home > products > Benchmarks

productsBoxCluster SeriesBoxClusterDSN BoxClusterNXi-24GG BoxClusterNXi BoxClusterMLRackmount SeriesRackMLHigh Speed InterconnectQuadrics QsNet^II Quadrics QsTenG Myricom Myri10G

Benchmarks >>

products

BoxCluster Series

BoxClusterML/NXi Benchmarks

STAR-CCM+ and BoxClusterNXi integration test results:
The test results indicate:
the simulation speed is almost 1.5 times faster than the previous generation system, BoxClusterML (16core), and it is more than 4 times after than the quad core Dell Precision T7400 system.

System Config:

BoxClusterNXi:

CPU: 4 * Core i7 940 (2.93GHz, L3 Cache 8MB, Bus speed 4.8GT/s)

MEN: 24 * 2GB DDR3 1333, Non-ECC

HDD: 5 * Seagate 250GB SATAII

STAR-CCM+ version 4.02

OS: RedHat 5.3

1. Civil_trim_20m

Cells	20,229,005
Interior Faces	60,349,506
Vetrices	20,274,462

2. Lemans_poly_17m

Cells	17,231,394
Interior Faces	114,159,241
Vetrices	95,305,955

3. DPW3_a0.5

Cells	10,985,953
Interior Faces	327,701,643
Vetrices	10,984,645

4. turbo

Cells	2,873,232
Interior Faces	15,892,433
Vetrices	12,606,218

Moldex3D R9.1 and BoxClusterML integration test results:
The test results indicate: the analysis speed is almost 5 times faster under Moldex3D R9.1 and one BoxClusterML (8-core), and almost 10 times faster under Moldex3D R9.1 and two BoxClusterMLs (16-core).

System Config:

CPU: Intel Xeon E3110 3.0GHz x 1 /node
Memory: DDR2-667 2G /node
HDD: 250GB SATA x 2 for Host node, x1 for Client
OS: CentOS 4.6 HPC Systems Edition

1. 16-Cavity Optical Lens

Mesh	3D Solid Mesh
Element of cavity and runner	1,066,448
Cooling Channels	33,600
Moldbase	1,347,283

2. Small CD Frame

Mesh	3D Solid Mesh
Element of cavity and runner	974,517
Cooling Channels	570
Moldbase	637,842

3. Automotive Component

Mesh	3D Solid Mesh
Element of cavity and runner	1,105,024

4. Electronic Fan

Mesh	3D Solid Mesh
Element of cavity and runner	496,405
Cooling Channels	17,952
Moldbase	539,670

Moldex3D R9.1 and BoxClusterNX integration test results:
The test results indicate: the analysis speed is more than 10 times faster under Moldex3D R9.1 and one BoxClusterNX (16 cores).

Software Information: Moldex3D/Solid-Flow R9.1 enhanced version
Hardware Information: BoxClusterNX (Intel Core i7)

1. Automotive grill

Mesh	3D Solid Mesh
Element of cavity and runner	713,588

2. 16-Cavity Optical Lens

Mesh	3D Solid Mesh
Element of cavity and runner	1,066,448
Cooling Channels	33,600
Moldbase	1,347,283

FLUENT

System Config	Ver.	#of Nodes	#of CPUs	#of cores	Small Class Job			Medium Class Job			Large Class Job
					FL5S1	FL5S2	FL5S3	FL5M1	FL5M2	FL5M3	FL5L1	FL5L2	FL5L3
					(rate)	(rate)	(rate)	(rate)	(rate)	(rate)	(rate)	(rate)	(rate)
Red Hat 5.3 Corei7 940 BoxClusterNXi	6.3.26	1	1	1	4,758	7,385	2,865	1,177	2,139	491	301	209	40
		1	1	4	15,610	16,991	10,135	3,365	7,053	1,662	977	778	145
		2	2	8	9,841	16,271	11,676	5,252	10,410	1,268	1,762	1,389	262
		4	4	16	11,636	12,896	14,521	6,926	15,498	1,156	3,072	2,570	481
Core2 Extreme QX9650 3.00GHz/6MBx2(L2) 8GBx4node BoxClusterML	6.3.26	1	1	1	4,120	5,363	2,267	952	1,680	418	249	171	32
		2	2	2	6,211	8,438	4,414	1,716	3,040	585	455	347	64
		4	4	4	8,205	11,676	7,909	2,713	5,503	915	876	674	122
		4	4	8	9,643	14,546	10,634	4,651	9,024	1,271	1,544	1,094	198
		4	4	16	12,308	14,281	12,993	6,912	13,714	1,203	2,667	1,752	295
Core2 Duo E6850 3.00GHz/4MB(L2) 4GB x8node BoxClusterMLx2	6.3.26	1	1	1	3,554	4,275	2,051	876	1,562	393	209	160
		1	1	2	6,330	6,021	3,383	1,472	2,566	636	368	268
		2	2	4	8,268	8,462	6,185	2,425	4,763	936	716	537
		4	4	8	9,526	13,671	10,082	4,359	8,252	1,343	1,391	1,055	192
		8	8	16	12,135	11,551	12,152	6,198	13,981	1,753	2,591	2,031	371

Abaqus

BoxClusterNXi Intel Core i7-940 / 12G *4		ABAQUS/Explicit						ABAQUS/Standard
	Core	e1	e2	e3	e4	e5	e6	s1	s2a	s3a	s3b	s4a	s4b	s4d	s5	s6
	1	18,455	10,319	7,727	3,535	4,526	8,940	65	2,103	299	2,478	636	9,595	3,790	1,061	5,009
	2	9,538	5,539	4,044	1,769	2,173	4,980	46	1,101	220	4,184	412	4,159	2,081	614	2,882
	4	5,721	2,893	2,348	1,003	1,121	2,993	40	589	164	2,191	296	1,695	1,253	371	1,838
	8	3,029	1,651	1,276	613	591	1,880	48	431	*	*	293	1,925	1,117	375	2,320
	16	1,961	1,064	755	370	348	1,256	46	339	*	*	249	1,185	787	267	1,609

*solver not for distribute system

VASP

System Config	#node	#CPU	#Core	Message Passing Device	Time Elapsed (sec)	VASP Ver.	Compiler	Math Library
CPU: Intel Xeon E3110 3.0GHz x 1 /node Memory: DDR2-667 2G /node HDD: 250GB SATA x 2 for Host node, x1 for Client OS: CentOS 4.6 HPC Systems Edition	4	4	8	GbE	20.238	4.6.28	ifc9.1	MKL 9.1
	2	2	4	GbE	28.228	4.6.28	ifc9.1	MKL 9.1
	1	1	2	SMP	43.592	4.6.28	ifc9.1	MKL 9.1
	1	1	1	-	66.915	4.6.28	ifc9.1	MKL 9.1

LS-DYNA / Model: neon_refined_revised

System Config	#of Total Cores	#of cores	#of node	Elapsed Time (sec)	Speed Up factor
OS:CentOS 4.6 64bit (BoxCluster original configuration) HP-MPI: hpmpi-2.02.05.01-20070708r LS-DYNA Ver. : mpp971_s_R3.1_Intel_linux86-64_hpmpi	8	2	4	1410	4.88
	4	1	4	2001	3.44
	4	2	2	2493	2.76
	2	1	2	3886	1.77
	2	2	1	4616	1.49
	1	1	1	6883	1.00

The result of 8core configuration is listed on www.topcrunch.org with neon_refined_revised model.

HPL

System Config	#of Nodes	#of cores	Matrix size	Result
OpenMPI 1.2.4 OS: RedHat ELWS 4.6 (x86_64) ICC: 9.1.052 IFORT: 9.1.052 MKL: 9.1enh023	1	4	31000 N	37.9 Gflops
	4	16	60000 N	127.8 Gflops

STAR-CD

System	# Node	# CPU	# Core	Time(sec)	benchmark name
System	# Node	# CPU	# Core	Time(sec)	aclass	ahmed	asmo	engine	interior	sbend
Core2 Extreme QX9650 (Yorkfield) 3.00GHz/6MBx2(L2) BoxClusterML STAR-CD v3.26.059	4	4	16	CPU	502	55	50	26	51	47
	4	4	16	Elapsed	540	66	63	32	60	71
	4	4	8	CPU	558	83	85	45	82	84
	4	4	8	Elapsed	587	96	91	54	97	99
	4	4	4	CPU	871	161	192	90	167	156
	4	4	4	Elapsed	909	175	201	97	176	171
	2	2	2	CPU	1,745	331	482	179	320	311
	2	2	2	Elapsed	1,759	342	491	190	331	323
	1	1	1	CPU	3,490	703	1,110	374	632	629
	1	1	1	Elapsed	3,490	703	1,110	374	632	629
Core2 Duo E6850 3.0GHz/4M(L2) BoxClusterMLx2 STAR-CD v3.26.019	8	8	16	CPU	333
	8	8	16	Elapsed	426
	4	4	8	CPU	660	182	144	54	102	212
	4	4	8	Elapsed	707	263	169	63	118	238