Addressing Key FSI Needs
Financial services industry (FSI) institutions need high performance computing solutions that enable faster trading and transactions, speed access to market intelligence, and improve customer experiences—while keeping TCO down, system management streamlined, and security intact. Discover how AMD solutions address these imperatives across the four major FSI segments: banking, capital markets, insurance, and payments.
Performance
AMD powers solutions that can drive exceptionally low latency for FSI applications, high performance per core, and leadership acceleration / compute offload.
Efficiency
By using AMD processor-based solutions, organizations may be able to reduce their server footprint—driving power and cost efficiencies that can help address sustainability goals.
Security and Compliance
AMD processor-powered solutions are built on cutting-edge security features that help protect sensitive financial data and transactions—including real-time system memory encryption, hardware-based protection, and more.
Use Cases
- HPC & Grid Computing
- Big Data & AI
- Electronic Trading
- Employee Productivity
HPC / Grid for Risk Management and Compliance
Grid computing requires dynamic workload balancing, high availability, and robust parallel processing. As a result, organizations need high-density server solutions that can enable faster simulations, more extensive algorithm analysis, and cheaper calculations for regulatory compliance and risk management.
AMD EPYC™ server processor-based solutions offer exceptional performance per core and core density, and come with large L3 cache that keeps data closer to cores—all ideal for Risk Analysis. AMD also offers a dedicated accelerator card for ultra-low latency pre-trade risk assessment and regulatory compliance—the Alveo™ UL3524.
Big Data and AI
AI is revolutionizing every industry, and FSI is no exception. Organizations need technologies that can support robust AI-driven capabilities such as customer chatbots, automated lending, personalized recommendations, real-time fraud prevention, and more.
These capabilities demand high memory and high memory bandwidth solutions. AMD EPYC processors and AMD Instinct™ accelerator-based solutions enable exactly that. AMD also offers a dedicated accelerator card that helps provide real-time market data reliably for brokerages and exchanges—the Alveo UL3524.
Electronic Trading
AMD products deliver powerful performance with high clock speeds ideal for electronic trading, plus high core counts for multithreaded processing power ideal for trading simulations. Depending on the type of trading, latency requirements may vary. AMD offers solutions to address the full spectrum of requirements.
Ultra-Low Latency Trading
Proprietary traders, hedge funds, and exchanges requiring ultra-low latency rely on Alveo accelerator cards to enable electronic trading at nanosecond speeds.
High Frequency Trading
Sophisticated, automated algorithms are used in high frequency trading to execute large numbers of transactions at a time. The combination of AMD Ryzen™ 7950X processors with AMD Alveo™ X3 or AMD Solarflare™ X2 Ethernet adapters delivers compelling performance for the STAC-N1 benchmark, which measures how network stacks perform in a simulated market data environment.Read the blog.
Other Trading Formats
Some electronic trading involves holding periods that are longer than those in ultra-low latency or high frequency trading scenarios. Different trading frequencies may imply different trading strategies, which may also influence the underlying hardware. The 5th Gen AMD EPYC family includes high frequency SKUs, denoted with an “F” at the end of the model’s name, that can offer competitive advantages for latency-constrained trading. AMD Ryzen PRO processors also offer advanced compute capability and minimal latency for algorithmic trading.
Employee Productivity
FSI workforces require business systems that yield maximal employee productivity to keep pace with the demands of ever-changing market landscapes. Unlock the full potential of your users by arming them with the latest technology to experiment, iterate, and innovate at speed.
Portfolio
- EPYC
- Alveo
- Instinct
- Ryzen PRO
- Radeon PRO
- Solarflare
AMD EPYC Processors
AMD EPYC server processors play an instrumental role in enabling compute for FSI both on-premises and in the cloud. EPYC processors deliver exceptional performance, power efficiency, and x86 compatibility for simplified migration.
Each EPYC processor also comes equipped withAMD Infinity Guard1—a full suite of modern security features that help decrease potential attack surfaces as software is booted, executed, and processes critical data.
Featured Partner
Meet CompatibL
CompatibL is a leading provider of trading and risk software in the financial services market. In considering the industry's ongoing pursuit of digital transformation and competitive differentiation, CompatibL tested their CompatibL Platform solution on AMD 4th Gen EPYC processors in AWS and observed that their application ran performantly and scaled linearly on AMD CPUs with no code changes required.
Featured Benchmark
AMD EPYC on Quantlib
In the realm of quantitative finance, QuantLib is an open-source suite that serves as a proxy for HPC / grid algorithms. AMD EPYC processor-based cloud instances can outperform the competition when running QuantLib benchmarks.2,3
AWS
Up to 210%
Up to 210% max higher performance/$ with AWS EC2 M6a instances powered by EPYC vs. competitive solutions2
Microsoft Azure
Up to 170%
Up to 170% max higher performance/$ with Ds Azure x86 instances powered by EPYC vs. competitive solutions3
AMD Alveo Accelerators
AMD Alveo adaptive accelerators address the unique requirements of the financial market, which often demands a convergence of low latency networking and hardware acceleration for algorithmic trading, pre-trade risk analysis, and market data delivery services at nanosecond speeds.
The Alveo X3522PV, Alveo UL3422, and Alveo UL3524 accelerators target low-latency and ultra-low-latency trading, respectively, for algorithms that prioritize trade execution performance vs. algorithmic complexity. The Alveo V80, Alveo U55C, and Alveo U50 accelerator cards' compute accelerators offer more resources for complex trading algorithms, risk and price modeling, and data analytics. The VCK5000 development card features AI engines for AI-based trading at deterministic latency.
AMD Instinct Accelerators
The AMD Instinct™ MI350 Series GPUs set a new standard for Generative AI and high-performance computing (HPC) in data centers. Built on the new cutting-edge 4th Gen AMD CDNA™ architecture, these GPUs deliver exceptional efficiency and performance for training massive AI models, high-speed inference, and complex HPC workloads like scientific simulations, data processing, and computational modeling.
Featured Benchmark
AMD Ryzen PRO on Cinebench
AMD Ryzen PRO processor-powered systems can offer better performance vs competitive alternatives when running the Cinebench R23 nT benchmark.4
Up to 2.4X
Lenovo ThinkPad Z13 Gen 2 with AMD Ryzen 7 PRO 7840 enables up to 2.4X better performance vs Dell Latitude 7340 with Intel core i7-1265U4
Up to 76%
Lenovo ThinkPad Z13 Gen 2 with AMD Ryzen 7 PRO 7840 delivers up to 76% higher performance vs. Dell XPS 13+ with Intel core i7 1360P4
AMD Radeon™ PRO Graphics
Viewing the world through financial data means intensive visualization to gain market insights through analytics. Integrated AMD Radeon graphics available in professional AMD Ryzen processor-powered laptops and desktops deliver immersive visuals, 4K streaming, and support the latest high-res displays and multimonitor setups.
AMD Solarflare Ethernet Adapters
AMD Solarflare X-Series Ethernet adapters, coupled with the Onload™ family of protocol acceleration software, have been deployed in leading banks, trading firms, stock exchanges, hedge funds and market makers worldwide for over 15 years. They offer field-proven low latency, are built on optimized architecture for reliable operation and trade execution, and support 10/25/40/100GbE link speeds.
AMD Partners for FSI Systems
AMD partners offer proven products and acceleration solutions, leveraging AMD technology to accelerate the time to production of complex systems in the cloud or on-premises.
Case Studies
*All performance and cost savings claims are provided by the company or organization mentioned herein and have not been independently verified by AMD. Performance and cost benefits are impacted by a variety of variables. Results herein are specific to the respective organization or company and may not be typical. GD-181.
Resources
Footnotes- GD-183: AMD Infinity Guard features vary by EPYC Processor generations. Infinity Guard security features must be enabled by server OEMs and/or Cloud Service Providers to operate. Check with your OEM or provider to confirm support of these features. Learn more about Infinity Guard at https://www.amd.com/en/technologies/infinity-guard.
- SP5C-001: Testing as of 8.13.2023 using an AMD benchmark based on the Quantlib 1.30 library. The AMD benchmark differs from the upstream Quantlib benchmark and as such results are not comparable.
Testing measured QuantLib 8400 tasks benchmark comparison based on AMD internal testing on AWS EC2 M6 and m6a instances using AMD AOCC4.1 and OneAPI 2023.0.0 Cloud performance results presented are based on the test date in the configuration and are in alignment with AMD internal bare-metal testing factoring in cloud service provider overhead. Results may vary due to changes to the underlying configuration, and other conditions such as the placement of the VM and its resources, optimizations by the cloud service provider, accessed cloud regions, co-tenants, and the types of other workloads exercised at the same time on the system. AWS On-Demand pricing source: https://aws.amazon.com/ec2/pricing/on-demand/ (US-East 2)
Comparisons: AMD Instances vs alternative x86 instances Arithmetic Perf/$ Uplift Cost reduction
m6a.8xlarge vs m6i.8xlarge 32% 24.5%
m6a.12xlarge vs m6i.12xlarge 30% 23.3%
m6a.16xlarge vs m6i.16xlarge 46% 31.3%
m6a.24xlarge vs m6i.24xlarge 16% 13.7%
m6a.32xlarge vs m6i.32xlarge 41% 28.8%
m6a.48xlarge vs m6i.32xlarge 14% 12.5%
hpc6a.48xlarge vs m6i.32xlarge 216% 68.3%
Median 32% 25%
Average 56% 29% - SP5C-002: Testing as of 8.13.2023 using an AMD benchmark based on the Quantlib 1.30 library. The AMD benchmark differs from the upstream Quantlib benchmark and as such results are not comparable.
Testing measured QuantLib 8400 tasks benchmark comparison based on AMD internal testing on Azure Standard D-series v5 and HB120 v3 instances and using AMD AOCC4.1 and OneAPI 2023.0.0. Cloud performance results presented are based on the test date in the configuration and are in alignment with AMD internal bare-metal testing factoring in cloud service provider overhead. Results may vary due to changes to the underlying configuration, and other conditions such as the placement of the VM and its resources, optimizations by the cloud service provider, accessed cloud regions, co-tenants, and the types of other workloads exercised at the same time on the system. Azure Pay-as-you-go pricing https://azure.microsoft.com/en-us/pricing/details/virtual-machines/linux/#pricing.
Comparisons: AMD Instances vs alternative x86 instances Perf/$ Uplift Cost reduction
Standard_D32as_v5 vs Standard_D32s_v5 29% 22.4%
Standard_D48as_v5 vs Standard_D48s_v5 38% 27.8%
Standard_D64as_v5 vs Standard_D64s_v5 22% 18.1%
Standard_D96as_v5 vs Standard_D96s_v5 28% 21.6%
Standard_HB120rs_v3 vs Standard_D96s_v5 172% 63.2%
Median 29% 22%
Average 58% 31% - PHXP-70: Testing as of 12/27/23 by BOXX Technologies, comissioned by AMD, utilizing Dell Latitude 7340 with Intel Core i7 1365U processor, Intel Integrated graphics, 16GB RAM, 256GB NVMe SSD and Windows 11 Pro, Dell XPS 13+ with Intel Core i7 1360P processor, Intel Integrated graphics, 16GB RAM, 1TB NVMe SSD and Windows 11 Pro, and Lenovo ThinkPad Z13 Gen2 with AMD Ryzen 7 PRO 7840U processor, Integrated Radeon Graphics, 32GB RAM, 1TB NVMe SSD, Windows 11 Pro. Using the following tests: Cinebench R23 1T, Cinebench R23 nT, 3DMark Night Raid Graphics, Passmark 11 Overall, Passmark 11 CPU Mark, PCMark 10 Extended, Puget Adobe Photoshop, and Puget Adobe Premiere. PCMark and 3DMark are registered trademarks of UL Solutions. PC manufacturers may vary configurations yielding different results. Results may vary.
- GD-183: AMD Infinity Guard features vary by EPYC Processor generations. Infinity Guard security features must be enabled by server OEMs and/or Cloud Service Providers to operate. Check with your OEM or provider to confirm support of these features. Learn more about Infinity Guard at https://www.amd.com/en/technologies/infinity-guard.
- SP5C-001: Testing as of 8.13.2023 using an AMD benchmark based on the Quantlib 1.30 library. The AMD benchmark differs from the upstream Quantlib benchmark and as such results are not comparable.
Testing measured QuantLib 8400 tasks benchmark comparison based on AMD internal testing on AWS EC2 M6 and m6a instances using AMD AOCC4.1 and OneAPI 2023.0.0 Cloud performance results presented are based on the test date in the configuration and are in alignment with AMD internal bare-metal testing factoring in cloud service provider overhead. Results may vary due to changes to the underlying configuration, and other conditions such as the placement of the VM and its resources, optimizations by the cloud service provider, accessed cloud regions, co-tenants, and the types of other workloads exercised at the same time on the system. AWS On-Demand pricing source: https://aws.amazon.com/ec2/pricing/on-demand/ (US-East 2)
Comparisons: AMD Instances vs alternative x86 instances Arithmetic Perf/$ Uplift Cost reduction
m6a.8xlarge vs m6i.8xlarge 32% 24.5%
m6a.12xlarge vs m6i.12xlarge 30% 23.3%
m6a.16xlarge vs m6i.16xlarge 46% 31.3%
m6a.24xlarge vs m6i.24xlarge 16% 13.7%
m6a.32xlarge vs m6i.32xlarge 41% 28.8%
m6a.48xlarge vs m6i.32xlarge 14% 12.5%
hpc6a.48xlarge vs m6i.32xlarge 216% 68.3%
Median 32% 25%
Average 56% 29% - SP5C-002: Testing as of 8.13.2023 using an AMD benchmark based on the Quantlib 1.30 library. The AMD benchmark differs from the upstream Quantlib benchmark and as such results are not comparable.
Testing measured QuantLib 8400 tasks benchmark comparison based on AMD internal testing on Azure Standard D-series v5 and HB120 v3 instances and using AMD AOCC4.1 and OneAPI 2023.0.0. Cloud performance results presented are based on the test date in the configuration and are in alignment with AMD internal bare-metal testing factoring in cloud service provider overhead. Results may vary due to changes to the underlying configuration, and other conditions such as the placement of the VM and its resources, optimizations by the cloud service provider, accessed cloud regions, co-tenants, and the types of other workloads exercised at the same time on the system. Azure Pay-as-you-go pricing https://azure.microsoft.com/en-us/pricing/details/virtual-machines/linux/#pricing.
Comparisons: AMD Instances vs alternative x86 instances Perf/$ Uplift Cost reduction
Standard_D32as_v5 vs Standard_D32s_v5 29% 22.4%
Standard_D48as_v5 vs Standard_D48s_v5 38% 27.8%
Standard_D64as_v5 vs Standard_D64s_v5 22% 18.1%
Standard_D96as_v5 vs Standard_D96s_v5 28% 21.6%
Standard_HB120rs_v3 vs Standard_D96s_v5 172% 63.2%
Median 29% 22%
Average 58% 31% - PHXP-70: Testing as of 12/27/23 by BOXX Technologies, comissioned by AMD, utilizing Dell Latitude 7340 with Intel Core i7 1365U processor, Intel Integrated graphics, 16GB RAM, 256GB NVMe SSD and Windows 11 Pro, Dell XPS 13+ with Intel Core i7 1360P processor, Intel Integrated graphics, 16GB RAM, 1TB NVMe SSD and Windows 11 Pro, and Lenovo ThinkPad Z13 Gen2 with AMD Ryzen 7 PRO 7840U processor, Integrated Radeon Graphics, 32GB RAM, 1TB NVMe SSD, Windows 11 Pro. Using the following tests: Cinebench R23 1T, Cinebench R23 nT, 3DMark Night Raid Graphics, Passmark 11 Overall, Passmark 11 CPU Mark, PCMark 10 Extended, Puget Adobe Photoshop, and Puget Adobe Premiere. PCMark and 3DMark are registered trademarks of UL Solutions. PC manufacturers may vary configurations yielding different results. Results may vary.