2nd Gen Intel® Xeon® Platinum 9200 Processors

Architected to deliver performance leadership across the widest range of demanding workloads with unprecedented native DDR4 memory bandwidth and more memory channels than any other Intel® Xeon® processor. The Intel® Xeon® Platinum 9200 processor delivers unprecedented performance scaling up to 112 cores in a two socket system with 24 channels of memory representing ~400GB/sec of memory bandwidth.

Leadership Performance

General & Technical Compute

With Intel® Advanced Vector Extensions 512 (Intel® AVX-512), get 2x better FLOPS than Intel® Advanced Vector Extensions 2 (Intel® AVX2) and utilize 2x memory channels per CPU vs prior generation.

HPC at Scale

Unprecedented performance per rack which is especially compelling for large scale HPC compute clusters.

Artificial Intelligence

Up to 30x improvement in inference performance on Intel® Xeon® Platinum 9282 processor (56 cores) w/ Intel® Deep Learning Boost (Intel® DL Boost) for ResNet-50 (image classification workload) vs. Intel® Xeon® Platinum 8180 processor at launch.

Where to Buy


Experience the power of a system based on 2nd Gen Intel® Xeon® Platinum 9200 processors.

Find a partner ›

General & Technical Compute

General computing workloads, such as SPECrate2017*, characterize compute, memory, and software characteristics which serves as a yardstick for a very wide variety of server usages such as compression, compilation, AI, gene sequencing, and many other integer-based applications workloads. Technical computing is the application of the mathematical and computational principles of scientific computing to solve practical problems of industrial interest. The distinction between technical computing and scientific computing is similar to the distinction between high performance computing (HPC) and high-performance technical computing. Technical computing may involve the use of large-scale high-performance systems.

Configs for General & Technical Compute

1.72X Integer throughput - base: 1-node, 2x Intel® Xeon® Platinum 9282 processor, https://www.spec.org/cpu2017/results/res2019q2/cpu2017-20190318-11207.html, SPECrate2017_int_base, score: 628, test by Intel on Mar 2019. 1-node, 2x Intel® Xeon® Platinum 8180 processor, https://www.spec.org/cpu2017/results/res2019q2/cpu2017-20190429-12779.html, SPECrate2017_int_base, score: 364, test by ASUSTeK on Apr 2019.

1.72X Integer throughput - peak: 1-node, 2x Intel® Xeon® Platinum 9282 processor, https://www.spec.org/cpu2017/results/res2019q2/cpu2017-20190513-13798.html, SPECrate2017_int_peak, score: 657, test by Intel on May 2019. 1-node, 2x Intel® Xeon® Platinum 8180 processor, https://www.spec.org/cpu2017/results/res2019q2/cpu2017-20190429-12779.html, SPECrate2017_int_peak, score: 381, test by ASUSTeK on Apr 2019.

1.73X Floating point throughput – base: 1-node, 2x Intel® Xeon® Platinum 9282 processor, https://www.spec.org/cpu2017/results/res2019q2/cpu2017-20190318-11208.html, SPECrate2017_fp_base, score: 522, test by Intel on Mar 2019. 1-node, 2x Intel® Xeon® Platinum 8180 processor, https://www.spec.org/cpu2017/results/res2019q2/cpu2017-20190429-12775.html, SPECrate2017_fp_base, score: 301, test by ASUSTeK on Apr 2019.

1.81X Floating point throughput – peak: 1-node, 2x Intel® Xeon® Platinum 9282 processor, https://www.spec.org/cpu2017/results/res2019q2/cpu2017-20190513-13797.html, SPECrate2017_fp_peak, score: 554, test by Intel on May 2019. 1-node, 2x Intel® Xeon® Platinum 8180 processor, https://www.spec.org/cpu2017/results/res2019q2/cpu2017-20190429-12775.html, SPECrate2017_fp_peak, score: 306, test by ASUSTeK on Apr 2019.

1.97X Linpack performance: 1-node, 2x Intel® Xeon® Platinum 9282 processor on Walker Pass with 768 GB (24x 32GB 2933) total memory, ucode 0x400000A on RHEL7.6, 3.10.0-957.el7.x86_64, IC19u1, AVX512, HT off, Turbo on, score: 6411, test by Intel on 2/16/2019. 1-node, 2x Intel® Xeon® Platinum 8180 processor, on Wolf Pass with 384 GB (12 X 32GB 2666) total memory, ucode 0x200004D on RHEL7.6, 3.10.0-957.el7.x86_64, IC19u1, AVX512, HT off, Turbo on, score: 3238, test by Intel on 1/29/2019.

1.99X Stream Triad: 1-node, 2x Intel® Xeon® Platinum 9282 processor on Walker Pass with 768 GB (24x 32GB 2933) total memory, ucode 0x400000A on RHEL7.6, 3.10.0-957.el7.x86_64, IC19u1, AVX512, HT off, Turbo on, score: 407, test by Intel on 2/16/2019. 1-node, 2x Intel® Xeon® Platinum 8180 processor, on Wolf Pass with 384 GB (12 X 32GB 2666) total memory, ucode 0x200004D on RHEL7.6, 3.10.0-957.el7.x86_64, IC19u1, AVX512, HT off, Turbo on, score: 204, test by Intel on 1/29/2019.

Technical Compute Benchmarks

Open MP Performance with SPEC OMP2012*

Containing a suite of 14 end-user applications that focuses on parallel computing performance using the OpenMP parallelism standard, this benchmark is based on compute-intensive applications that emphasize the performance of the computer processor, memory architecture, parallel support libraries, and the compilers. The performance metric shown is calculated from the geometric mean of 14 normalized ratios when the applications are compiled with consistent compiler options across all programs of a given language (eight use Fortran, five use C, and one uses C++).

SPEC OMP2012* Benchmark Leadership Configuration Details

1.53X 2S SPECompG_base2012: 1-node, 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, Intel® C++ Compiler 2019u2; 2 threads per core; HT=ON, Turbo=ON, geomean: 36 test by Intel on 5/14/2019; results submitted June 2019 and pending review by SPEC. 1-node, 2x Intel® Xeon® Platinum 8180 processor, source: https://www.spec.org/omp2012/results/res2017q4/omp2012-20171010-00130.html, SPECompG_base2012, geomean: 23.5, test by Cisco on Oct 2017.

1.68X 2S SPECompG_peak2012: 1-node, 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, , geomean: 43 test by Intel on 5/14/2019; results submitted June 2019 and pending review by SPEC. 1-node, 2x Intel® Xeon® Platinum 8180 processor, source: https://www.spec.org/omp2012/results/res2017q3/omp2012-20170718-00108.html, SPECompG_peak2012, geomean: 25.5, test by Intel on Jul 2017.

MPI Performance with SPEC MPI2007*

Containing a suite of 18 end-user applications that evaluate MPI-parallel, floating point, compute-intensive performance across a wide range of cluster and SMP hardware. This benchmark is based on compute-intensive applications that emphasize the performance of the computer processor, memory architecture, communication interconnect, the compilers, and the shared file system and are developed in compliance with Message-Passing Interface (MPI) 2.1 standard. The performance metric shown is calculated from the geometric mean of 18 normalized ratios when the applications are compiled with consistent compiler options across all programs of a given language (including Fortran, C, and C++).

SPEC MPI2007* Benchmark Leadership Configuration Details

1-node, 2x Intel® Xeon® Platinum 8180 processor, source: http://spec.org/mpi2007/results/res2018q3/mpi2007-20180828-00601.html, SPECmpiM_base2007, geomean score: 18.1, test by Lenovo on Aug 2018.

1-node, 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, Intel® C++ Compiler 2018u2, Intel® MPI Library 2019u2; 1 thread per core; HT=OFF; Turbo=ON, geomean: 30.6 test by Intel on 5/15/2019; results submitted June 2019 and pending review by SPEC.

2-node, 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, Intel® C++ Compiler 2018u2, Intel® MPI Library 2019u2; 1 thread per core; HT=OFF; Turbo=ON, geomean: 52.2 test by Intel on 5/15/2019; results submitted June 2019 and pending review by SPEC.

4-node, 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, Intel® C++ Compiler 2018u2, Intel® MPI Library 2019u2; 1 thread per core; HT=OFF; Turbo=ON, geomean: 78 test by Intel on 5/15/2019; results submitted June 2019 and pending review by SPEC.

8-node, 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, Intel® C++ Compiler 2018u2, Intel® MPI Library 2019u2; 1 thread per core; HT=OFF; Turbo=ON, geomean score: 93.5 test by Intel on 5/21/2019; results submitted June 2019 and pending review by SPEC.

HPC at Scale

Unprecedented performance per rack which is especially compelling for large scale HPC compute clusters. Performance leadership is shown for industry standard benchmarks (i.e., STREAM-Triad, HPCG, SPEC CPU2017* and High Performance LINPACK), Physics (i.e., MILC), Earth Science Modeling (i.e., WRF – weather research and forecast), Life and Material Science (VASP*, LAMMPS, NAMD, RELION), manufacturing (i.e., GROMACS, OpenFOAM*), and Financial Services Industry (i.e., 3 FSI workloads).

Configuration: HPC at Scale

Up to 2.25x avg gain w/ Intel® Xeon® Platinum 9200 processors vs. prior generation: geomean of STREAM OMP - Triad, HPCG, SPECrate2017_fp_base*, High Performance LINPACK, MILC - ks_imp_rhmc w/ 32^4 lattice, WRF - conus-2.5km, VASP* - geomean (3 workloads), LAMMPS - geomean (4 workloads), NAMD - geomean (2 workloads), RELION - Plasmodium Ribosome, GROMACS - geomean (4 workloads), OpenFOAM - 42M_cell_motorbike, FSI - geomean (3 workloads).

 Intel® Xeon® Gold 6148 processor: Intel Reference Platform with 2S Intel® Xeon® Gold 6148 processors (2.4GHz, 20C), 12x16GB DDR4-2666, 1 SSD, Cluster File System: Panasas* (124 TB storage) Firmware v6.3.3.a & Intel® Omni-Path Architecture (Intel® OPA) based IEEL Lustre, BIOS: SE5C620.86B.00.01.0015.110720180833, Microcode: 0x200004d, Oracle* Linux Server release 7.6 (compatible with Red Hat Enterprise Linux* (RHEL) 7.6) on a 7.5 kernel using ksplice for security fixes, Kernel: 3.10.0-862.14.4.el7.crt1.x86_64, OFED stack: OFED OPA 10.8 on RHEL 7.5 with Lustre* v2.10.4, HBA: 100Gbps Intel® OPA 1 port PCIe x16, Switch: Intel® OPA Edge Switch 100 Series 48 Port.

Intel® Xeon® Gold 6148 Processor

 

 

 

 

 

 

 

Application

Workload

Score

Units

Test Date

Compiler

Other S/W

BIOS settings

STREAM OMP 5.10

Triad

192.2471

GB/s, higher is better

2/6/2019

Intel® C++ Compiler 2019u2

 

HT=ON, Turbo=OFF, 1 thread per core

HPCG 2018u3

Binary included MKL

37.7557

GF/s, higher is better

2/6/2019

Intel® C++ Compiler 2019u1

Intel® MPI Library 2019u1, Intel® Math Kernel Library (Intel® MKL) 2019u1

HT=ON, Turbo=OFF, 1 thread per core

SPECrate2017_fp_base

 

216 (Note 1)

higher is better

 

 

 

 

HPL 2.1

Intel® Distribution for MP LINPACK for Clusters

 2284.08

GF/s, higher is better

2/6/2019

Intel® C++ Compiler
2019u1

Intel® Math Kernel Library (Intel® MKL) 2019, Intel® MPI Library 2019u1

HT=ON, Turbo=OFF, 2 threads per core

MILC 7.8.1

ks_imp_rhmc, with lattice volume = 32^4

55.304

GF/s, higher is better

1/30/2019

Intel® C++ Compiler 2019u2

Intel® MPI Library 2019u1

HT=ON, Turbo=OFF, 1 thread per core

WRF 3.9.1.1

conus-2.5km

3.283959

seconds, lower is better

2/11/2019

Intel® C++ Compiler
2018u3

Intel® MPI Library
2018u3

HT=ON, HT=ON, 1 threads per core

GROMACS 2018.2

ion_channel_pme

48.202

ns/day, higher is better

2/5/2019

Intel® C++ Compiler
2019u2

Intel® Math Kernel Library (Intel® MKL) 2019u2, Intel® MPI Library
2019u2

HT=ON, Turbo=OFF, 2 threads per core

 

lignocellulose_rf

2.51

ns/day, higher is better

2/5/2019

Intel® C++ Compiler
2019u2

Intel® Math Kernel Library (Intel® MKL) 2019u2, Intel® MPI Library
2019u2

HT=ON, Turbo=OFF, 2 threads per core

 

water_pme

4.289

ns/day, higher is better

2/5/2019

Intel® C++ Compiler 2019u2

Intel® Math Kernel Library (Intel® MKL) 2019u2, Intel® MPI Library
2019u2

HT=ON, Turbo=OFF, 2 threads per core

 

water_rf

7.978

ns/day, higher is better

2/5/2019

Intel® C++ Compiler
2019u2

Intel® Math Kernel Library (Intel® MKL) 2019u2, Intel® MPI Library
2019u2

HT=ON, Turbo=OFF, 2 threads per core

VASP 5.4.4

CuC

266.678

LOOP+/sec, lower is better

2/5/2019

Intel® C++ Compiler
2018u3

Intel® MPI Library
2018u3

HT=ON, Turbo=OFF, 1 thread per core

 

Pd04

80.21

LOOP+/sec, lower is better

2/5/2019

Intel® C++ Compiler
2018u3

Intel® MPI Library
2018u3

HT=ON, Turbo=OFF, 1 thread per core

 

Si

1360.462

LOOP+/sec, lower is better

2/5/2019

Intel® C++ Compiler
2018u3

Intel® MPI Library
2018u3

HT=ON, Turbo=OFF, 2 threads per core

LAMMPS 12 Dec 2018

Copper

123.17

Timestamp/sec, higher is better

2/5/2019

Intel® C++ Compiler
2019u2

Intel® MPI Library
2019u2

HT=ON, Turbo=ON, 2 threads per core

 

Liquid Crystal

26.267

Timestamp/sec, higher is better

2/5/2019

Intel® C++ Compiler
2019u2

Intel® MPI Library
2019u2

HT=ON, Turbo=ON, 2 threads per core

 

Protein

23.044

Timestamp/sec, higher is better

2/5/2019

Intel® C++ Compiler
2019u2

Intel® MPI Library
2019u2

HT=ON, Turbo=ON, 2 threads per core

 

Water

112.054

Timestamp/sec, higher is better

2/5/2019

Intel® C++ Compiler
2019u2

Intel® MPI Library
2019u2

HT=ON, Turbo=ON, 2 threads per core

NAMD 2.13

apoa1

8.58068

simulations ns/day, higher is better

2/4/2019

Intel® C++ Compiler
2019

Intel® MPI Library
2019

HT=ON, Turbo=OFF, 2 threads per core

 

smtv

0.700624

simulations ns/day, higher is better

2/4/2019

Intel® C++ Compiler
2019

Intel® MPI Library
2019

HT=ON, Turbo=OFF, 2 threads per core

RELION 3.0-beta-2

Plasmodium Ribosome

15685

seconds, lower is better

1/30/2019

Intel® C++ Compiler
2018u3

Intel® Math Kernel Library (Intel® MKL) 2018u3, Intel® MPI Library
2018u3

HT=ON, Turbo=OFF, 2 threads per core

OpenFOAM 6.0

42M_cell_motorbike

2352

time to solution, lower is better

1/30/2019

Intel® C++ Compiler
2019u1

Intel® MPI Library
2018u3

HT=ON, Turbo=OFF, 1 thread per core

FSI

Binomial Options

154.949529

Koptions/sec, higher is better

2/11/2019

Intel® C++ Compiler
2019u1

 

HT=ON, Turbo=ON, 2 threads per core

 

Black Scholes

4496.647324

Koptions/sec, higher is better

2/15/2019

Intel® C++ Compiler
2019u1

 

HT=ON, Turbo=ON, 2 threads per core

 

Monte Carlo

91.524463

Koptions/sec, higher is better

2/15/2019

Intel® C++ Compiler
2019u1

 

High Performance Computing (HPC) - IntelHT=ON, Turbo=ON, 2 threads per core

Configuration: Single/Multi Node Intel® Xeon® Platform Generational HPC Performance

Intel® Xeon® Platinum 9242 processor: Intel Reference Platform with 2S Intel® Xeon® 9242 processors (2.2GHz, 48C), 24x16GB DDR4-2933, 1 SSD, Cluster File System: 2.12.0-1 (server) 2.11.0-14.1 (client), BIOS: PLYXCRB1.86B.0572.D02.1901180818, Microcode: 0x4000017, CentOS 7.6, Kernel: 3.10.0-957.5.1.el7.x86_64, OFED stack: N/A, HBA: 100Gbps Intel® Omni-Path Architecture (Intel® OPA) 1 port PCIe x16, Switch: Intel® OPA Edge Switch 100 Series 48 Port.

Intel® Xeon® Platinum 9242

 

1-node

2-node

4-node

8-node

 

 

 

 

 

Application

Workload

Score

Score

Score

Score

Units

Test Date

Compiler

Other S/W

BIOS settings

STREAM OMP 5.10

Triad

397.7742

 

 

 

GB/s, higher is better

2/26/2019

Intel® C++ Compiler 2019u2

 

HT=ON, Turbo=ON, 1 thread per core

HPCG 2018u3

Binary included MKL

81.9093

161.842

322.874

638.223

GF/s, higher is better

3/5/2019

Intel® C++ Compiler 2019u1

Intel®
MPI 2019u1, Intel®
MKL 2019u1

HT=ON, Turbo=ON, 1 thread per core

SPECrate2017_fp_base

 

485 (Note 1)

 

 

 

higher is better

 

 

 

 

HPL 2.1

Intel® Distribution for MP LINPACK for Clusters

5314

10245

20408

39921

GF/s, higher is better

3/5/2019

Intel® C++ Compiler 2019u1

Intel® Math Kernel Library (Intel® MKL) 2019, Intel®
MPI 2019u1

HT=ON, Turbo=ON, 2 threads per core

MILC 7.8.1

ks_imp_rhmc, with lattice volume = 32^4

118.1088

227.44

456.53

864.53

GF/s, higher is better

3/7/2019

Intel® C++ Compiler 2019u2

Intel® MPI Library
2019u1

HT=ON, Turbo=ON, 1 thread per core

WRF 3.9.1.1

conus-2.5km

1.4403

0.7711

0.4086

0.2266

seconds, lower is better

3/2/2019

Intel® C++ Compiler 2018u3

Intel® MPI Library
2018u3

HT=ON, HT=ON, 1 threads per core

 

conus-12km

0.131243

0.07113

0.03791

0.022752

seconds, lower is better

3/2/2019

Intel® C++ Compiler 2018u3

Intel® MPI Library
2018u3

HT=ON, HT=ON, 1 threads per core

GROMACS 2018.2

ion_channel_pme

108.5

155.1

224.5

321.1

ns/day, higher is better

3/3/2019

Intel® C++ Compiler 2019u2

Intel® Math Kernel Library (Intel® MKL) 2019u2, Intel®
MPI 2019u2

HT=ON, Turbo=ON, 2 threads per core

 

lignocellulose_rf

6.145

11.87

23.247

41.847

ns/day, higher is better

3/3/2019

Intel® C++ Compiler 2019u2

Intel® Math Kernel Library (Intel® MKL) 2019u2, Intel® MPI Library
2019u2

HT=ON, Turbo=ON, 2 threads per core

 

water_pme

12.007

21.755

37.889

63.568

ns/day, higher is better

3/3/2019

Intel® C++ Compiler 2019u2

Intel® Math Kernel Library (Intel® MKL) 2019u2, Intel® MPI Library
2019u2

HT=ON, Turbo=ON, 2 threads per core

 

water_rf

20.762

39.046

71.86

117.375

ns/day, higher is better

3/3/2019

Intel® C++ Compiler 2019u2

Intel® Math Kernel Library (Intel® MKL) 2019u2, Intel® MPI Library
2019u2

HT=ON, Turbo=ON, 2 threads per core

VASP 5.4.4

CuC

133.96

73.55

40.81

24.15

LOOP+/sec, lower is better

3/1/2019

Intel® C++ Compiler 2018u3

Intel® MPI Library
2018u3

HT=ON, Turbo=ON, 1 thread per core

 

Pd04

38.54

23.07

14.02

9.16

LOOP+/sec, lower is better

3/1/2019

Intel® C++ Compiler 2018u3

Intel® MPI Library
2018u3

HT=ON, Turbo=ON, 1 thread per core

 

Si

620.67

331.81

190.06

112.08

LOOP+/sec, lower is better

3/1/2019

Intel® C++ Compiler 2018u3

Intel® MPI Library
2018u3

HT=ON, Turbo=ON, 2 threads per core

LAMMPS 12 Dec 2018

Copper

253.005

474.217

780.521

1162.656

Timestamp/sec, higher is better

3/6/2019

Intel® C++ Compiler 2019u2

Intel® MPI Library
2019u2

HT=ON, Turbo=ON, 2 threads per core

 

Liquid Crystal

61.455

122.73

230.805

404.286

Timestamp/sec, higher is better

3/6/2019

Intel® C++ Compiler 2019u2

Intel® MPI Library
2019u2

HT=ON, Turbo=ON, 2 threads per core

 

Protein

48.307

86.92

146.65

171.822

Timestamp/sec, higher is better

3/6/2019

Intel® C++ Compiler 2019u2

Intel® MPI Library
2019u2

HT=ON, Turbo=ON, 2 threads per core

 

Water

276.104

493.935

858.324

1435.068

Timestamp/sec, higher is better

3/6/2019

Intel® C++ Compiler 2019u2

Intel® MPI Library
2019u2

HT=ON, Turbo=ON, 2 threads per core

NAMD 2.13

apoa1

19.9

34.7

48.5

78.6

simulations ns/day, higher is better

2/28/2019

Intel® C++ Compiler 2019

Intel® MPI Library
2019

HT=ON, Turbo=ON, 2 threads per core

 

smtv

1.71

3.32

6.3

11.68

simulations ns/day, higher is better

2/28/2019

Intel® C++ Compiler 2019

Intel® MPI Library
2019

HT=ON, Turbo=ON, 2 threads per core

RELION 3.0-beta-2

Plasmodium Ribosome

6256

3586

2466

1647

seconds, lower is better

2/28/2019

Intel® C++ Compiler 2018u3

Intel® Math Kernel Library (Intel® MKL) 2018u3, Intel® MPI Library
2018u3

HT=ON, Turbo=ON, 2 threads per core

OpenFOAM 6.0

42M_cell_motorbike

1106

574

277

137

time to solution, lower is better

3/3/2019

Intel® C++ Compiler 2019u1

Intel® MPI Library
2018u3

HT=ON, Turbo=ON, 1 thread per core

FSI

Binomial Options

382.87

 

 

 

Koptions/sec, higher is better

3/6/2019

Intel® C++ Compiler 2019u1

 

HT=ON, Turbo=ON, 2 threads per core

 

Black Scholes

9044.32

 

 

 

Koptions/sec, higher is better

3/6/2019

Intel® C++ Compiler 2019u1

 

HT=ON, Turbo=ON, 2 threads per core

 

Monte Carlo

227.62

 

 

 

Koptions/sec, higher is better

3/6/2019

Intel® C++ Compiler 2019u1

 

HT=ON, Turbo=ON, 2 threads per core

ANSYS® FLUENT®

ANSYS® FLUENT® software contains the broad, physical modeling capabilities needed to model flow, turbulence, heat transfer, and reactions for industrial applications. These range from air flow over an aircraft wing to combustion in a furnace, from bubble columns to oil platforms, from blood flow to semiconductor manufacturing and from clean room design to wastewater treatment plants. Fluent spans an expansive range, including special models, with capabilities to model in-cylinder combustion, aero-acoustics, turbomachinery, and multiphase systems.

Learn more about ANSYS® ›

ANSYS® FLUENT® Multi-Node Configurations

rotor_3m: 1 to 16-node, Each node: 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, ANSYS® FLUENT® 2019 R1 - rotor_3m, Intel® C++ Compiler 2017u3, Intel® MPI Library 2018u3; 1 thread per core; HT-ON, Turbo ON, score (higher is better): 1-node=9988.4, 2-node=17454.5, 4-node=29288.1, 8-node=43746.8, 16-node=48000 test by Intel on 4/30/2019 - 5/3/2019.

fluidized_bed_2m: 1 to 16-node, Each node: 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, ANSYS® FLUENT® 2019 R1 - fluidized_bed_2m, Intel® C++ Compiler 2017u3, Intel® MPI Library 2018u3; 1 thread per core; HT-ON, Turbo ON, score (higher is better): 1-node=6898.2, 2-node=12521.7, 4-node=20945.5, 8-node=32914.3, 16-node=39724.1 test by Intel on 4/30/2019 - 5/6/2019.

sedan_4m: 1 to 16-node, Each node: 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, ANSYS® FLUENT® 2019 R1 - sedan_4m, Intel® C++ Compiler 2017u3, Intel® MPI Library 2018u3; 1 thread per core; HT-ON, Turbo ON, score (higher is better): 1-node=11220.8, 2-node=22012.7, 4-node=41142.9, 8-node=59586.2, 16-node=76800 test by Intel on 4/30/2019 - 5/6/2019.

oil_rig_7m: 1 to 16-node, Each node: 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, ANSYS® FLUENT® 2019 R1 - oil_rig_7m, Intel® C++ Compiler 2017u3, Intel® MPI Library 2018u3; 1 thread per core; HT-ON, Turbo ON, score (higher is better): 1-node=1986.2, 2-node=3891.9, 4-node=6736.8, 8-node=11520, 16-node=15567.6 test by Intel on 5/1/2019 - 5/6/2019.

pump_2m: 1 to 16-node, Each node: 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, ANSYS® FLUENT® 2019 R1 - pump_2m, Intel® C++ Compiler 2017u3, Intel® MPI Library 2018u3; 1 thread per core; HT-ON, Turbo ON, score (higher is better): 1-node=7854.5, 2-node=15026.1, 4-node=28097.6, 8-node=50823.5, 16-node=61714.3 test by Intel on 4/30/2019 - 5/3/2019.

landing_gear_15m: 1 to 16-node, Each node: 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, ANSYS® FLUENT® 2019 R1 - landing_gear_15m, Intel® C++ Compiler 2017u3, Intel® MPI Library 2018u3; 1 thread per core; HT-ON, Turbo ON, score (higher is better): 1-node=598.8, 2-node=1205, 4-node=2400, 8-node=4547.4, 16-node=8554.5 test by Intel on 5/1/2019 - 5/3/2019.

aircraft_wing_14m: 1 to 16-node, Each node: 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, ANSYS® FLUENT® 2019 R1 - aircraft_wing_14m, Intel® C++ Compiler 2017u3, Intel® MPI Library 2018u3; 1 thread per core; HT-ON, Turbo ON, score (higher is better): 1-node=1631.7, 2-node=3254.2, 4-node=6595.4, 8-node=13140.7, 16-node=23834.5 test by Intel on 5/1/2019 - 5/3/2019.

exhaust_system_33m: 1 to 16-node, Each node: 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, ANSYS® FLUENT® 2019 R1 - exhaust_system_33m, Intel® C++ Compiler 2017u3, Intel® MPI Library 2018u3; 1 thread per core; HT-ON, Turbo ON, score (higher is better): 1-node=583.2, 2-node=1152.4, 4-node=2365.5, 8-node=4727.8, 16-node=8771.6 test by Intel on 5/1/2019 - 5/6/2019.

combustor_71m: 1 to 16-node, Each node: 2x Intel® Xeon® Platinum 9242 processor (48 core) on Walker Pass with 384 GB (24x16GB 2933Mhz DDR4) total memory, ucode 0x4000021 on Oracle* Linux Server release 7.6, 3.10.0-957.5.1.el7.crt1.x86_64, ANSYS® FLUENT® 2019 R1 - combustor_71m, Intel® C++ Compiler 2017u3, Intel® MPI Library 2018u3; 1 thread per core; HT-ON, Turbo ON, score (higher is better): 1-node=50.4, 2-node=100.5, 4-node=201.8, 8-node=403.7, 16-node=788.3 test by Intel on 4/30/2019 - 5/3/2019.

Artificial Intelligence

Designed for high performance computing (HPC), advanced artificial intelligence and analytics, and high density infrastructures Intel® Xeon® Platinum 9200 processors deliver breakthrough levels of performance. Using Intel® Deep Learning Boost (Intel® DL Boost) combined with Intel® Optimization for Caffe*, new breakthrough levels of performance can be achieved. Here we show the throughput on an image classification topology – ResNet-50 on the 2nd Generation Intel® Xeon® Scalable processor. Up to 30x improvement in inference performance on Intel® Xeon® Platinum 9282 processor (56 cores) w/ Intel® Deep Learning Boost (Intel® DL Boost) for ResNet-50 (image classification workload) vs. Intel® Xeon® Platinum 8180 processor at launch.

ResNet-50 Performance with Intel® Optimization for Caffe*

Designed for high performance computing (HPC), advanced artificial intelligence and analytics, and high density infrastructures Intel® Xeon® Platinum 9200 processors deliver breakthrough levels of performance. Using Intel® Deep Learning Boost (Intel® DL Boost) combined with Intel® Optimization for Caffe*, new breakthrough levels of performance can be achieved. Here we show the throughput on an image classification topology – ResNet-50 on the 2nd Generation Intel® Xeon® Scalable processor.

Max Inference Throughput

Intel® Xeon® Platinum 8280 processor: Tested by Intel as of 3/04/2019. 2S Intel® Xeon® Platinum 8280 processor (28 cores per socket), HT ON, turbo ON, Total Memory 384 GB (12 slots/ 32 GB/ 2933 MHz), BIOS: SE5C620.86B.0D.01.0348.011820191451, Centos* 7 Kernel 3.10.0-957.5.1.el7.x86_64, Intel® Deep Learning Framework: Intel® Optimization for Caffe* version: https://github.com/intel/caffe Commit id: 362a3b3, ICC 2019.2.187 for build, MKL DNN version: v0.17 (commit hash: 830a10059a018cd2634d94195140cf2d8790a75a), model: https://github.com/intel/caffe/blob/master/models/intel_optimized_models/int8/resnet50_int8_full_conv.prototxt, BS=10, syntheticData:3x224x224, 14 instance/2 socket, Datatype: INT8.

Intel® Xeon® Platinum 9242 processor: Tested by Intel as of 3/04/2019 2S Intel® Xeon® Platinum 9242 processor (48 cores per socket), HT ON, turbo ON, Total Memory 768 GB (24 slots/ 32 GB/ 2933 MHz), BIOS: SE5C620.86B.0D.01.0403.022020190327, Centos* 7 Kernel 3.10.0-957.5.1.el7.x86_64, Intel® Deep Learning Framework: Intel® Optimization for Caffe* version: https://github.com/intel/caffe Commit id: 362a3b3, ICC 2019.2.187 for build, MKL DNN version: v0.17 (commit hash: 830a10059a018cd2634d94195140cf2d8790a75a), model: https://github.com/intel/caffe/blob/master/models/intel_optimized_models/int8/resnet50_int8_full_conv.prototxt, BS=8, synthetic Data:3x224x224, 24 instance/2 socket, Datatype: INT8.

Intel® Xeon® Platinum 9282 processorTested by Intel as of 3/04/2019. DL Inference: Platform: Dragon rock 2S Intel® Xeon® Platinum 9282 processor (56 cores per socket), HT ON, turbo ON, Total Memory 768 GB (24 slots/ 32 GB/ 2933 MHz), BIOS: SE5C620.86B.0D.01.0241.112020180249, Centos* 7 Kernel 3.10.0-957.5.1.el7.x86_64, Intel® Deep Learning Framework: Intel® Optimization for Caffe*
version: https://github.com/intel/caffe Commit id: 362a3b3, ICC 2019.2.187 for build, MKL DNN version: v0.17 (commit hash: 830a10059a018cd2634d94195140cf2d8790a75a), model: https://github.com/intel/caffe/blob/master/models/intel_optimized_models/int8/resnet50_int8_full_conv.prototxt, BS=8, synthetic Data:3x224x224, 28 instance/2 socket, Datatype: INT8.

BKMs for running multi-stream configurations on Xeon: https://www.intel.ai/wp-content/uploads/sites/69/TensorFlow_Best_Practices_Intel_Xeon_AI-HPC_v1.1_Q119.pdf.

Notices

Performance results are based on testing as of the dates shown in configurations and may not reflect all publicly available security updates. See configuration disclosure for details. No product or component can be absolutely secure.

Software and workloads used in performance tests may have been optimized for performance only on Intel® microprocessors. Performance tests, such as SYSmark* and MobileMark*, are measured using specific computer systems, components, software, operations, and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products. For more information go to www.intel.com.tr/benchmarks.

Intel's compilers may or may not optimize to the same degree for non-Intel microprocessors for optimizations that are not unique to Intel® microprocessors. These optimizations include SSE2 and SSE3 instruction sets and other optimizations. Intel does not guarantee the availability, functionality, or effectiveness of any optimization on microprocessors not manufactured by Intel.

Microprocessor-dependent optimizations in this product are intended for use with Intel® microprocessors. Certain optimizations not specific to Intel® microarchitecture are reserved for Intel® microprocessors. Please refer to the applicable product user and reference guides for more information regarding the specific instruction sets covered by this notice. Notice revision #20110804.

This offering is not approved or endorsed by OpenCFD Limited, producer and distributor of the OpenFOAM software via www.openfoam.com, and owner of the OPENFOAM® and OpenCFD® trademark.

SPEC* and the benchmark names SPECint*, SPECfp*, SPECjbb*, SPECjEnterprise*, SPECvirt_sc*, SPECpower_ssj*, SPECompG*, and SPECmpi* are registered trademarks of the Standard Performance Evaluation Corporation.

ANSYS® and FLUENT® are trademarks of ANSYS Inc., used by permission.