Firefly and PC GAMESS Benchmarks

Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://classic.chem.msu.su/gran/gamess/performance.html
Äàòà èçìåíåíèÿ: Sun Apr 17 23:48:20 2011
Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 19:56:25 2012
Êîäèðîâêà:

Firefly and PC GAMESS Benchmarks

Firefly and PC GAMESS Benchmarks

Two years later: benchmarking Firefly version 8.0.0 beta on Intel Core i7 2600K AVX-enabled system
Not bad but not phenomenal at all: Firefly v. 7.1.G benchmarks on AMD Quad-core Phenom II X4 955 Black Edition 3.2 GHz system
Just CUDing: initial results for CUDA-enabled Firefly v. 7.1.G
Testing the monster: Firefly v. 7.1.G benchmarks on dual Intel Xeon W5580-based system

Hyperthreading that works: testing Intel Core i7 940 CPU with PC GAMESS/Firefly v. 7.1.F
Not kidding at all: testing Intel Core i7 CPU with PC GAMESS/Firefly v. 7.1.E
Just kidding: testing Intel Atom CPU with PC GAMESS/Firefly v. 7.1.C
PC GAMESS v. 7.1.9 performance and scalability on 24-core Intel Dunnington (Xeon L7455)-based system
PC GAMESS v. 7.1.5 benchmarks on AMD Athlon vs. AMD Phenom processors, as well as other useful documents on related topics by Marek Checinski.

PC GAMESS v. 7.1.4 benchmarks and scalability: Barcelona vs. Clovertown vs. Harpertown.
PC GAMESS v. 7.1.1 performance and scalability on Windows CCS (Compute Cluster Server)-based Opteron cluster.
PC GAMESS v. 7.1 performance and scalability on 16-core Intel Tigerton (Xeon X7350)-based system
PC GAMESS v. 7.1 new parallel MP2 gradient code benchmark and scalability on 16-node Pentium 4 Gigabit Ethernet Windows cluster
PC GAMESS v. 7.0.7 performance and scalability: MPICH-MX case
PC GAMESS v. 7.0.7 performance and scalability: comparing dynamic vs. static load balancing approaches
PC GAMESS v. 7.0.4 benchmarks and scalability on SKIF K-1000 (large Opteron Infiniband Linux cluster)
PC GAMESS v. 7.0.4 benchmarks and scalability on 21-node Pentium 4 Infiniband Linux cluster
PC GAMESS v. 7.0 standard MCQDPT2 benchmark on modern hardware.
PC GAMESS v. 7.0 standard MP4(SDTQ) benchmark on modern hardware.
PC GAMESS v. 7.0.2 benchmarks on four-core Intel Kentsfield (Core 2 Quadro) system
PC GAMESS v. 7.0.1 benchmarks and scalability on different eight-core computer systems
Preliminary PC GAMESS v. 7.0 benchmarks on Intel Conroe (Core 2 Duo) and PC GAMESS version 7.0.1 benchmarks on Intel Woodcrest and AMD dual-core Opteron pages also gracefully provided by Dr. Alex Popov.
PC GAMESS v. 7.0 benchmarks on different AMD Opteron processors.
PC GAMESS v. 7.0 benchmarks on different Intel Xeon (Netburst core) processors.
PC GAMESS v. 7.0 benchmarks on different Intel Pentium 4 family (Netburst) processors.

The PC GAMESS input and output of geometry optimization run of fullerene C₆₀ using our new canonical MP2 gradient code on Windows 2008 Server/CCP 2.0 cluster. The hardware used was exactly the same as described on this page. Each compute node has only one physical disk per four CPU cores (i.e., four PC GAMESS instances) thus seriously degrading IO performance, reducing CPU usage and increasing wall clock time. Having four physical HDDs per node would halve execution time.
The PC GAMESS input and output files for really large MP4(full) job (more than 7*10¹⁵ double precision floating point operations). The average performance of the PC GAMESS was higher than 60% of the theoretical maximum performance of the four-cpu computer system used for calculations. We are grateful to Dr. Kalju Kahn for providing us by this example.

The PC GAMESS vs. WinGamess performance comparison page gracefully provided by Dr. Alex Popov.

The PC GAMESS v. 5.2 SMP scaling results for large MP4-SDQ job. Note both the program version (May 1999) and hardware (four-CPU Pentium II Xeon 450 MHz SMP system) are ancient. However, we hope these data may be still of some historical interest.