Radar Interferometry Group Computer Resources

Uniprocessors

Name	mflops	SPECfp2000	Vendor	Model	CPU	Clock (MHz)	Cache (kByte)	Memory (MByte)	LAN (Mbit/s)	OS
pfase fogo etna mitchell461-dell2	901	679	Dell	Precision WorkStation 340	Pentium IV	1800	256	512	1000 or 100	Fedora Core 1
drummer0 : drummer9	789	615	Dell	Precision WorkStation 340	Pentium IV	1500	256	512	1000	Fedora Core 1
hzmac	369		Apple	PowerMac G4	PowerPC	400	1024	1280	100	MacOS 10.3
elvis frank kurt tupac wendy	207		Dell	Optiplex GX1	Pentium III	500	512	256	100	Fedora Core 1

Name	mflops	SPECfp_rate95 or SPECfp_rate2000	Vendor	Model	CPU	Clock (MHz)	Cache (kByte)	#CPU	Memory (MByte)	LAN (Mbit/s)	OS
jukebox	453	7.7 (2000)	Dell	PowerEdge 6400	Pentium III Xeon	700	1024	4	3072	1000 and 100	Fedora Core 1
buddy cass duane janice jimi keith	317	243	Dell	Precision Workstation 610	Pentium III Xeon	550	512	2	512	100	Fedora Core 1
rocknroll	290	n/a	Dell	PowerEdge 6300	Pentium III Xeon	500	512	4	512	1000 and 100	Red Hat 7.3
chirp	261	233	Dell	Precision Workstation 610	Pentium II Xeon	450	512	2	512	100	Fedora Core 1

mflops: Floating point performance reported by fftw_test -s 2048 for a single precision 2048 point FFT. Multi-threading was not enabled, even on multiprocessor machines. Threfore this should represent the performance of any machine running our single-threaded radar processor, which is FFT intensive. See the benchFFT page for more info. benchFFT methodology for more info.
SPECfp: Floating point performance for a uniprocessor reported by the manufacturer to SPEC. See the SPEC home page for more info.
SPECfp_rate: Floating point performance for a multiprocessor reported by the manufacturer to SPEC. Roughly scales with the number of processors in the machine. Must write a parallel application (using threads, shared memory constructs, or sockets) or have a parallelizing compiler and threaded libraries to take advantage of all processors. See the SPEC home page for more info.