Most Feature-Rich and Secure Xeon Platform Yet
Signifying the biggest platform transformation Intel has delivered, not only is 4th Gen Xeon a marvel of acceleration, but it is also an achievement in manufacturing, combining up to four Intel 7-built tiles on a single package, connected using Intel EMIB (embedded multi-die interconnect bridge) packaging technology and delivering new features including increased memory bandwidth with DDR5, increased I/O bandwidth with PCIe5.0 and Compute Express Link (CXL) 1.1 interconnect.
At the foundation of it all is security. With 4th Gen Xeon, Intel is delivering the most comprehensive confidential computing portfolio of any data center silicon provider in the industry, enhancing data security, regulatory compliance and data sovereignty. Intel remains the only silicon provider to offer application isolation for data center computing with Intel® Software Guard Extensions (Intel® SGX), which provides today’s smallest attack surface for confidential computing in private, public and cloud-to-edge environments. Additionally, Intel’s new virtual-machine (VM) isolation technology, Intel® Trust Domain Extensions (Intel® TDX), is ideal for porting existing applications into a confidential environment and will debut with Microsoft Azure, Alibaba Cloud, Google Cloud and IBM Cloud.
Finally, the modular architecture of 4th Gen Xeon allows Intel to offer a wide range of processors across nearly 50 targeted SKUs for customer use cases or applications, from mainstream general-purpose SKUs to purpose-built SKUs for cloud, database and analytics, networking, storage, and single-socket edge use cases. The 4th Gen Xeon processor family is On Demand-capable and varies in core count, frequency, mix of accelerators, power envelope and memory throughput as is appropriate for target use cases and form factors addressing customers’ real-world requirements.
SKU TABLE: SKUs for 4th Gen Xeon and Intel Xeon CPU Max Series
About Intel
Intel (Nasdaq: INTC) is an industry leader, creating world-changing technology that enables global progress and enriches lives. Inspired by Moore’s Law, we continuously work to advance the design and manufacturing of semiconductors to help address our customers’ greatest challenges. By embedding intelligence in the cloud, network, edge and every kind of computing device, we unleash the potential of data to transform business and society for the better. To learn more about Intel’s innovations, go to newsroom.intel.com and intel.com.
1 Geomean of following workloads: RocksDB (IAA vs ZTD), ClickHouse (IAA vs ZTD), SPDK large media and database request proxies (DSA vs out of box), Image Classification ResNet-50 (AMX vs VNNI), Object Detection SSD-ResNet-34 (AMX vs VNNI), QATzip (QAT vs zlib)
2 1-node, Intel Reference Validation Platform, 2x Intel® Xeon 8480+ (56C, 2GHz, 350W TDP), HT On, Turbo ON, Total Memory: 1 TB (16 slots/ 64GB/ 4800 MHz), 1x P4510 3.84TB NVMe PCIe Gen4 drive, BIOS: 0091.D05, (ucode:0x2b0000c0), CentOS Stream 8, 5.15.0-spr.bkc.pc.10.4.11.x86_64, Java Perf/Watt w/ openjdk-11+28_linux-x64_bin, 112 instances, 1550MB Initial/Max heap size, Tested by Intel as of Oct 2022.
3 ResNet50 Image Classification
New Configuration: 1-node, 2x pre-production 4th Gen Intel® Xeon® Scalable 8490H processor (60 core) with Intel® Advanced Matrix Extensions (Intel AMX), on pre-production SuperMicro SYS-221H-TNR with 1024GB DDR5 memory (16x64 GB), microcode 0x2b0000c0, HT On, Turbo On, SNC Off, CentOS Stream 8, 5.19.16-301.fc37.x86_64, 1x3.84TB P5510 NVMe, 10GbE x540-AT2, Intel TF 2.10, AI Model=Resnet 50 v1_5, best scores achieved: BS1 AMX 1 core/instance (max. 15ms SLA), using physical cores, tested by Intel November 2022. Baseline: 1-node, 2x production 3rd Gen Intel Xeon Scalable 8380 Processor ( 40 cores) on SuperMicro SYS-220U-TNR, DDR4 memory total 1024GB (16x64 GB), microcode 0xd000375, HT On, Turbo On, SNC Off, CentOS Stream 8, 5.19.16-301.fc37.x86_64, 1x3.84TB P5510 NVMe, 10GbE x540-AT2, Intel TF 2.10, AI Model=Resnet 50 v1_5, best scores achieved: BS1 INT8 2 cores/instance (max. 15ms SLA), using physical cores, tested by Intel November 2022.
For a 50 server fleet of 3rd Gen Xeon 8380 (RN50 w/DLBoost), estimated as of November 2022:
CapEx costs: $1.64M
OpEx costs (4 year, includes power and cooling utility costs, infrastructure and hardware maintenance costs): $739.9K
Energy use in kWh (4 year, per server): 44627, PUE 1.6
Other assumptions: utility cost $0.1/kWh, kWh to kg CO2 factor 0.42394
For a 17 server fleet of 4th Gen Xeon 8490H (RN50 w/AMX), estimated as of November 2022:
CapEx costs: $799.4K
OpEx costs (4 year, includes power and cooling utility costs, infrastructure and hardware maintenance costs): $275.3K
Energy use in kWh (4 year, per server): 58581, PUE 1.6
AI -- 55% lower TCO by deploying fewer 4th Gen Intel® Xeon® processor-based servers to meet the same performance requirement. See [E7] at intel.com/processorclaims: 4th Gen Intel Xeon Scalable processors. Results may vary.
Database -- 52% lower TCO by deploying fewer 4th Gen Intel® Xeon® processor-based servers to meet the same performance requirement. See [E8] at intel.com/processorclaims: 4th Gen Intel Xeon Scalable processors. Results may vary.
HPC -- 66% lower TCO by deploying fewer Intel® Xeon® CPU Max processor-based servers to meet the same performance requirement. See [E9] at intel.com/processorclaims: 4th Gen Intel Xeon Scalable processors. Results may vary.
4 Geomean of HP Linpack, Stream Triad, SPECrate2017_fp_base est, SPECrate2017_int_base est. See [G2, G4, G6] at intel.com/processorclaims: 4th Gen Intel Xeon Scalable.
5 Up to 10x higher PyTorch real-time inference performance with built-in Intel® Advanced Matrix Extensions (Intel® AMX) (BF16) vs. the prior generation (FP32)
PyTorch geomean of ResNet50, Bert-Large, MaskRCNN, SSD-ResNet34, RNN-T, Resnext101.
6 Up to 10x higher PyTorch training performance with built-in Intel® Advanced Matrix Extensions (Intel® AMX) (BF16) vs. the prior generation (FP32)
PyTorch geomean of ResNet50, Bert-Large, DLRM, MaskRCNN, SSD-ResNet34, RNN-T.
7 Estimated as of 8/30/2022 based on 4th generation Intel® Xeon® Scalable processor architecture improvements vs 3rd generation Intel® Xeon® Scalable processor at similar core count, socket power and frequency on a test scenario using FlexRAN™ software. Results may vary.
8 Up to 95% fewer cores and 2x higher level 1 compression throughput with 4th Gen Intel Xeon Platinum 8490H using integrated Intel QAT vs. prior generation.
8490H: 1-node, pre-production platform with 2x 4th Gen Intel® Xeon Scalable Processor (60 core) with integrated Intel QuickAssist Accelerator (Intel QAT), QAT device utilized=8(2 sockets active), with Total 1024GB (16x64 GB) DDR5 memory, microcode 0xf000380, HT On, Turbo Off, SNC Off, Ubuntu 22.04.1 LTS, 5.15.0-47-generic, 1x 1.92TB Intel® SSDSC2KG01, QAT v20.l.0.9.1 , QATzip v1.0.9 , ISA-L v2.3.0, tested by Intel September 2022.
8380: 1-node, 2x 3rd Gen Intel Xeon Scalable Processors( 40 cores) on Coyote Pass platform, DDR4 memory total 1024GB (16x64 GB), microcode 0xd000375, HT On, Turbo Off, SNC Off, Ubuntu 22.04.1 LTS, 5.15.0-47-generic, 1x 1.92TB Intel SSDSC2KG01,QAT v1.7.l.4.16, QATzip v1.0.9 , ISA-L v2.3.0, tested by Intel October 2022.
9 Up to 3x higher RocksDB performance with 4th Gen Intel Xeon Platinum 8490H using integrated Intel IAA vs. prior generation.
8490H: 1-node, pre-production Intel platform with 2x 4th Gen Intel Xeon Scalable Processor (60 cores) with integrated Intel In-Memory Analytics Accelerator (Intel IAA), HT On, Turbo On, Total Memory 1024GB (16x64GB DDR5 4800), microcode 0xf000380, 1x 1.92TB INTEL SSDSC2KG01, Ubuntu 22.04.1 LTS, 5.18.12-051812-generic, QPL v0.1.21,accel-config-v3.4.6.4, ZSTD v1.5.2, RocksDB v6.4.6 (db_bench), tested by Intel September 2022.
8380: 1-node, 2x 3rd Gen Intel Xeon Scalable Processors( 40 cores) on Coyote Pass platform, HT On, Turbo On, SNC Off, Total Memory 1024GB (16x64GB DDR4 3200), microcode 0xd000375, 1x 1.92TB INTEL SSDSC2KG01, Ubuntu 22.04.1 LTS, 5.18.12-051812-generic, ZSTD v1.5.2, RocksDB v6.4.6 (db_bench), tested by Intel October 2022.
10 Intel® Xeon® 8380: Test by Intel as of 10/7/2022. 1-node, 2x Intel® Xeon® 8380 CPU, HT On, Turbo On, Total Memory 256 GB (16x16GB 3200MT/s DDR4), BIOS Version SE5C620.86B.01.01.0006.2207150335, ucode revision=0xd000375, Rocky Linux 8.6, Linux version 4.18.0-372.26.1.el8_6.crt1.x86_64, YASK v3.05.07
Intel® Xeon® CPU Max Series: Test by Intel as of ww36’22. 1-node, 2x Intel® Xeon® CPU Max SeriesHT On, Turbo On, SNC4, Total Memory 128 GB (8x16GB HBM2 3200MT/s), BIOS Version SE5C7411.86B.8424.D03.2208100444, ucode revision=0x2c000020, CentOS Stream 8, Linux version 5.19.0-rc6.0712.intel_next.1.x86_64+server, YASK v3.05.07.
11 Up to 20% system power savings utilizing 4th Gen Xeon Scalable with Optimized Power mode on vs off on select workloads including SpecJBB, SPECINT and NIGNX key handshake.
12 AMD Milan: Tested by Numenta as of 11/28/2022. 1-node, 2x AMD EPYC 7R13 on AWS m6a.48xlarge, 768 GB DDR4-3200, Ubuntu 20.04 Kernel 5.15, OpenVINO 2022.3, BERT-Large, Sequence Length 512, Batch Size 1
