SPEC® ACCEL™ ACC Result

IBM Corporation (Test Sponsor: NVIDIA Corporation)

Tesla V100

IBM Power Systems AC922 for High Performance
Computing (8335-GTH)

ACCEL license:	019	Test date:	Aug-2018
Test sponsor:	NVIDIA Corporation	Hardware Availability:	May-2018
Tested by:	NVIDIA Corporation	Software Availability:	Aug-2018

Hardware
CPU Name:	POWER9, altivec supported
CPU Characteristics:
CPU MHz:	3400
CPU MHz Maximum:	3800
FPU:	Integrated
CPU(s) enabled:	40 cores, 2 chips, 20 cores/chip, 4 threads/core
CPU(s) orderable:	1,2 chips
Primary Cache:	64 KB I + 64 KB D on chip per core
Secondary Cache:	512 KB I+D on chip per core
L3 Cache:	100 MB I+D on chip per chip shared NUCA / 20 cores
Other Cache:	None
Memory:	128 GB (16 x 8 GB 1Rx4 PC4-2666V-R)
Disk Subsystem:	1 x 1TB 7200 RPM SATA HDD
Other Hardware:	None

Accelerator
Accel Model Name:	Tesla V100-SXM2-16GB
Accel Vendor:	NVIDIA Corporation
Accel Name:	Tesla V100
Type of Accel:	GPU
Accel Connection:	NVLink
Does Accel Use ECC:	Yes
Accel Description:	See notes
Accel Driver:	NVIDIA UNIX ppc64le Kernel Module 396.26

Software
Operating System:	Red Hat Enterprise Linux Server release 7.5 (Maipo) 4.14.0-49.8.1.el7a.ibmnvidia.6.1.ppc64le
Compiler:	PGI Professional Edition, Release 18.7
File System:	xfs
System State:	Run level 3 (multi-user)
Other Software:	None

Results Table

Benchmark	Base						Peak
Benchmark	Seconds	Ratio	Seconds	Ratio	Seconds	Ratio	Seconds	Ratio	Seconds	Ratio	Seconds	Ratio
Results appear in the order in which they were run. Bold underlined text indicates a median measurement.
303.ostencil	12.8	11.3	12.8	11.4	13.7	10.6	12.8	11.3	12.8	11.4	13.7	10.6
304.olbm	35.6	12.8	35.6	12.8	35.6	12.8	35.6	12.8	35.6	12.8	35.6	12.8
314.omriq	42.4	22.6	42.4	22.6	42.2	22.6	42.4	22.6	42.4	22.6	42.2	22.6
350.md	11.2	22.4	11.4	22.2	11.4	22.2	11.2	22.4	11.4	22.2	11.4	22.2
351.palm	110	3.38	110	3.37	110	3.37	110	3.38	110	3.37	110	3.37
352.ep	48.2	11.0	48.1	11.0	48.3	11.0	48.2	11.0	48.1	11.0	48.3	11.0
353.clvrleaf	37.0	12.0	37.0	12.0	36.9	12.0	37.0	12.0	37.0	12.0	36.9	12.0
354.cg	30.2	13.5	30.2	13.5	30.3	13.4	30.2	13.5	30.2	13.5	30.3	13.4
355.seismic	28.3	13.1	28.3	13.1	28.3	13.1	28.3	13.1	28.3	13.1	28.3	13.1
356.sp	22.2	12.4	22.2	12.4	22.4	12.3	22.2	12.4	22.2	12.4	22.4	12.3
357.csp	20.3	13.3	20.3	13.3	20.3	13.3	20.3	13.3	20.3	13.3	20.3	13.3
359.miniGhost	46.5	7.93	46.4	7.96	46.1	8.01	46.5	7.93	46.4	7.96	46.1	8.01
360.ilbdc	31.6	11.6	31.1	11.8	31.2	11.8	31.6	11.6	31.1	11.8	31.2	11.8
363.swim	33.0	6.96	33.0	6.97	33.4	6.89	33.0	6.96	33.0	6.97	33.4	6.89
370.bt	9.98	22.4	9.71	23.0	10.1	22.2	9.98	22.4	9.71	23.0	10.1	22.2

Platform Notes

 Sysinfo program /local/home/colgrove/SPECACCEL/Docs/sysinfo
 $Rev: 6965 $ $Date:: 2015-04-21 #$ c05a7f14b1b1765e3fe1df68447e8a35
 running on wsn1 Wed Aug  1 09:48:43 2018

 This section contains SUT (System Under Test) info as seen by
 some common utilities.  To remove or add to this section, see:
   http://www.spec.org/accel/Docs/config.html#sysinfo

 From /proc/cpuinfo
    clock : 3000.000000MHz
    clock : 3616.000000MHz
    machine : PowerNV 8335-GTC........
    model : 8335-GTC........
    platform : PowerNV
    revision : 2.2 (pvr 004e 1202)
    cpu : POWER9, altivec supported
 *
 * 0 "physical id" tags found.  Perhaps this is an older system,
 * or a virtualized system.  Not attempting to guess how to
 * count chips/cores for this system.
 *
       160 "processors"
    cores, siblings (Caution: counting these is hw and system dependent.  The
    following excerpts from /proc/cpuinfo might not be reliable.  Use with
    caution.)

 From /proc/meminfo
    MemTotal:       199796800 kB
    HugePages_Total:       0
    Hugepagesize:       2048 kB

 /usr/bin/lsb_release -d
    Red Hat Enterprise Linux Server release 7.5 (Maipo)

 From /etc/*release* /etc/*version*
    os-release:
       NAME="Red Hat Enterprise Linux Server"
       VERSION="7.5 (Maipo)"
       ID="rhel"
       ID_LIKE="fedora"
       VARIANT="Server"
       VARIANT_ID="server"
       VERSION_ID="7.5"
       PRETTY_NAME="Red Hat Enterprise Linux Server 7.5 (Maipo)"
    redhat-release: Red Hat Enterprise Linux Server release 7.5 (Maipo)
    system-release: Red Hat Enterprise Linux Server release 7.5 (Maipo)
    system-release-cpe: cpe:/o:redhat:enterprise_linux:7.5:ga:server

 uname -a:
    Linux wsn1 4.14.0-49.8.1.el7a.ibmnvidia.6.1.ppc64le #1 SMP Tue Jun 5 13:56:12
    -03 2018 ppc64le ppc64le ppc64le GNU/Linux

 run-level 3 Jun 29 16:37

 SPEC is set to: /local/home/colgrove/SPECACCEL
    Filesystem                 Type  Size  Used Avail Use% Mounted on
    /dev/mapper/rhel_wsn1-root xfs   927G   95G  832G  11% /

 (End of data from sysinfo program)
 Information from pgaccelinfo
 CUDA Driver Version:           9020
 NVRM version:                  NVIDIA UNIX ppc64le Kernel Module  396.26
 Device Number:                 0
 Device Name:                   Tesla V100-SXM2-16GB
 Device Revision Number:        7.0
 Global Memory Size:            16911433728
 Number of Multiprocessors:     80
 Concurrent Copy and Execution: Yes
 Total Constant Memory:         65536
 Total Shared Memory per Block: 49152
 Registers per Block:           65536
 Warp Size:                     32
 Maximum Threads per Block:     1024
 Maximum Block Dimensions:      1024, 1024, 64
 Maximum Grid Dimensions:       2147483647 x 65535 x 65535
 Maximum Memory Pitch:          2147483647B
 Texture Alignment:             512B
 Clock Rate:                    1530 MHz
 Execution Timeout:             No
 Integrated Device:             No
 Can Map Host Memory:           Yes
 Compute Mode:                  default
 Concurrent Kernels:            Yes
 ECC Enabled:                   Yes
 Memory Clock Rate:             877 MHz
 Memory Bus Width:              4096 bits
 L2 Cache Size:                 6291456 bytes
 Max Threads Per SMP:           2048
 Async Engines:                 4
 Unified Addressing:            Yes
 Managed Memory:                Yes
 Concurrent Managed Memory:     Yes
 Preemption Supported:          Yes
 Cooperative Launch:            Yes
   Multi-Device:                Yes
 PGI Default Target:            -ta=tesla:cc70

General Notes

Yes: The test sponsor attests, as of date of publication, that CVE-2017-5754 (Meltdown)
is mitigated in the system as tested and documented.
Yes: The test sponsor attests, as of date of publication, that CVE-2017-5753 (Spectre variant 1)
is mitigated in the system as tested and documented.
Yes: The test sponsor attests, as of date of publication, that CVE-2017-5715 (Spectre variant 2)
is mitigated in the system as tested and documented.

Base Compiler Invocation

C benchmarks:

pgcc

Fortran benchmarks:

pgfortran

Benchmarks using both Fortran and C:

pgcc pgfortran

Base Optimization Flags

C benchmarks:

-fast -Mfprelaxed -acc -ta=tesla:cc70

Fortran benchmarks:

-fast -Mfprelaxed -acc -ta=tesla:cc70

Benchmarks using both Fortran and C:

353.clvrleaf:	-fast -Mfprelaxed -acc -ta=tesla:cc70
359.miniGhost:	-fast -Mfprelaxed -acc -ta=tesla:cc70 -Mnomain

Peak Optimization Flags

C benchmarks:

303.ostencil:	basepeak = yes
304.olbm:	basepeak = yes
314.omriq:	basepeak = yes
352.ep:	basepeak = yes
354.cg:	basepeak = yes
357.csp:	basepeak = yes
370.bt:	basepeak = yes

Fortran benchmarks:

350.md:	basepeak = yes
351.palm:	basepeak = yes
355.seismic:	basepeak = yes
356.sp:	basepeak = yes
360.ilbdc:	basepeak = yes
363.swim:	basepeak = yes

Benchmarks using both Fortran and C:

353.clvrleaf:	basepeak = yes
359.miniGhost:	basepeak = yes

The flags files that were used to format this result can be browsed at
https://www.spec.org/accel/flags/PGI-Platform-Multicore-OMP.html,
https://www.spec.org/accel/flags/pgi2018_flags.html.

You can also download the XML flags sources by saving the following links:
https://www.spec.org/accel/flags/PGI-Platform-Multicore-OMP.xml,
https://www.spec.org/accel/flags/pgi2018_flags.xml.