Red Hat ブログ
Last week, Intel launched the new family of Intel® Xeon® scalable processors with new features such as Intel Advanced Vector Extensions 512 (Intel AVX-512), which boost performance of computationally intensive tasks, a new Intel Mesh Architecture for reduced system latency, Intel QuickAssist Technology for hardware acceleration of cryptography and data compression operations and integrated high-speed fabric with Intel Omni-Path Architecture. According to Intel,
the combination of these features delivered 1.65x higher performance, on average, compare to the prior generation of processors and resulted in a long list of world records on various benchmarks.
Red Hat enabled the support for the new Intel® Xeon® Scalable processors in Red Hat Enterprise Linux Red Hat Enterprise Linux 7.3 and 6.9. It is also possible to “test drive” these new systems with Red Hat Enterprise Linux 7.4, currently in Beta. Compare to previous generation, the new generation of CPUs offer up to 28 cores and 56 threads per processor, significant increase in memory speed and the number of DDR4 channels, as well as additional PCIe lanes for better expansion capabilities.
Red Hat and Intel have enjoyed a long history of collaboration across a full spectrum of enterprise IT - covering a wide range of use cases, from applications running on physical servers to virtualized and cloud-based deployments. During months leading to this announcement Red Hat Performance team has validated performance improvements ranging from 1.2x to 1.6x across many key industry-standard workloads, so it should not come as a surprise that Red Hat Enterprise Linux was used by our Original Equipment Manufacturer (OEM) partners to produce 17 new world record results. See table below.
|Significance||OEM Platform||Category and Benchmark name||OS version|
|2-socket world record||Cisco UCS C240 M5||Server-side Java: SPECjbb*2015 Composite Maxl-jOPS||Red Hat Enterprise Linux 7.3|
|2-socket world record||Cisco UCS C240 M5||Server-side Java: SPECjbb*2015 Multi-JVM Critical-jOPS||Red Hat Enterprise Linux 7.3|
|2-socket world record (Linux)||Dell PowerEdge R740xd||Business Processing - Enterprise Resource Planning (ERP): Two-tier SAP Sales and Distribution||Red Hat Enterprise Linux 7.3|
|4-socket x86 world record||Dell PowerEdge R940||Business Processing - Enterprise Resource Planning (ERP): Two-tier SAP Sales and Distribution||Red Hat Enterprise Linux 7.3|
|1-socket x86 World record||HPE ProLiant DL380 Gen10||General Computing: SPECint_base2006 (SPEED)||Red Hat Enterprise Linux 7.3|
|4-socket x86 World record||HPE ProLiant DL580 Gen 10||General Computing:
|Red Hat Enterprise Linux 7.3|
|2-socket x86 World record||HPE ProLiant DL380 Gen10||Big Data Analytics: TPC Benchmark Express Big Bench @ 10000GB||Red Hat Enterprise Linux 7.3|
|2-socket x86 World record||HPE ProLiant ML380 Gen10||Server-side Java: SPECjbb2015 Distributed Critical-jOPS||Red Hat Enterprise Linux 7.3|
|4-socket x86 World record||HPE Synergy 660 Gen10||Server-side Java: SPECjbb2015 Composite Critical-jOPS||Red Hat Enterprise Linux 7.3|
|4-socket x86 World record||HPE Synergy 660 Gen10||Server-side Java: SPECjbb2015 Distributed Critical-jOPS||Red Hat Enterprise Linux 7.3|
|4-socket x86 world record||HPE Synergy 660 Gen10||Server-side Java: SPECjbb2015 Multi-JVM Max-jOPS||Red Hat Enterprise Linux 7.3|
|2-socket x86 world record||Lenovo ThinkSystem SR650||Server-side Java: SPECjbb2015 Composite critical-jOPS||Red Hat Enterprise Linux 7.3|
|8-socket x86 world record||Lenovo ThinkSystem SR950||Server-side Java: SPECjbb2015 Distributed critical-jOPS||Red Hat Enterprise Linux 7.3|
|8-socket x86 World record||Lenovo ThinkSystem SR950||General Computing:
|Red Hat Enterprise Linux 7.3|
|2-socket x86 World record||Lenovo ThinkSystem SR650||STAC M3 (Antuco - 15/17)||Red Hat Enterprise Linux 7.3|
|4-socket x86 World record||Lenovo ThinkSystem SR950||STAC M3 (Shasta 9/11 + 2 options)||Red Hat Enterprise Linux 7.3|
|4-socket x86 World record||Huawei 2488H V5||Technical Computing SPECompG_base2007||Red Hat Enterprise Linux 7.3|
These benchmarks cover a broad spectrum of workloads ranging from compute-intensive technical computing problems to Big Data analytics and from enterprise Java applications to ERP and provide solid proof that Red Hat Enterprise Linux excels at handling scalable workloads and thrives in ever-changing environments, as enterprise customers place ever increasing demands on their infrastructure.
Moreover, because of Red Hat’s continued commitment to deliver the latest enterprise computing features, our customers can count on Red Hat Enterprise Linux to be an excellent choice for demanding enterprise workloads running across, physical, virtual or cloud deployments, as they are considering the adoption of these new Xeon-based servers.
Special thanks to our OEM partners that made Red Hat Enterprise Linux their platform of choice for producing these results on new server hardware. We can not emphasise enough the importance of close collaboration across the ecosystem, as many of these records are the culmination of combined silicon, hardware and software improvements.
Enterprises and customers continue to place increasing demands on their infrastructure requiring stable and optimized platform for best performance. As proven by these top performance scores, Red Hat Enterprise Linux runs workloads more efficiently while achieving better results than other operating systems and delivers industry-leading capabilities and the latest technology innovation to our customers and partners, as they are optimizing their IT infrastructure.
All results as of July 11, 2017. SPEC and the benchmark names SPEC CPU, SPECjbb, SPEComp and SPECvirt_sc are registered trademarks of the Standard Performance Evaluation Corporation. For more information about SPEC and its benchmarks see: www.spec.org. TPC is a trademark of the Transaction Processing Performance Council (TPC). For more information about the TPC and it's benchmarks see www.tpc.org. SAP and SAP NetWeaver are the registered trademarks of SAP AG in Germany and in several other countries. See www.sap.com/benchmark for more information.
About the author
Yan Fisher is a Global evangelist at Red Hat where he extends his expertise in enterprise computing to emerging areas that Red Hat is exploring.