» Download all images (ZIP, 817 KB)
This week, Intel announced an unmatched silicon portfolio for 5G infrastructure with the launch of the Intel Atom® P5900, a 10nm system-on-chip (SoC) for wireless base stations, and the introduction of both a structured ASIC for 5G network processing and a 5G network-optimized Ethernet network adapter, offering high-precision, GPS-based network service synchronization.
More: Intel Announces Unmatched Portfolio for 5G Network Infrastructure (News Release) | Intel Reinforces Data Center Leadership with New 2nd-Gen Intel Xeon Scalable Processors (News Byte) | Intel Announces Unmatched Portfolio for 5G Network Infrastructure (Press Kit)
Additionally, Intel announced the expansion of its 2nd gen Intel® Xeon® Scalable processor lineup as the company continues to deliver workload-optimized performance, integrated artificial intelligence (AI) acceleration and hardened security at the platform level.
Launching Intel Atom P5900 Platform
In a highly integrated 10nm SoC, the Intel Atom P5900 delivers what customers require for future-ready 5G base stations, including ultra-low latency, accelerated throughput and precise load balancing.
Performance improvements include up to 5.6 times more secured network communication with the Intel Atom® P5952B processor and Intel® QuickAssist Technology (versus software alone)1 and up to 3.7 times more packet processing with the Intel Atom P5952B processor and Intel® Dynamic Load Balancer (versus software alone)2.
Key features include:
- Up to 24 Intel Atom P5000 processor cores, based on the Tremont microarchitecture, using 10nm process technology.
- L1 cache of 32KB/core, L2 cache of 4.5MB per 4-core cluster and shared LLC cache up to 15MB.
- Base frequency support of up to 2.2 GHz, single-thread performance.
- Memory capacity of up to 128GB of DDR4 up to 2,933 MT/s with server-class reliability, availability and serviceability (RAS) and support for RDIMM, UDIMM, SODIMM and memory down.
- Up to 100 Gbps throughput with integrated Intel Ethernet 800 series technology.
- Up to 100 Gbps security processing with Intel QuickAssist Technology.
- Up to 440 Gbps network switching connectivity support with up to 20 fully integrated Ethernet SerDes.
- Up to 16 lanes of flexible high-speed IO configured as PCIe, SATA and/or USB 3.0.
- Up to 16 lanes of PCIe Gen 3.0.
- Up to 4 USB 2.0 ports, eMMC 5.1, LPC or eSPI.
- Extended temperature support of minus 40 C to 85 C with full dynamic temperature range.
New 2nd Gen Intel Xeon Scalable Processors
The new 2nd Gen Intel Xeon Scalable processors launched today deliver an average of 36% more performance and an average of 42% more performance per dollar3 than the prior generation Intel Xeon Gold and increased value for customers across their cloud, network and edge needs.
- These new processors add more cores, cache and/or frequency, providing more performance and value across a wide range of mainstream price points.
- Industry-leading frequency, with a new processor featuring base frequency of 3.9 GHz and turbo frequency up to 4.5 GHz.
- Continued hardware-based mitigations against side-channel attacks, plus built-in encryption acceleration and key protection technologies.
- Eighteen Intel® Select Solutions will be updated in the coming weeks with new processor options to take advantage of the improved performance and value. Intel Select Solution are workload-optimized configurations for a variety of popular workloads that reduce the guesswork and accelerate time-to-deployment.
- Find additional information and SKU chart on “Intel Reinforces Data Center Leadership with New 2nd Gen Intel Xeon Scalable Processors.”
Introducing “Diamond Mesa,” Intel’s First Next-Generation Structured ASIC for 5G Network Acceleration
Structured ASICs like “Diamond Mesa” provide a minimum-risk optimization path for workloads that do not require the full programmability of FPGAs, targeting double the performance efficiency versus the prior generation, and uniquely position Intel as the only provider delivering a full silicon platform foundation for network infrastructure.
Key features include:
- The first eASIC device to leverage all-new IP from the Intel Programmable Solutions Group, a quad-core ARM A53 processor and secure device manager subsystem.
- Structured ASICs like Diamond Mesa strike a balance between the programmability and fast time-to-market of FPGAs and the efficient performance of purpose-built ASICs.
- Only Intel offers developers a full range of silicon solutions to optimize platforms for performance, cost, power, flexibility and time-to-market – from Intel Xeon processors to purpose-built ASICs.
Introducing the Intel Ethernet 700 Series Network Adapter with Hardware-Enhanced Precision Time Protocol (code-named “Edgewater Channel”)
Intel introduced the Intel® Ethernet 700 Series network adapter with hardware-enhanced Precision Time Protocol (PTP), which is designed for 5G, edge and other use cases with very tight latency and timing requirements.
Intel Ethernet 700 Series is compliant with the IEEE 1588 PTP standard and increases phase accuracy through a combination of hardware enhancements and software optimizations with within 100 nanoseconds or less.
Key features include:
- Onboard high-precision oscillator with up to 5,000 times tighter accuracy: Provides greater phase
accuracy than a standard Ethernet adapter and better holdover between clock interruption. (5000x=20 ppb TCXO oscillator (10 °C to 70 °C) vs. typical 100 ppm crystal)
- Dual SubMiniature version A (SMA) coaxial connectors: Allow connection to external timing sources, such as GPS signaling devices and to timing sinks/recipients, allowing the NIC to act as
a grandmaster and as a timing source for other equipment.
- Software-defined pins (SDPs): User-configurable I/O pins drive the SMAs and provide built-in timing coordination between ports.
- Standard Linux support: Linux PTP utility provides support for multiple PTP profiles and use of SDPs for sync via 1 pulse per second (PPS) input and output signals.
Performance results are based on testing as of dates in configuration and may not reflect all publicly available security updates. See configuration disclosure for details. No product can be absolutely secure. For more complete information about performance and benchmark results, visit www.intel.com/benchmarks. Refer to http://software.intel.com/en-us/articles/optimization-notice for more information regarding performance and optimization choices in Intel software products. Results have been estimated or simulated. Int, Integer Throughput: estimated SPECrate®2017_int_base based on Intel internal measurements; fp, Floating Point Throughput: estimated SPECrate®2017_fp_base based on Intel internal measurements. SPEC®, SPECrate® and SPEC CPU® are registered trademarks of the Standard Performance Evaluation Corporation. See www.spec.org for more information.
1 Get up to 5.6X more packet security throughput with Intel QAT and the integrated switch, versus software: Tested by Intel on 1/27/2020, 1x Intel Atom® P5952B processor (Running under A-3 Eng Silicon 20C, 2.2GHz) on Intel internal Frost Creek platform, 16GB DDR4 2933MHz, OS: Ubuntu 18.04 with Kernel: 5.2.10-rt5, BIOS: JBVLCRB1.86B.0011.D44.1909191126, uCode: 0x90010006, Benchmark: DPDK IPSec (using Intel® QAT) (1420B packet size) (1C/1T/1P) (Estimated), Software: DPDK 19.05, Compiler: GCC 7.3.0 with MKL, Network: Onboard 100GbE NIC compared to 1x Intel Atom® P5952B processor (Running under A-3 Eng Silicon 20C, 2.2GHz) on Intel internal Frost Creek platform, 16GB DDR4 2933MHz, OS: Ubuntu 18.04 with Kernel: 5.2.10-rt5, BIOS: JBVLCRB1.86B.0011.D44.1909191126, uCode: 0x90010006, Benchmark: DPDK IPSec (using software cipher algorithm AES-128-CBC) (1420B packet size) (1C/1T/1P) (Estimated), Software: DPDK 19.05, Compiler: GCC 7.3.0 with MKL, Network: Onboard 100GbE NIC.
2 Achieve up to 3.7X more packet balancing throughput versus software queue management: Tested by Intel on 1/27/2020, 1x Intel Atom® P5952B processor (Running under A-3 Eng Silicon 20C, 2.2GHz) on Intel internal Victoria Canyon platform, 16GB DDR4 2933MHz ,OS: Ubuntu 18.04 with Kernel: 5.2.10-rt5, BIOS: JBVLCRB1.86B.0012.D17.1911070324, uCode: 0x90040006, Benchmark: DPDK Event Device Ordered Scheduling (3 stages) (Estimated), Software: RDK19.11, Compiler: GCC 7.3.0 with MKL, Network: 1x Intel® Ethernet Network Adapter X710-DA4 (10GbE) compared to 1x Intel Atom® P5952B processor (Running under A-3 Eng Silicon 20C, 2.2GHz) on Intel internal Victoria Canyon platform, 16GB DDR4 2933MHz, OS: Ubuntu 18.04 with Kernel: 5.2.10-rt5, BIOS: JBVLCRB1.86B.0012.D17.1911070324, uCode: 0x90040006, Benchmark: DPDK Event Device Ordered Scheduling (3 stages) (Estimated), Software: RDK19.11, Compiler: GCC 7.3.0 with MKL, Network: 1x Intel® Ethernet Network Adapter X710-DA4 (10GbE).
3 Configuration details updated on March 12, 2020, to reflect the latest SPEC guidelines.
36% More Performance and 42% More Performance/Dollar: Geomean of Integer Throughput, Floating Point Throughput, STREAM Triad, and Intel® Distribution for LINPACK Across Ten New 2-socket 2nd Gen Intel® Xeon® Gold Processors Vs. First Generation. 2nd Gen Intel® Xeon® Gold R processors: 1-node, 2x 2nd Gen Intel Xeon Gold processor (62xxR/$$) on Intel Reference platform with 384GB (12 slots / 32 GB / 62xx@2933,52xx@2666) total memory, ucode 0x500002c, HT on for all except off for STREAM (GB/s), LINPACK (GFLOPS/s), Turbo on, with Ubuntu19.10, 5.3.0-24-generic, 6258R/$3950: int=323, fp=262, STREAM=224, LINPACK =3305; 6248R/$2700: int=299, fp=248, STREAM=224, LINPACK =3010; 6246R/$3286: int=238, fp=217, STREAM=225, LINPACK =2394; 6242R/$2529: int=265, fp=231, STREAM=227, LINPACK =2698; 6240R/$2200: int=268, fp=228, STREAM=223, LINPACK=2438; 6238R/$2612: int=287, fp=240, STREAM=222, LINPACK =2545; 6230R/$1894: int=266, fp=227, STREAM=222, LINPACK =2219; 6226R/$1300: int=208, fp=192, STREAM=200, LINPACK =2073; 5220R/$1555: int=257, fp=220, STREAM=210, LINPACK =1610; 5218R/$1273: int=210, fp=188, STREAM=199, LINPACK =1290, test by Intel on 12/25/2019. First Gen Intel® Xeon® Gold processor: 1-node, 2x Intel Xeon Gold processor (61xx/$$) on Intel Reference platform with 384GB (12 slots / 32 GB / 61xx@2666,51xx@2400) total memory, ucode 0x500002c, HT on for all except off for STREAM (GB/s), LINPACK (GFLOPS/s), Turbo on, with Ubuntu19.10, 5.3.0-24-generic, 6152/$3655: int=224, fp=198, STREAM=200, LINPACK =1988; 6148/$3072: int=225, fp=198, STREAM=197, LINPACK =2162; 6146/$3286: int=161, fp=175, STREAM=185, LINPACK =1896; 6142/$2946: int=193, fp=176, STREAM=185, LINPACK =1895; 6140/$2445: int=202, fp=183, STREAM=188, LINPACK =1877; 6138/$2612: int=189, fp=195, STREAM=189, LINPACK =1976; 6130/$1894: int=172, fp=165, STREAM=185, LINPACK =1645; 6126(proj)/$1776: int=141, fp=157, STREAM=170, LINPACK =1605; 5120(proj)/$1555: int=148, fp=148, STREAM=159, LINPACK =924, 5118/$1273: int=134, fp=132, STREAM=149, LINPACK =818, test by Intel on 2/18/2020. Your costs and results may vary.