Computational Chemistry and Biochemistry
The following applications have been accelerated using the ClearSpeed CATS system:
- Molpro: molecular electronic structure code
- Sire: molecular modeling and simulation using QM/MM free energy code
- BUDE: molecular dynamics-based drug design
- Amber: molecular dynamics simulation
Molpro / Sire
Accelerated Sire/Molpro was used for modeling of neuraminidase inhibitors to vaccinate against the Influenza virus.
- Sustained >220 double precision GFLOPS on the Molpro application per CATS node
- With 10 CATS nodes achieved 2.2 TFLOPS sustained in a single rack
- Executed ~60x1015 64-bit floating point operations on Molpro in ~17 hours spread over 3 days during SC07
- 7.3X speedup per CATS node
- Enables 1 ligand per day with QM/MM levels of accuracy: 55 QM atoms, ~1600 MM atoms, DFT BLYP VDZ
The acceleration of Molpro is described in more detail in Massively Multicore Parallelization of Kohn-Sham Theory by Philip Brown, Christopher Woods, Simon McIntosh-Smith, and Frederick R. Manby.
For more information see: Molpro and Sire.
BUDE
The Bristol University Docking Engine (BUDE) uses Molecular Dynamics to perform Evolutionary Monte Carlo (EMC) based drug docking. Used for research into peptide-based protease inhibitors.
- Millions of docking operations required
- The energy of billions of poses must be calculated
- 10.2 x speedup and 5.6x less energy consumed per CATS node compared to the latest 3.0GHz quad core CPU
BUDE is currently under development as a generic molecular docking program under the direction of Dr. Richard Sessions and Professor Anthony Clarke of the Bristol Protein Folding Group.
Acceleration of this algorithm has been obtained on the ClearSpeed Advance X620 accelerator board by porting the energy calculation and geometry routines to the ClearSpeed Advance accelerator.
Amber
ClearSpeed acceleration of Amber 9 has been shown to deliver a speedup of between 3.4x and 9.4x for compute intensive sections of the code using a single Advance X620 accelerator board.
The Amber 9 methods that include the effective radius and force calculation of AMBER’s Generalized Born (GB) models, 1, 2, and 6 have been modified to take advantage of ClearSpeed’s Advance accelerator board. Supported options include constant pH7 and analytical linearized Poission Boltzmann (ALPB) as well as options that do not directly change the force calculation, including NMR restraints.
While the genborn module of Amber is a small part of the sander executable, it typically amounts for 95-97% of the CPU compute time for GB simulations. The CPU compute time is mainly spent in three loops: effective radii calculations, diagonal and off-diagonal force calculations.
The overall structure of the code was maintained. A thin layer written in C, using ClearSpeed’s CSAPI library, was added to handle the communication between the host and board.
For more information: Amber.
Experience ClearSpeed Accelerated Amber for yourself! Click here to get started.