Hi Filip,
I will have benchmarks for the AMBER website soon. Expect about 20% or so
performance boost over a K20X - so something around 115 to 120ns/day or so
for DHFR NVE on a single GTX 780 <cough>Titan</cough> card.
The double precision performance makes little difference these days for
AMBER since we switched to our fixed precision model:
Grand, S.L.; Goetz, A.W.; Walker, R.C.; "SPFP: Speed without compromise -
a mixed precision model for GPU accelerated molecular dynamics
simulations", Chem. Phys. Comm., 2013, 184, pp374-380, DOI:
10.1016/j.cpc.2012.09.022
http://www.wmd-lab.org/papers/2012_10_Comp_Phys_Comm_SPFP_Paper.pdf
So the thing to look at mostly is the single precision clock speed, core
count and memory bandwidth.
I hope to have numbers for a 4 way box of these in a month or so. It
should be awesome! although availability will be a pain in the butt for a
few months I am sure.
All the best
Ross
On 2/19/13 8:40 AM, "filip fratev" <filipfratev.yahoo.com> wrote:
>Hi Ross and all,
>
>After a lot of rumors the GTX Titan card was
>released today. For the first time on any consumer-level NVIDIA card,
>double precision (FP64)
>performance is uncapped. That means 1/3 FP32 performance, or
>roughly 1.3TFLOPS theoretical FP64 performance. Thus this card is very
>similar
>(same for Amber use) to Tesla K20x, but costs 1000$ and will be much much
>faster! I suppose that Titan will break the 100+ ns threshold on JAC test
>with
>Amber 12 and have no patience to see some test results!
>
>All the best,
>Filip
>_______________________________________________
>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 Tue Feb 19 2013 - 10:00:03 PST
