Also don't forget that the fixed cost of the rest of the machine is added
to the per-gpu price. With slower GPUs you're adding the same fixed price
(maybe 1/2 or 1/4 of the node cost) to a slower gpu, making the cost per ns
even less favorable with a slower gpu.
On Thu, Apr 18, 2019, 5:28 PM David Cerutti <dscerutti.gmail.com> wrote:
> I should work just fine; run the test suit after installing, then
> benchmarks. Personally my recommendation would be to get RTX-2080Tis :
> within the GTX and RTX hardware lines, you tend to get what you pay for:
> more FLOPS for more $, roughly in proportion to the cost. If you got
> Tesla, you pay way more per ns of MD, but you get support for your cluster
> and other things not so immediately obvious with the gaming cards. But, as
> I said, with the gaming cards if you're getting more ns of MD for spending
> proportionately more money, wouldn't you want the fastest card before the
> payoff starts to diminish? Your human time is valuable as well, and you
> don't want to wait on results any longer than you need to.
>
> Don't scrimp on your hardware unless it's a fundamental shift in your
> apparatus (i.e. switching to T4s on a completely different server
> configuration and power budget).
>
> Dave
>
>
> On Thu, Apr 18, 2019 at 4:30 PM Yeyue Xiong <xiongyy.vt.edu> wrote:
>
> > Hello!
> >
> > I have a quick question about GPU RTX-2070. Our group is currently
> looking
> > to upgrade some of our workstations and the RTX-2070 seems a good choice,
> > but on this link <http://ambermd.org/GPUHardware.php> RTX-2070 is listed
> > as
> > supported (untested). We are a bit confused about whether it will work
> with
> > AMBER.
> >
> > Has anyone had any experience with running AMBER on RTX-2070? Are there
> any
> > issues using that card?
> >
> > Best regards,
> > Yeyue Xiong
> >
> > --
> > Yeyue Xiong
> > PhD student,
> > Structural Bioinformatics and Computational Molecular Biophysics Lab,
> > Department of Computer Science;
> > Engineering Mechanics,
> > Department of Biomedical Engineering and Mechanics,
> > Virginia Tech
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Apr 19 2019 - 05:30:02 PDT
