Cores

From FaHWiki
(Difference between revisions)
Jump to: navigation, search
(Gromacs SMP2 core)
(Add a6 minimal info)
Line 123: Line 123:
 
| [[#Gromacs SMP2 bigadv core|Gromacs SMP2 bigadv]] (Version 2.27)
 
| [[#Gromacs SMP2 bigadv core|Gromacs SMP2 bigadv]] (Version 2.27)
 
| GRO-A5
 
| GRO-A5
 +
| SMP2 multithread bigadv
 +
| [[SMP client]]
 +
|-
 +
| FahCore_a6
 +
| [[#Gromacs SMP2 bigadv core|Gromacs SMP2 bigadv]] (Version 2.28)
 +
| GRO-A6
 
| SMP2 multithread bigadv
 
| SMP2 multithread bigadv
 
| [[SMP client]]
 
| [[SMP client]]

Revision as of 04:44, 23 May 2011

Contents

Introduction

The 'core' is the program that performs the calculations. The same core is used by various versions of the client and is automatically updated whenever necessary. This provides an easy way for the scientific calculations to be improved without requiring you to install a new version of the client.

There are several different cores:

FahCore # Scientific Name Project Summary Name Special Optimizations Client
FahCore_11 GPU2 (NV Version 1.31) GROGPU2 ATI CAL / NV CUDA GPU client
FahCore_12 GPU2 ATI-DEV ATI DEVELOPMENT GPU client
FahCore_13 GPU2 NVIDIA-DEV NVIDIA DEVELOPMENT GPU client
FahCore_14 GPU2 GROGPU2-MT NV CUDA GPU client
FahCore_15 GPU3 (Version 2.09) OPENMMGPU NV OPENMM GPU client
FahCore_16 GPU3 OPENMMGPU NV OPENMM / ATI OpenCL GPU client
FahCore_20 SHARPEN SHARPEN  ?? SMP client
FahCore_78 Gromacs (Version 1.90) GROMACS SSE CPU client
FahCore_79 Double Gromacs DGROMACS SSE2 CPU client
FahCore_7a GB Gromacs GBGROMACS SSE CPU client
FahCore_7b Double Gromacs B DGROMACSB SSE2 CPU client
FahCore_7c Double Gromacs C DGROMACSC SSE2 CPU client
FahCore_80 Gromacs SREM GROST SSE CPU client
FahCore_81 Gromacs Simulated Tempering GROSIMT SSE CPU client
FahCore_82 Amber (Version 1.03) AMBER None CPU client
FahCore_a0 Gromacs 33 GROMACS33 SSE CPU client
FahCore_a3 Gromacs SMP2 (Version 2.22) GRO-A3 SMP2 multithread SMP client
FahCore_a4 GB Gromacs GRO-A4 SSE CPU client
FahCore_a5 Gromacs SMP2 bigadv (Version 2.27) GRO-A5 SMP2 multithread bigadv SMP client
FahCore_a6 Gromacs SMP2 bigadv (Version 2.28) GRO-A6 SMP2 multithread bigadv SMP client
FahCore_b4 ProtoMol (Version 25) ProtoMol SSE2, SSE3, SSSE3, SSE4.1 and SSE4.2 CPU client
FahCore_?? Desmond Desmond?  ? CPU client
Unknown Gromacs PS3 GRO-PS3 Cell PS3 client
...Retired...
FahCore_10 GPU GROGPU Retired
FahCore_65 Tinker TINKER Retired
FahCore_96 QMD QMD Retired
FahCore_a1 Gromacs SMP (Version 1.74) GRO-SMP SSE, SMP multicore SMP client
FahCore_a2 Gromacs SMP (Version 1.94) GROCVS SSE, SMP multicore, Linux/Mac Only SMP client


Note: The numbers associated with the names of the cores are arbitrary identifiers. They do not indicate a version number which can be used to tell an "old" core from a "new" one. Version 1.86 of FahCore_78 may or may not be older than Version 2.53 of FahCore_65.

See also: Core MD5 Sums

Gromacs core

Gromacs (Groningen Machine for Chemical Simulations) is a single-precision scientific core that is used for the majority of F@H calculations. It is developed by the University of Groningen under the GPL and is writen in Fortran77-based GROMOS in C. Gromacs takes advantage of SIMD, including SSE and AltiVec instructions in supported processors to provide nearly a threefold increase in calculation speed. Modern processors such as the Pentium 3, Pentium 4, Athlon XP and Athlon 64 series, as well as their mobile counterparts, are SSE capable. The PowerPC G4 and G5 series are Altivec capable.

More information can be found here:
http://folding.stanford.edu/gromacs.html
Gromacs Home page


Core Information
Core number 78
Core Name Folding@Home Gromacs Core
Operating System/Architecture Windows/x86, Linux/x86, Mac OS X/PPC


Double Gromacs core

The DoubleGromacs core is very similiar to the Gromacs core except it performs double-precision rather than single-precision calculations. DoubleGromacs takes advantage of SSE2 instructions in supported processors to provide about a 2x increase in calculation speed. Newer processors such as the Pentium 4 and the Athlon 64 have SSE2 capability.


Core Information
Core number 79
Core Name Folding@Home Double Gromacs Core
Operating System/Architecture Windows/x86, Linux/x86, Mac OS X/PPC


GB Gromacs core

The GB Gromacs core is very similiar to the Gromacs core. The Pande Group added the "Generalized Born implicit solvent" model to GB Gromacs codebase; this should make it possible to drop or reduce the use of Tinker in the future.

A new version of the a4 fahcore (v2.25), based on Gromacs v4.5.1, was released in October 2010. Folding Forum: P10412 now on advanced


Core Information
Core number 7a
Core Name Folding@Home GB Gromacs Core
Operating System/Architecture Windows/x86, Linux/x86, Mac OS X/PPC


Core Information
Core number a4
Core Name Folding@Home GB Gromacs Core
Operating System/Architecture Windows/x86


Double Gromacs B core

The Double Gromacs B core is very similiar to the Double Gromacs core. This new FahCore_7b has several scientific additions to the core. Initially released only to the Linux platform in August 2007, it will eventually be available for all platforms.


Core Information
Core number 7b
Core Name Folding@Home Double Gromacs Core B
Operating System/Architecture Windows/x86, Linux/x86, Mac OS X/PPC


Double Gromacs C core

The Double Gromacs C core is very similiar to the Double Gromacs core. Initially released for Linux and Windows in April 2008.


Core Information
Core number 7c
Core Name Folding@Home Double Gromacs Core C
Operating System/Architecture Windows/x86, Linux/x86, Linux/x86_64


Gromacs 33 core

The Gromacs 33 core contains features from more recent versions of Gromacs which have been ported to Folding@home, allowing a broader range of simulations to be run.


Core Information
Core number a0
Core Name Folding@Home Gromacs 3.3 Core
Operating System/Architecture Windows/x86, Linux/x86, Mac OS X/x86


Gromacs SMP core

The Gromacs SMP core supports running on multiprocessor or multicore processor systems. This core is currently available for Mac OS X on Intel, 64-bit Linux and Windows. 32-bit Linux support is being worked on.

This core makes running multiple clients for each CPU or CPU-core obsolete. The core will make use of the two or more cores available in the system.

See: Folding-community: SMP client open beta and Folding@home on multi-core/SMP boxes (SMP FAQ)


Core Information
Core number a1
Core Name Folding@Home Gromacs SMP Core
Operating System/Architecture Linux/x86_64, Mac OS X/x86, Windows/x86


Gromacs CVS SMP core

The Gromacs CVS SMP core supports running on multiprocessor or multicore processor systems. This core is currently available for only Mac OS X on Intel and 64-bit Linux. A Windows version, and a 32-bit Linux support is being worked on.

This core makes running multiple clients for each CPU or CPU-core obsolete. The core will make use of the two or more cores available in the system.

See: Folding-community: SMP client open beta and Folding@home on multi-core/SMP boxes (SMP FAQ)

This core looks almost identical to the Gromacs SMP core (a1) because it uses mostly the same code base from Gromacs. The Gromacs CVS SMP core (a2) only uses more recent code from Gromacs than the Gromacs SMP core (a1) core does, and supports some more features (like more CPUs) than the Gromacs SMP core (a1) does at current.

Specifying the -bigadv flag requests Really BIG Work Units (WU) using 8+ fast cores, 6+ GB RAM, tight deadlines and more network bandwidth. See the announcement Extra-large work units (bigadv) and the New bonus plan TRIAL (bigadv)


Core Information
Core number a2
Core Name Folding@Home Gromacs SMP Core
Operating System/Architecture Linux/x86_64, Mac OS X/x86


Gromacs SMP2 core

First SMP2 core to use a threads-based parallelization instead of MPI for its Inter-process communication

See: Folding Forum: SMP2 with passkey (core_a3) V6.29Beta Folding@home Blog: SMP2 core update

As of approximately 3/3/2011, this core a3 no longer runs bigadv WU's, allowing optimizations for non-bigadv SMP WU's. See the new Gromacs SMP2 bigadv core for running bigadv WU's.

Core Information
Core number a3
Core Name Folding@Home Gromacs SMP Core
Operating System/Architecture Linux/x86_64, Mac OS X/x86, Windows/x86


Gromacs SMP2 bigadv core

Derivative SMP2 core using a threads-based parallelization instead of MPI for its Inter-process communication optimized for bigadv WU's.

See: Folding Forum: upcoming core changes (bigadv will switch from A3 to A5) Folding@home Blog: Introduction of a new SMP core, changes to bigadv

Specifying the -bigadv flag requests Really BIG Work Units (WU) using 8+ fast cores, 6+ GB RAM, tight deadlines and more network bandwidth. See the announcement Extra-large work units (bigadv) and the New bonus plan TRIAL (bigadv)

Core Information
Core number a5
Core Name Folding@Home Gromacs SMP Core
Operating System/Architecture Linux/x86_64, Mac OS X/x86, Windows/x86


SHARPEN core

See: Folding@Home #N FAQ

Core Information
Core number 20
Core Name Folding@Home SHARPEN Core
Operating System/Architecture Windows/x86


Tinker core

The Tinker core was the second-most used core for F@H calculations (superseded by AMBER). It does not use any special instructions to accelerate floating-point operations on modern processors (such as SSE or 3dNow+). WorkUnits that use the Tinker core usually perform well on AMD processors.


Core Information
Core number 65
Core Name Folding@Home Client Core
TINKER: Software Tools for Molecular Design
Operating System/Architecture Windows/x86, Linux/x86, Mac OS X/PPC


QMD core

QMD uses a quantum chemical method for calculations and uses a large amount of memory (sometimes in excess of 512 MiB of RAM).

Note: Because of the licensing issues (FAH & QMD & AMD64 & SSE2) QMD WUs are currently not assigned to AMD CPUs.

Note: The QMD core is multi-thread capable. Folding-community: OMP_NUM_THREADS

There are no projects at the moment utilizing the QMD core: Folding-community: Vijay Pandes post in "Where did they go?? [QMDs"]

But it may become active again, see Folding-community: Vijay Pande's post in: What are we ACTUALLY work on?


More information can be found here: http://folding.stanford.edu/QMD.html


Core Information
Core number 96
Core Name Folding@Home QMD Core
Operating System/Architecture Windows/x86, Linux/x86


AMBER core

The AMBER core (Assisted Model Building and Energy Refinement) is a molecular dynamics program originally developed by Peter Kollman's group at the University of California, San Francisco. Its main use is in force field calculations. The project is now coordinated by David A. Case at Scripps Research Institute. The AMBER core does not take advantage of SSE to provide an increase in calculation speed, but hopefully will use SSE2 in the future.

There is a project to integrate at least a portion of the AMBER force field algorithm into the Gromacs core (ffamber). It is not clear if this will negate the use of the AMBER core.
http://folding.stanford.edu/ffamber/

More information can be found here:
http://folding.stanford.edu/AMBER.html
http://en.wikipedia.org/wiki/AMBER


Core Information
Core number 82
Core Name Folding@Home PMD Core
Operating System/Architecture Windows/x86, Linux/x86


GPU core

The GPU core (Graphics Processor Unit) uses the graphics chip of modern video cards to do molecular dynamics. The first GPU client supports ATI 1xxx series of GPUs.
"The GPU Gromacs core isn't a port of Gromacs, but rather we've taken key elements from Gromacs we need and enhanced them based on the unique capabities of GPU's. Thus, it's really a new and different core on the inside, but with wrapped Gromacs on the outside" -- Vijay Pande, Getting close to the GPU beta launch


Core Information
Core number 10
Core Name Folding@Home GPU Core
Operating System/Architecture Windows/x86


GPU2 core

The GPU core (Graphics Processor Unit) uses the graphics chip of modern video cards to do molecular dynamics. This second generation GPU client supports ATI 2xxx/3xxx or later, and nVidia 8xxx or later series of GPUs. This new client uses AMD's CAL or nVidia's CUDA instead of Microsoft's DX9.


Core Information
Core number 11
Core Name Folding@home GPU2 Core
Operating System/Architecture Windows/x86


Core Information
Core number 12
Core Name Folding@home GPU2 Core
Operating System/Architecture Windows/x86


Core Information
Core number 13
Core Name Folding@home GPU2 Core
Operating System/Architecture Windows/x86


Core Information
Core number 14
Core Name Folding@home GPU2 Core
Operating System/Architecture Windows/x86


GPU3 core

The GPU core (Graphics Processor Unit) uses the graphics chip of modern video cards to do molecular dynamics. This third generation GPU client/core was mentioned in the project NEWS. More details to follow.


Core Information
Core number 15
Core Name Folding@home GPU3 Core
Operating System/Architecture Windows/x86


Core Information
Core number 16
Core Name Folding@home GPU3 Core
Operating System/Architecture Windows/x86


Gromacs SREM core

The Gromacs Serial Replica Exchange Method core, also known as GroST (Gromacs Serial replica exchange with Temperatures), uses Replica Exchange Method in its simulations also known as REMD, Replica Exchange Molecular Dynamics.

See: Folding-community: uncle_fungus in GROST

See also: "Replica-exchange molecular dynamics method for protein folding"


Core Information
Core number 80
Core Name Folding@Home Gromacs SREM Core
Operating System/Architecture Windows/x86, Linux/x86


Gromacs Simulated Tempering core

GROSimT is a new (May 2007) Gromacs core that performs simulated tempering. Simulated tempering is one of the sampling algorithms, of which the basic idea is to enhance sampling by periodically raising and lowering temperature. Using this new core, we are hoping to sample more efficiently the transitions between folded and unfolded conformations of proteins.

See also: Simulated tempering: a new Monte Carlo scheme. Authors:, Marinari, E.; Parisi, G. Publication:, Europhysics Letters, Vol. 19, p.451.


Core Information
Core number 81
Core Name Folding@home Gromacs Simulated Tempering Core
Operating System/Architecture Linux/x86, Windows/x86


ProtoMol core

We have been working on another approach to speeding dynamics greatly, based on a new technique called Normal Mode Langevin (NML) dynamics. This method uses the same style models as normal MD (same force fields, etc) and thus should have the same accuracy, but with a pretty significant speedup due to algorithmic advances. NML is complementary to our other methods, so we're hoping to add it to everything else (in particular to the GPU core). To start, we will be testing it in a new core, based on the Protomol software. Protomol is designed to allow for rapid prototyping of molecular simulations, which is perfect for NML.

See http://protomol.sourceforge.net/

Core Information
Core number b4
Core Name Folding@home Protomol Core
Operating System/Architecture Linux/x86, Linux/x86_64, Windows/x86


Desmond core

Place holder for a new fahcore mentioned on the project News page here. And a link to the developer D E Shaw Research


Core Information
Core number  ??
Core Name Folding@home Desmond Core?
Operating System/Architecture Linux/x86, Linux/x86_64, Windows/x86 ?


Links

Personal tools