Relational Database Management Systems (RDBMS) have a half century of history. They laid the foundation for modern business computing. Today, many types of data stores and data management systems are deployed. Still, RDBMS remains the core of enterprise applications for transaction processing, business analytics, and decision support systems – all part of the enterprise business.
While the foundational aspects of RDBMS remain the same, many enterprises demand NoSQL systems, Object Stores and others for real-time processing over vast amounts of data. With “Data-driven decision making” an increasingly common theme across businesses today, smart data processing helps to bring mission-critical business insights to the fingertips. Plus, thanks to the technological innovations for enabling the democratization of data, many applications and data that were once available only for resource-rich enterprises, are now available to businesses of all sizes.
Trading, fraud and anomaly detection, recommendation engines, logistics management, transportation route planning, financial modeling, activity trackers, and many more applications require extreme compute power to consume a huge amount of data in real-time in order to bring insights to modern businesses. These businesses have realized that real-time analytics capabilities can provide a competitive advantage in today’s data-driven world. Their urgency for faster competitive insights from data is driving greater demand for computing power in enterprise data centers across the globe. In response to this growing demand, we at AMD have introduced three new processors in the 2nd Generation AMD EPYC Processor family.
AMD EPYC 7Fx2 Processors
With up to 500MHz of additional base frequency over existing 2nd Gen EPYC processors and large amounts of L3 cache per core, AMD EPYC 7Fx2 features the world’s highest per-core performance x86 CPU1. These new processors are taking computing performance to new heights by pushing the limits of computing throughput in every AMD “Zen 2” core to deliver the most performance. With the increased L3 cache that helps to keep data close to the execution core, the new AMD EPYC 7Fx2 processors bring leadership performance to relational databases for transactional processing and real-time analytics.
Cores / Processor
Maximum Memory / Socket (DDR4-3200)
PCIe Gen4 Lanes / System
AMD EPYC 7F32
AMD EPYC 7F52
At AMD, we have worked closely with ecosystem partners in optimizing the performance of leading RDBMS on AMD EPYC processors to offer companies like yours the best performance and low TCO.
Another unique aspect of the AMD 7Fx2 EPYC processor is its ability to support up to 4TB memory per processor. That is 8TB of memory in a standard two-processor system places a large amount of data close to the processors enabling real-time analytics over large datasets. In addition, the 2nd Gen EPYC family’s industry-first support for PCIe 4 enables high-speed network connectivity, NVMe storage and connectivity to accelerators (FPGA, GPU, etc.).
Let me highlight three examples.
1) Microsoft SQL Server is a leading RDBMS. SQL Server 2019 builds on previous releases to grow as a platform that gives you choices of development languages, data types, on-premises or cloud environments, and operating systems.
The results below demonstrate how AMD EPYC based systems deliver high performance for Online Transaction Processing (OLTP) performance with Microsoft SQL Server 2019.
2) Another example I’d like to bring up here is with AMD EPYC processors with Oracle Database, a multi-model database management system. Oracle Database continues to deliver leading-edge innovations, including machine learning, to enable self-driving data management. This enterprise-proven, database cloud service is designed to support mixed workloads through any deployment strategy, on-premises, or in the cloud.
Best performance for database applications is the synergic outcome of number of vCPUs, size of memory, storage with high throughput IOPS and network speed configured on the instance types used on the infrastructure side. A performance leadership similar to Microsoft SQL Server was found when we tested AMD EPYC 7Fx2 Processors with Oracle Database 19c on RHEL 7.7 using HammerDB.
3) I can quote many more examples on the performance leadership of AMD EPYC 7Fx2 family of processors but will bring up one more here. Read the results from the test on MySQL using IBM Cloud Bare Metal Servers to see how capable IBM Cloud Bare Metal Instances are at optimizing the I/O throughput for database applications.
While we focused on bringing the highest possible performance to your data center, we kept a laser focus on helping ensure your cost efficiency. AMD EPYC processors enable sustained transaction throughput and linear scaling that allows you to right-size the compute power for your application needs to more easily achieve a lower total cost of ownership -- you pay only for the cores you actually need and optimize your core-based software licensing model costs.
With innovative architecture and security features, the new AMD EPYC 7Fx2 processors can provide enterprise data centers running transactional databases on SQL Server with up to 10% higher TPM per-core performance at an estimated 35% lower CPU cost per TPM2. We are here to help you derive faster insights from your data center.
Contact your preferred IT infrastructure provider and start accelerating your time to insight.
- Highest per core performance in the world based on EPYC 7F32 (8-cores) having the highest SPECrate 2017_fp_base score divided by total core count, of all SPEC publications as of 4/14/2020. 1x EPYC 7F32 (8-cores) scoring 12.875 base result per core (103 SPECrate 2017_fp_base/16 total cores, www.spec.org/cpu2017/results/res2020q2/cpu2017-20200316-21228.pdf) compared to the next highest result 1x AMD EPYC 7262 (8-cores) scoring 11.54 base result per core (92.3 SPECrate 2017_fp_base/8 total cores, http://spec.org/cpu2017/results/res2020q1/cpu2017-20191220-20435.pdf) See www.spec.org/cpu2017/results for full ranking. SPEC and SPECrate are trademarks of the Standard Performance Evaluation Corporation. Learn more at www.spec.org ROM-570
- Testing as of 3.20.2020 by AMD Performance Labs. Up to 10% higher SQL Server tpm-per-core/lower cost-per-tpm. Workload: HammerDB 3.3 (TPC-C profile - The workload is derived from the TPC-C Benchmark, and as such is not comparable to published TPC-C Benchmark results, as the HammerDB OLTP workload results do not comply with the TPC-C benchmark). Configurations: 2x EPYC 7F32 (16C total, $4200) scoring 2,692,958 tpm (168,310 tpm per core at $0.00156 per tpm). 2x Xeon Gold 6244 (16C total, $5850) scoring 2,446,340 tpm (152,896 tpm per core at $0.00239 per tpm). Results may vary. ROM-572