[Pw_forum] QE-GPU performance
Rolly Ng
rollyng at gmail.com
Mon Dec 11 16:46:31 CET 2017
Dear Phanikumar,
You are welcome and please try to reply to your question so it is easy for tracking.
When you have Intel PSXE installed, please make sure the environmental variables are loaded when you log into the system. After this is done, you may check that mpirun is pointed to the correct path by:
[somebody at somenode ~]$ mpirun -V
Intel(R) MPI Library for Linux* OS, Version 2017 Update 1 Build 20161016 (id: 16418)
Copyright (C) 2003-2016, Intel Corporation. All rights reserved.
Please refer to Ubuntu guide on intel PSXE installation on known to make the these persistent. I believe this can be done via editing your /etc/profile.local and using the source comment, something like these,
source <install_dir>/parallel_studio_xe_2017.<update number>.<package number>/bin/psxevars.sh intel64
logout and login again to see the effect.
Regards,
Rolly
PhD, Research Fellow,
Department of Materials Science and Engineering,
City University of Hong Kong
Tel: +852 3442 4000
Fax: +852 3442 0892
From: pw_forum-bounces at pwscf.org [mailto:pw_forum-bounces at pwscf.org] On Behalf Of Phanikumar Pentyala
Sent: Monday, December 11, 2017 10:54 AM
To: PWSCF Forum
Subject: Re: [Pw_forum] Pw_forum Digest, Vol 125, Issue 8
Thank you Rolly for your comments
Previously I used both intel MKL and MPI. MPI (intel) was not running at all so that I switched to Openmpi. current version of my intel MKL library was "l_mkl_2018.1.163"
My linux-OS was Ubuntu-16.04 serever, Is OS also create some problem??
Can you explain Is there any difference between Parallel Studio XE inetel and above intel MKL (above version)??
(sorry , since it was so long time using pw-forum so I forgot that, This is my affiliation)
Phanikumar
Research scholar
Department of Chemical engineering
Indian Institute of Technology Kharagpur
West Bengal
India
Message: 4
Date: Sun, 10 Dec 2017 09:01:59 +0530
From: Phanikumar Pentyala <phani12.chem at gmail.com>
Subject: [Pw_forum] QE-GPU performance
To: PWSCF Forum <pw_forum at pwscf.org>
Message-ID:
<CAOgLYHHDQWV7JeYe17KBTwGwv4NVyNTJ-6XpqKfkVjXYbj8ELQ at mail.gmail.com>
Content-Type: text/plain; charset="utf-8"
Dear users and developers
Currently I am using two Tesla K40m cards for my computational work on
quantum espresso (QE). My GPU enabled QE code running very slower than
normal version. My question was weather particular application will be fast
only in some versions of CUDA toolkit? (as mentioned in previous post:
http://qe-forge.org/pipermail/pw_forum/2015-May/106889.html) OR is there
any other reason hindering performance (memory) of GPU? (when I am hitting
top command in my server, option of 'VIRT' showing different values (top
command pasted in attached file))
Some error was generating while submitting code that "A high-performance
Open MPI point-to-point messaging module was unable to find any relevant
network interfaces: Module: OpenFabrics (openib) Host: XXXX Another
transport will be used instead, although this may result in lower
performance". Is this MPI thread hindering GPU performance ?
(P.S: We don't have any Infiband adapter HCA in server)
Current details of server are (full details attached):
Server: FUJITSU PRIMERGY RX2540 M2
CUDA version: 9.0
NVIDIA driver: 384.9
openmpi version: 2.0.4 with intel mkl libraries
QE-gpu version : 5.4.0
Thanks in advance
Regards
Phanikumar
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://pwscf.org/pipermail/pw_forum/attachments/20171210/91bedf7a/attachment-0001.html
-------------- next part --------------
##################################################################################################################################################
SERVER architecture information (from "lscpu" command in terminal)
##################################################################################################################################################
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 40
On-line CPU(s) list: 0-39
Thread(s) per core: 2
Core(s) per socket: 10
Socket(s): 2
NUMA node(s): 2
Vendor ID: GenuineIntel
CPU family: 6
Model: 79
Model name: Intel(R) Xeon(R) CPU E5-2640 v4 @ 2.40GHz
Stepping: 1
CPU MHz: 1200.000
BogoMIPS: 4788.53
Virtualization: VT-x
L1d cache: 32K
L1i cache: 32K
L2 cache: 256K
L3 cache: 25600K
NUMA node0 CPU(s): 0-9,20-29
NUMA node1 CPU(s): 10-19,30-39
##################################################################################################################################################
After I run device quiry in CUDA_samples I got this information about my GPU accelerators
##################################################################################################################################################
CUDA Device Query (Runtime API) version (CUDART static linking)
Detected 2 CUDA Capable device(s)
Device 0: "Tesla K40m"
CUDA Driver Version / Runtime Version 9.0 / 9.0
CUDA Capability Major/Minor version number: 3.5
Total amount of global memory: 11440 MBytes (11995578368 bytes)
(15) Multiprocessors, (192) CUDA Cores/MP: 2880 CUDA Cores
GPU Max Clock rate: 745 MHz (0.75 GHz)
Memory Clock rate: 3004 Mhz
Memory Bus Width: 384-bit
L2 Cache Size: 1572864 bytes
Maximum Texture Dimension Size (x,y,z) 1D=(65536), 2D=(65536, 65536), 3D=(4096, 4096, 4096)
Maximum Layered 1D Texture Size, (num) layers 1D=(16384), 2048 layers
Maximum Layered 2D Texture Size, (num) layers 2D=(16384, 16384), 2048 layers
Total amount of constant memory: 65536 bytes
Total amount of shared memory per block: 49152 bytes
Total number of registers available per block: 65536
Warp size: 32
Maximum number of threads per multiprocessor: 2048
Maximum number of threads per block: 1024
Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
Max dimension size of a grid size (x,y,z): (2147483647, 65535, 65535)
Maximum memory pitch: 2147483647 bytes
Texture alignment: 512 bytes
Concurrent copy and kernel execution: Yes with 2 copy engine(s)
Run time limit on kernels: No
Integrated GPU sharing Host Memory: No
Support host page-locked memory mapping: Yes
Alignment requirement for Surfaces: Yes
Device has ECC support: Enabled
Device supports Unified Addressing (UVA): Yes
Supports Cooperative Kernel Launch: No
Supports MultiDevice Co-op Kernel Launch: No
Device PCI Domain ID / Bus ID / location ID: 0 / 2 / 0
Compute Mode:
< Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >
Device 1: "Tesla K40m"
CUDA Driver Version / Runtime Version 9.0 / 9.0
CUDA Capability Major/Minor version number: 3.5
Total amount of global memory: 11440 MBytes (11995578368 bytes)
(15) Multiprocessors, (192) CUDA Cores/MP: 2880 CUDA Cores
GPU Max Clock rate: 745 MHz (0.75 GHz)
Memory Clock rate: 3004 Mhz
Memory Bus Width: 384-bit
L2 Cache Size: 1572864 bytes
Maximum Texture Dimension Size (x,y,z) 1D=(65536), 2D=(65536, 65536), 3D=(4096, 4096, 4096)
Maximum Layered 1D Texture Size, (num) layers 1D=(16384), 2048 layers
Maximum Layered 2D Texture Size, (num) layers 2D=(16384, 16384), 2048 layers
Total amount of constant memory: 65536 bytes
Total amount of shared memory per block: 49152 bytes
Total number of registers available per block: 65536
Warp size: 32
Maximum number of threads per multiprocessor: 2048
Maximum number of threads per block: 1024
Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
Max dimension size of a grid size (x,y,z): (2147483647, 65535, 65535)
Maximum memory pitch: 2147483647 bytes
Texture alignment: 512 bytes
Concurrent copy and kernel execution: Yes with 2 copy engine(s)
Run time limit on kernels: No
Integrated GPU sharing Host Memory: No
Support host page-locked memory mapping: Yes
Alignment requirement for Surfaces: Yes
Device has ECC support: Enabled
Device supports Unified Addressing (UVA): Yes
Supports Cooperative Kernel Launch: No
Supports MultiDevice Co-op Kernel Launch: No
Device PCI Domain ID / Bus ID / location ID: 0 / 129 / 0
Compute Mode:
< Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >
> Peer access from Tesla K40m (GPU0) -> Tesla K40m (GPU1) : No
> Peer access from Tesla K40m (GPU1) -> Tesla K40m (GPU0) : No
deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 9.0, CUDA Runtime Version = 9.0, NumDevs = 2
Result = PASS
##################################################################################################################################################
GPU performance after 'nvidia-smi' command in terminal
##################################################################################################################################################
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 384.90 Driver Version: 384.90 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 Tesla K40m Off | 00000000:02:00.0 Off | 0 |
| N/A 42C P0 75W / 235W | 11381MiB / 11439MiB | 83% Default |
+-------------------------------+----------------------+----------------------+
| 1 Tesla K40m Off | 00000000:81:00.0 Off | 0 |
| N/A 46C P0 75W / 235W | 11380MiB / 11439MiB | 87% Default |
+-------------------------------+----------------------+----------------------+
##################################################################################################################################################
TOP command if my server
##################################################################################################################################################
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
20019 xxxxx 20 0 0.158t 426080 152952 R 100.3 0.3 36:29.44 pw-gpu.x
20023 xxxxx 20 0 0.158t 422380 153328 R 100.0 0.3 36:29.42 pw-gpu.x
20025 xxxxx 20 0 0.158t 418256 153376 R 100.0 0.3 36:27.74 pw-gpu.x
20042 xxxxx 20 0 0.158t 416912 153104 R 100.0 0.3 36:24.63 pw-gpu.x
20050 xxxxx 20 0 0.158t 412564 153084 R 100.0 0.3 36:25.68 pw-gpu.x
20064 xxxxx 20 0 0.158t 408012 153100 R 100.0 0.3 36:25.54 pw-gpu.x
20098 xxxxx 20 0 0.158t 398404 153436 R 100.0 0.3 36:27.92 pw-gpu.x
------------------------------
Message: 5
Date: Sun, 10 Dec 2017 17:07:59 +0800
From: Rolly Ng <rollyng at gmail.com>
Subject: Re: [Pw_forum] QE-GPU performance
To: pw_forum at pwscf.org
Message-ID: <225411b4-1c48-6f24-954f-5d0af115e76f at gmail.com>
Content-Type: text/plain; charset="utf-8"
Dear Phanikumar,
Please include your affiliation when posting to the forum.
In my experience with QE-GPU v5.3.0 and v5.4.0, the working combination
of software is,
1) Intel PSXE 2017
2) CUDA 6.5 or 7.0
3) Centos 7.1
Please try the above combination.
Regards,
Rolly
PhD. Research Fellow,
Dept. of Physics & Materials Science,
City University of Hong Kong
Tel: +852 3442 4000
Fax: +852 3442 0538
On 12/10/2017 11:31 AM, Phanikumar Pentyala wrote:
> Dear users and developers
>
> Currently I am using two Tesla K40m cards for my computational work on
> quantum espresso (QE). My GPU enabled QE code running very slower than
> normal version. My question was weather particular application will be
> fast only in some versions of CUDA toolkit? (as mentioned in previous
> post: http://qe-forge.org/pipermail/pw_forum/2015-May/106889.html) OR
> is there any other reason hindering performance (memory) of GPU? (when
> I am hitting top command in my server, option of 'VIRT' showing
> different values (top command pasted in attached file))
>
> Some error was generating while submitting code that "A
> high-performance Open MPI point-to-point messaging module was unable
> to find any relevant network interfaces: Module: OpenFabrics (openib)?
> Host: XXXX Another transport will be used instead, although this may
> result in lower performance". Is this MPI thread hindering GPU
> performance ?
>
> (P.S: We don't have any Infiband adapter HCA in server)
>
>
> Current details of server are (full details attached):
>
> Server: FUJITSU PRIMERGY RX2540 M2
> CUDA version: 9.0
> NVIDIA driver: 384.9
> openmpi version: 2.0.4 with intel mkl libraries
> QE-gpu version : 5.4.0
>
>
> Thanks in advance
>
> Regards
> Phanikumar
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://pwscf.org/pipermail/pw_forum/attachments/20171210/35e7e383/attachment-0001.html
------------------------------
_______________________________________________
Pw_forum mailing list
Pw_forum at pwscf.org
http://pwscf.org/mailman/listinfo/pw_forum
End of Pw_forum Digest, Vol 125, Issue 8
****************************************
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20171211/048f8e3c/attachment.html>
More information about the users
mailing list