Alessandro Pedretti

A new distributed paradigm for parallel computing

UNIVERSITÁ DEGLI STUDI DI MILANOFacoltà di Scienze del Farmaco

Typical scenario in a lab

Internet

Firewall

Servers

PCs

Networkdevices

Ethernet infrastructure100-1000 Mbit/s

Several PCs with heterogeneous hardware / OSs.

Very high computational power “fragmented” on the local network.

Hard possibility to use all computational power to run a single complex calculation.

Parallel computing without cluster paradigm.

Client/server architecture with hot-plug capabilities.

Possibility to perform calculations with different pieces of software without changing the main code.

Expandable by scripting languages.

High-level database interface integrated in the main code and supporting the most common SQL database engines (Access, MySQL, SQLite, SQL Server, etc).

Easy configuration by graphic interface.

High performances and security.

Main features

… to develop WarpEngine:

What we need …

High-level database interface.

Fast customizable Web server.

Script engine.

Graphic environment.

Plug-in expandability

Scripting languages

Molecule editing

Surface mapping

File format conversion

Database engine

Graphic interface

Property calculation

MM / MD calculations

Trajectory analysis

Server scheme

UDP server HTTP server

Client manager

Project manager

Jobmanager

VEGA ZZcore

Databaseengine

IP filterPowerNetplug-in

Main program

To clientsTCP/IP, HTTP,broadcast

Optional encrypted tunnelprovided by WarpGate

Client scheme

UDP client HTTP client

Project manager

Multithreadedworker

VEGA ZZcore

PowerNet plug-in Main program

To the serverTCP/IP, HTTP, broadcast

Graphic interface

Hardware for the test

1 PC configured as client and server:Quad-core

9 PC configured as client:1 six-core7 quad-core1 dual-core1 single-core

37 cores42 Gb ram

> 3 Tb storage

Operating systems:6 Windows 7 Pro x643 Windows 7 Pro1 Windows XP Pro

Network connection:Ethernet 100 Mbs

Communication stress test:delivery of empty jobs to the clients and receive of the result from them.

79.651,78 jobs / min.

Preliminary performance test

Database stress test:extraction (by SQL query), decompression and delivery of molecules to the clients and answer.

41.115,00 molecules / min.

Apache Bench 2.0.41100 requests with concurrency level of 5.

3.205,13 pages / sec.1,560 ms / request

Microsoft IIS 6.01.066,67 pages / sec.

4,688 ms / request

Software & data for the test

APBS – Adaptive Poisson-Boltzmann SolverCalculation of solvation energy.

PLANTS – Protein-Ligand ANT systemStructure-based virtual screening.

Database of drugs in .mdb format174.398 molecules, average MW 353,70.

Human M2 muscarinic receptorPDB ID: 3UON.

Both programsare single-threaded

APBS – Solvation energy calculation.174.398 molecules, two APBS calculation for each molecule (reference and solvated state).

Time required by a single thread calculation: 13 days 5 hours

Time required by WarpEngine: 8 hours 36 minutes

WarpEngine speed: 339,10 jobs / min.

Real case tests

PLANTS – Virtual screening.174.398 molecules, M2 target, search speed 2.

Time required by a single thread calculation: 36 days 22 hours

Time required by WarpEngine: 1 day 0 hour 1 minute

WarpEngine speed: 121,00 jobs / min.

WarpEngine is easy expandable by scripting language, so it’s possible to add some other calculation types:

The future …

Semi-empirical calculations MOPAC

Ab-initio calculations FireFly / PC GAMESS

Other virtual screening methods AutoDock, Vina

Rescore of docking poses VEGA, XScore

Molecular mechanics calculations AMBER, AMMP, NAMD

Other applications …