[Wien] Hardware for Wien2k ver. 19

Sat Jul 20 23:54:02 CEST 2019

Thank you very much Gavin - I'll use your expert input in trying to refine
the configuration.

Meanwhile -
Yes, SSD is the choice of the 1st HD; RAM is TBD;
CUDA - No;
NVIDIA is recommended by Wavefunc, but not necessarily high end; in Wien2k
I use the auxiliary software XCrysden and am guessing that
             Spartan18 requirements are more stringent.

Other - My current goal is to calculate Gibbs free energy G(P,T) for
chemical reactions, so the system could become large, possibly =< 100.
 atoms.  So far succeeded via EOS with G(P, T=0) and learning to extend for
G(P,T.ne.0).

On Sat, Jul 20, 2019 at 4:41 PM Gavin Abo <gsabo at crimson.ua.edu> wrote:

> What memory and hard drive(s) did you select?
>
> - Getting more RAM should increase the capability of doing WIEN2k
> calculations (of structures with more atoms) [1], but you have to decide if
> the additional cost is attainable or worthwhile to you or not.
>
> - SSD (solid state drive) as 1st hard drive I believe is preferred these
> days for the installation drive of Linux, WIEN2k, and for WIEN2k
> calculations (also is the recommendation given for Spartan'18 on its
> website [2]).  The flash memory of SSD is known to have faster data read
> than HDD (hard disk drive).  SSD is known to have slower data write than
> its own data read, but the data write might be faster or comparable to that
> of the HDD.  Look for drive having more buffer cache as this may be
> reducing that bottleneck of slower data write [3].  The flash of SSD breaks
> down over time [4], such that after many years you might notice capacity of
> the drive shrink as the memory storage cells fail.  HDD does not have that
> problem since it uses a magnetic disk instead of flash cell.  However, the
> mechanical read/write head of HDD can sometimes break causing it to fail
> sooner than a SSD.  SSD tends to have lower capacity (higher cost per bit)
> than a HDD.  Therefore, a low RPM HDD (e.g., 5,400 RPM) I would recommend
> as 2nd hard drive for data storage of completed WIEN2k calculation files.
> If you decided to go with a HDD for the 1st hard drive, I suggest trying to
> get a higher performance high RPM HDD (e.g., 10,000 RPM) for that.
>
> Does your 'other uses' require the "Nvidia Quadro RTX4000" or do you need
> it for its VGA support [5] for hooking up to your monitor?
>
> The Dell website [6] lists the following for me:
>
> AMD Single
> Radeon Pro WX 7100, 8GB, 4DP (7X20T) Included in price
> Radeon Pro WX 9100, 16GB, 6 mDP to DP adapter XX20T + $1,174.88
>
> nVIDIA Single
> Nvidia Quadro RTX4000, 8GB, 3DP, VirtualLink (XX20T) + $545.85
> Nvidia Quadro RTX5000, 16GB, 4DP, VirtualLink (XX20T) + $1,529.22
> Nvidia Quadro RTX6000, 24GB, 4DP, VirtualLink (XX20T) + $4,347.94
> NVIDIA® Quadro® P2000, 5GB, 4 DP (7X20T) + $55.22
> NVIDIA® Quadro® P4000, 8GB, 4 DP (7X20T) + $336.18
>
> If you have or are getting a monitor with DisplayPort 1.4, the lower end
> "Radeon Pro WX 7100" [7] would likely work fine for WIEN2k and would save
> you $545.85.
>
> I haven't noticed WIEN2k requiring much for graphics.  Usually, the modern
> lower end integrated graphics ports on personal computers have worked
> fine.  I haven't had to use higher end dedicated graphic cards like the
> Nvidia Quadro or Radeon Pro.
>
> It's the visualization programs that you use for WIEN2k that you will
> likely want to look into.  For example, if you use VESTA, its manual [8]
> has:
> Video RAM: 16 MB or more is desirable.
> Video card: A graphics card capable of hardware acceleration of the OpenGL
> instruction set is recommended.
> Display: A minimum resolution of 1024 x 768 pixels with ca. 65,000 orca.
> 16.7 million colors.
>
> Or are you getting the Nvidia Quadro RTX4000 because you plan to modify
> the WIEN2k code yourself to make it do calculations with CUDA [9]?  If so,
> the datasheet [5] shows 2304 CUDA cores. That looks significant compared to
> the NVIDIA Quadro P2000 having 1024 CUDA cores [10].
>
>
> [1] http://susi.theochem.tuwien.ac.at/hard+soft/index.html
> [2] https://www.wavefun.com/spartan
> [3]
> https://superuser.com/questions/309613/do-solid-state-disks-ssds-have-a-buffer-cache
> [4] https://www.ontrack.com/blog/2018/02/07/how-long-do-ssds-really-last/
> [5]
> https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/quadro-product-literature/quadro-rtx-4000-data-sheet-us-nvidia-830682-r6-web.pdf
> [6]
> https://www.dell.com/en-us/work/shop/workstations-isv-certified-dell/precision-7920-tower/spd/precision-7920-workstation/xctopt7920us_3?configurationid=734ac6d8-8978-4b7d-8392-5ec7d1a1d1bd
> [7]
> https://www.amd.com/en/products/professional-graphics/radeon-pro-wx-7100
> [8] http://jp-minerals.org/vesta/archives/VESTA_Manual.pdf
> [9] https://en.wikipedia.org/wiki/CUDA
> [10]
> https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/productspage/quadro/quadro-desktop/quadro-pascal-p2000-data-sheet-us-nvidia-704443-r2-web.pdf
>
>
> On 7/20/2019 10:55 AM, kklier wrote:
>
>
>    - Yes, I have SEARCHED the archives AND READ THE USERS GUIDE and the
>    FAQ pages, but I couldn't solve my problem that way.
>
> I am seeking a recommendation for suitability of a new multi-core machine
> for medium-level calculations with Wien2k ver. 19.  I have the following
> quote from Dell, herein abbreviated:
>
>
>
> *Precision 7920 Tower XCTO Base 210-AMRM*
>
> *Intel Xeon Gold 6130 2.1GHz, 3.7GHz Turbo, 16C, 10.4GT/s 3UPI, 22MB
> Cache, HT (125W) DDR4-2666 338-BMDV*
>
> *CPU clip, assemble CPU with heatsink 575-BBPB*
>
> *Red Hat Enterprise Linux WS v7.5 with 1 YR RHN 634-BRFB  *
>
> *Nvidia Quadro RTX4000, 8GB, 3DP, VirtualLink (XX20T) 490-BFHQ*
>
>
>
> *Question 1*: Is the proposed choice of processor, 16 cores, and the
> NVIDIA graphics card adequate for Wien2k ver. 19 calculations and
> graphics?
>
> *Question 2*: Any suggestion for a more suitable configuration would be
> highly appreciated.
>
> *Info:* Immediate application will be pressure AND temperature dependent
> EOS to follow up on Thermodynamically Controlled High-Pressure
> High-Temperature Synthesis of Crystalline Fluorinated sp3-Carbon Networks
> by Klier and Landskron, J. Phys. Chem. C, 119, 26086, 2015
>
> *Comment 1*: The proposed machine is intended for several other uses,
> including of Linux Spartan18 Parallel Suite for molecular calculations,
> hence the option 16C.
>
> *Comment 2*: No networking is intended – just stand-alone workstation.
>
>
>
>
>
>
>
>
>
> Sent from Mail <https://go.microsoft.com/fwlink/?LinkId=550986> for
> Windows 10
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at:
> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20190720/18f51edb/attachment.html>