NVMe for
Unprecedented Density and Capacity
Universal GPU Servers
Maximum Performance and Flexibility for
GPU/AI/ML Workloads

6. ©2022 Supermicro
4th Gen Intel® Xeon® Scalable Processors
(Formerly codenamed Sapphire Rapids)
Designed to deliver potent data infrastructure from the cloud, to the network, to the intelligent edge, 4th
Gen Intel® Xeon® Scalable processors, (Formerly Codenamed Sapphire Rapids), establish a new
standard in data center architecture.

7. ©2022 Supermicro
4th Gen Intel Xeon Scalable Processors - Technologies
High Bandwidth
Memory
CXL
PCIe 5.0 DDR5

8. ©2022 Supermicro
PCIe 5.0
• Doubling the I/O bandwidth every 3 years
• PCIe 5.0, 128GB/s (bidirectional)
• Faster Ethernet capability to eliminate
throughput bottlenecks for data
• FPGA/GPU accelerators to support
emerging machine learning and deep
learning workloads with uncompromised
PCIe 4.0/5.0 bandwidth and lower latency
• Next-generation PCIe 5.0 not only helps
today’s applications run faster, it also
enables new applications in AI/ML and
cloud computing
Source: https://www.extremetech.com/extreme/337527-towards-pcie-7-0-and-blazing-fast-storage-why-
engineers-have-hit-the-gas-on-interconnect-standards

9. ©2022 Supermicro
DDR5
• Delivers continued scaling of bandwidth,
capacity, and power efficiency - up to 1.5x
bandwidth improvement vs DDR4
• Up to 512GB per DIMM
• Achieves 30% greater power efficiency
• Delivers a significant benefit to Intel-based
platforms with over 2x performance of DDR4.
• Supports eight channels of DDR5 memory, with
four memory controllers supporting the eight
channels, and an integrated memory engine.

10. ©2022 Supermicro
Compute Express Link (CXL)
• Industry-supported cache-coherent
interconnect for processors, memory
expansion, and accelerators
• Cache coherent interface is designed
for data centers and enables shared
memory spaces that reduce overhead
and latency
• Provides a critical capability to create a
unified, coherent memory space
between CPUs and accelerators, and
this innovation will revolutionize how
data center server architectures will be
built for years to come. Source Link: CXL 2.0 white paper

11. ©2022 Supermicro
High Bandwidth Memory
• Platforms based on the Intel® Xeon® processors, formerly
codenamed Sapphire Rapids will provide versions with High
Bandwidth Memory (HBM) to accelerate memory bandwidth
sensitive workloads
• SKUs with high bandwidth memory (HBM) will integrate up
to 64 GB of HBM2e memory
• Offers applications up to 4x more memory bandwidth,
providing a generational performance improvement of up to
2.8x over 3rd Gen Intel® Xeon® Scalable processors1
• Two- to three-times performance increases across weather
research, energy, manufacturing and physics workloads2
• Up to 2x performance on real-world workloads from Ansys
Fluent and ParSeNet in comparison to 3rd Gen Intel® Xeon®
Scalable processors3
• HBM will have 3 modes: HBM only, Flat, Cache
1 Source: https://edc.intel.com/content/www/us/en/products/performance/benchmarks/vision-2022/, Demo Benchmark #15. Results may vary.
2 Source: https://www.intel.com/content/www/us/en/newsroom/opinion/accelerated-innovations-sustainable-open-hpc.html, Footnote #2. Results may vary.
3 Source: https://www.intel.com/content/www/us/en/newsroom/opinion/accelerated-innovations-sustainable-open-hpc.html, Footnote #3. Results may vary.

12. ©2022 Supermicro
Accelerator Engines in 4th Gen Intel® Xeon® Scalable
Processors
• Increasing effectiveness of cores, by enabling
offload of common mode tasks via seamlessly
integrated accelerator engines:
• Intel® Advanced Matrix Extensions
• Intel® Data Streaming Accelerator
• Intel® QuickAssist Technology
• Intel® vRAN Boost

13. ©2022 Supermicro
Intel® Advanced Matrix Extensions (Intel® AMX)
• Intel’s next-generation built-in Intel Deep Learning Boost (Intel
DL Boost) advancement for deep learning inference and
training performance
• Improves the performance of deep learning workloads on Intel®
Xeon® processors
• As tested with early samples, optimized internal matrix-multiply
micro benchmarks run up to 8x faster using the new Intel AMX
instruction set extensions compared to a version of the same
micro benchmark using VNNI instructions1
• As measured by the real-time visual inference algorithm SSD-
ResNet34 Intel AMX has been shown to execute up to 4.5x
INT8 image inferences per second and up to 6x BF16 image
inferences per second in comparison to 3rd Gen Intel® Xeon®
Scalable processors2
1 Source: https://edc.intel.com/content/www/us/en/products/performance/benchmarks/architecture-day-2021/, Sapphire Rapids -Advanced Matrix Extension (AMX) demo. Results may vary.
2 Source: https://edc.intel.com/content/www/us/en/products/performance/benchmarks/vision-2022/, Demo Benchmark #14. Results may vary.

14. ©2022 Supermicro
Intel® Data Streaming Accelerator (Intel® DSA)
Offloads the most common data movement tasks that cause the
overhead seen in data-center-scale deployments.
Improves the processing overhead tasks to deliver increased overall
workload performance and can move data among CPU, memory, and
caches, as well as all attached memory, storage, and network
devices.
Will drive significant performance improvements for high-performance
storage, networking, and data-processing-intensive workloads by
optimizing streaming-data-movement and transformation operations.
Has been shown in an example Open virtual Switch use case to
reduce CPU utilization by nearly 40% and deliver a 2.5x
improvement in data movement performance. This results in nearly
doubling the effective performance for this workload1
Delivers more than 1.7x I/O per second speedup and a 45% latency
decrease for storage usages in comparison to the same benchmark
without Intel DSA (128 KB transfer size)2
1 Source: https://edc.intel.com/content/www/us/en/products/performance/benchmarks/architecture-day-2021/, section beginning ‘On an Open virtual Switch use case…’ Results may vary.
2 Source: https://edc.intel.com/content/www/us/en/products/performance/benchmarks/vision-2022/, Session Benchmark #44. Results may vary.

15. ©2022 Supermicro
Intel® QuickAssist Technology (Intel® QAT)
• Significantly increases the performance and efficiency of standard
platform solutions in cloud, networking, big data, and storage.
• Applications utilizing high-performance accelerators enabled by
Intel QAT will see greatly reduced CPU utilization and greatly
increased performance for encryption, secure client tunnels, and/or
secure storage compression.
• Rearchitected to increase performance and lower latency, improve
usability, reduce platform design and implementation overhead, and
support advanced virtualization capabilities.
• Intel QAT Acceleration Engine offload yields a benefit of up to 98%
additional core workload capacity as well as a markedly reduced
compressed data footprint. 1
• New cryptographic instructions and optimizations and rearchitected
QAT have been shown to enable approximately 4.2x the NGINX
webserver connections per second while simultaneously freeing up
four CPU cores for other tasks (in comparison to default software
on CPU only).2
1 Source: https://edc.intel.com/content/www/us/en/products/performance/benchmarks/architecture-day-2021/, section beginning, ‘With the ZlibL9 compression algorithm…’Results may vary.
2 Source: https://edc.intel.com/content/www/us/en/products/performance/benchmarks/vision-2022/, Session Benchmark #43. Results may vary.

16. ©2022 Supermicro
Intel® vRAN Boost
• New, unique signal processing instruction enhancements
in the core of Intel® Xeon® processors
• Built to support RAN-specific signal processing – will
deliver up to two times capacity gains1 for vRAN and
support advanced capabilities like high-cell density for
64T64R Massive MIMO.
• Eliminates the need for external accelerator cards by
fully integrating vRAN acceleration directly into the Intel
Xeon SOC
• Reduce vRAN compute power consumption by up to
20%2 for service providers
1 Claim estimated as of 2/10/2022 based on Sapphire Rapids architecture improvements vs 3rd generation Intel Xeon Scalable at similar core counts on a test scenario using FlexRAN software. Results may vary.
2 Estimated as of 08/30/2022based on scenario design power (SDP) analysis on pre-production 4th Generation Intel® Xeon® Scalable processor with Intel® vRAN Boost (integrated vRAN acceleration) and pre-production 4th Generation Intel® Xeon® Scalable processor at the
same core count and frequency with an external vRAN accelerator card. Performance and power vary by use, configuration, and other factors. Learn more at www.Intel.com/PerformanceIndex
Reduce power
consumption by
~20%

17. Introducing Supermicro
X13 Servers
Powered by 4th Gen Intel® Xeon®
scalable Processors (formerly code-
named Sapphire Rapids)

18. ©2022 Supermicro
Supermicro Total IT Liquid Cooling Solution
Reduce PUE and
CRAC Cost
Complete Solution
Integration and
Testing
Reduced Lead
Time
Cooling Distribution Unit
Cold Plate

19. ©2022 Supermicro
Supermicro X13 JumpStart Early Access Program
Remote access to pre-release next-generation Supermicro systems powered by 4th Gen Intel® Xeon®
Scalable processors (formerly codenamed Sapphire Rapids)
1. Sign Up
Visit
http://www.supermicro.com/jumpstart/x13 to register
2. Get Qualified
Verify Supermicro and Intel NDA information
3. Access
We’ll provide your unique username, password and URL

20. ©2022 Supermicro
Supermicro X13 Early Ship Program
• Provides early access to the next generation of Intel CPUs for approved
customers
• Full production units
• Please contact your Supermicro sales representative for more information

21. Better Faster Greener™ © 2022 Supermicro
8U SuperBlade®
20/10 Single/Double-Width
Blades in 8U
6U SuperBlade
10/5 Single/Double-Width
Blades in 6U
X13 SuperBlade®
Highest Density Multi-Node
Architecture for HPC Applications

22. X13 GPU
Better Faster Greener™ © 2022 Supermicro
4U/5U Universal 4-GPU
Multi-Architecture Flexibility, Future-Proof
Open Standards-Based Design
8U 8-GPU
Next-Generation HGX H100 8-GPU
Accelerated Computing for AI/ML
and HPC
4U/5U 10 GPU
High Performance and Flexibility

23. Better Faster Greener™ © 2022 Supermicro
BigTwin®
Highest Performance 2U Twin
Architecture with 4 or 2
Nodes
FatTwin®
Highest Efficiency 4U Twin
Architecture with 4 or 8
Nodes
GrandTwin™
Front access 2U Twin
Architecture with 4 or 2
Nodes
X13 Twin
Innovative Multi-Node
Architecture with Reduced TCO
and TCE

24. X13 1U/2U Rackmount Server
Better Faster Greener™ © 2022 Supermicro
CloudDC
Data Center Optimized Server
Hyper
High-Performance Rackmount Architecture
Superior Performance, Efficiency
and Time-to-Market for Rapid
Adoption
WIO
UP I/O Optimized Server

25. X13 5G/Edge
Better Faster Greener™ © 2022 Supermicro
Hyper-E
Best-in-Class Performance
for Edge Data Centers
SuperEdge
High-Density Computing and
Flexibility at the Intelligent Edge
Compact, flexible architectures for
space-constrained deployments
Embedded
Compact Form Factors for
5G and Edge computing

26. X13 1U Petascale All-Flash
Better Faster Greener™ © 2022 Supermicro
24x E1.S (15mm)
32x E1.S (9.5mm) 20x E3S.1T (7.5mm)
Revolutionary Petascale NVMe for
Unprecedented Density and
Capacity

27. Learn More at www.supermicro.com/X13
Thank You