September 1996 Order Number: 242999-001 Pentium® Processor Workst

The Datasheet Archive - 100 Million Datasheets from 7500 Manufacturers.

September 1996 Order Number: 242999-001 Pentium® Processor Workst

Datasheet Thumbnail

Top Searches for this datasheet

Pentium® Processor Workstation Performance Brief
September 1996 Order Number: 242999-001
Pentium® Processor Workstation Performance Brief
Information this document provided connection with Intel products. license, express implied, estoppel otherwise, intellectual property rights granted this document. Except provided Intel's Terms Conditions Sale such products, Intel assumes liability whatsoever, Intel disclaims express implied warranty, relating sale and/or Intel products including liability warranties relating fitness particular purpose, merchantability, infringement patent, copyright other intellectual property right. Intel products intended medical, life saving, life sustaining applications. Intel make changes specifications product descriptions time, without notice. Pentium® processor contain design defects errors known errata. Current characterized errata available request. *Third-party brands names property their respective owners. Contact your local Intel sales office your distributor obtain latest specifications before placing your product order. Copies documents which have ordering number referenced this document, other Intel literature, obtained from: Intel Corporation P.O. 7641 Prospect, 60056-7641 call 1-800-879-4683
COPYRIGHT INTEL CORPORATION 1996
Pentium® Processor Workstation Performance Brief
CONTENTS
PAGE INTRODUCTION INTEL PENTIUM PROCESSOR PRODUCT FEATURE HIGHLIGHTS.6 SPEC.7 APPLICATION PERFORMANCE BENCHMARKS
SYSmark* Windows NT*. LINPACK* 100x100 Benchmark. MicroStation* Performance Benchmark. MSC/NASTRAN*. BCELL*. ANSYS* Parametric Technology Corporation Pro/ENGINEER*. Simulators
GRAPHICS PERFORMANCE SUMMARY
Viewperf*. Viewperf CDRS* Viewset Viewperf Design Review* Viewset Viewperf Data Explorer* Viewset. Advanced Visualizer* Viewset (AWadvs-01) Lightscape* Viewset (Light-01).
SUMMARY APPENDIX TEST CONFIGURATIONS
Website References.
Pentium® Processor Workstation Performance Brief
FIGURES
Figure SPECint_base95* Benchmark. Figure SPECfp_base95* Benchmark. Figure SPECint_rate95*. Figure SPECfp_rate95* Benchmark Figure SYSmark* Windows Benchmark Figure LINPACK MFLOPS Performance Figure MicroStation* Performance Benchmark. Figure MSC/NASTRAN* Power Train Vibrational Model Performance Figure MSC/NASTRAN* Power Train Vibrational Model Performance Figure MSC/NASTRAN* BCELL* Model Performance ~10000 Degrees Freedom Figure MSC/NASTRAN* BCELL* Model Performance ~20000 Degrees Freedom Figure ANSYS* Performance. Figure ANSYS* Performance. Figure ANSYS* 5.3/Intergraph ANSYS* 5.2/non-Intel systems Figure Pro/ENGINEER* Performance Figure Raynet-gate TI-Mult. Figure CPU-Mix Dmult Figure CDRS* Composite Scores. Figure Design Review* Composite Scores. Figure Data Explorer* Composite Scores Figure AWadvs-01 Composite Numbers Figure Light-01 Composite Scores.
TABLES
Table Summary CDRS* Tests Table Summary Design Review* Tests Table Summary Data Explorer* Tests. Table Summary Advanced Visualizer* Tests Table Summary Lightscape* Tests.
Pentium® Processor Workstation Performance Brief
INTRODUCTION
1This performance brief compares performance Pentium® processor systems other non-Intel processor-based workstations, such those based SPARC*, MIPS*, PowerPC*, PA-RISC*, Alpha* architectures. Traditionally, Intel Architecture processors have been used personal computers running operating systems like DOS* Windows* business personal productivity applications such word processing, spreadsheets, presentations. Workstations have been used scientific technical applications such mechanical computer-aided design analysis (MCAD), electronic computer-aided design (ECAD), scientific modeling analysis, high performance graphics imaging, other computationally intensive tasks. Workstations often include two-dimensional threedimensional graphics accelerators. distinction between personal computer workstation been based processor performance, operating system features, graphics capabilities, applications. Today, that distinction being blurred availability high performance Pentium processors that being used workstation server products from variety vendors. widespread adoption high speed peripheral interconnect allowed workstation-class graphics accelerators attached Pentium Pentium processor-based systems. availability UNIX*, Microsoft Windows NT*, SCSI disks high-speed network connectivity Intel architecture provides robust multitasking capabilities. high performance Intel processors caused many traditional workstation applications migrate Intel platform. Pentium Pentium processors provide workstation capabilities competitive performance levels superior price/performance. This performance summary documents performance Pentium processors variety workstation benchmarks. Competitive results included other non-Intel processor-based workstations which performance data available. This performance summary does attempt provide benchmark results workstations throughout price range workstations available. Instead focuses those price range Intel architecture-based workstation. Hence justification including particular workstation this report based list price workstation platform. Intel Pentium processor-based workstations targeted introductory volume workstation markets. This performance brief examines those platforms from other non-Intel processorbased workstations that considered these markets. some performance comparisons throughout this brief, platforms architectures appear that exceed this qualification. These included reference only since configuration architecture considered workstation platform. these cases, Intel processor-based workstation have been configured match certain characteristics other platforms. Intel Pentium processor-based systems denoted graphs with above system's when non-Intel processor-based platforms included. graphics portion paper, this symbol used represent graphics accelerator card tested Pentium processor-based platform.
INTEL PENTIUM® PROCESSOR
Pentium processor family next generation Intel's processor technology. Pentium processor family currently consists following processors:
Endnotes System configurations, including software hardware design, affect actual performance.
Pentium® Processor Workstation Performance Brief
Pentium processor (512K cache) Pentium processor (256K cache) Pentium processor (256K cache) Pentium processor (512K cache)
While fully compatible with existing Intel architecture-compatible software, Pentium processor specifically optimized 32-bit software. This includes demanding software like CAD/CAM, numerical analysis, multimedia authoring applications running workstations. Pentium processor delivers superior performance over wide-range applications that exist today applications that coming future. achieves this superior performance through innovative architectural technique called "Dynamic Execution". Dynamic Execution removes software bottlenecks incorporated features such out-of-order execution register-renaming. Pentium processor design also includes glueless logic support four processors multiprocessor design. Other high-performance features include tightly coupled cache advanced processor bus. Manufactured using Intel's high-volume well-proven 0.35 technology, Pentium processor delivers outstanding integer floating-point performance. (See Figures 1-4.) following numbers peak rating SPEC benchmarks represent results Fujitsu superserver J650I* containing single Pentium processor running with 512K cache.
SPECint95* rating 8.71 SPECfp95* rating 6.68
Pentium processor been designed with balanced integer floating-point performance. This balanced integer floating-point performance design gives edge workstation applications. example, MSC/NASTRAN*, state-of-the-art mechanical engineering tool that very floating point intensive, runs particularly well Pentium processor processor's balanced design. Results MSC/NASTRAN Intel processor-based workstations reported contrasted with other workstations later this document.
PRODUCT FEATURE HIGHLIGHTS
Fully compatible with entire library software based operating systems such Windows NT*, UNIX* SVR4*, UNIX*, NEXTSTEP*, Solaris*, OS/2*, well Windows 95*, Windows Workgroups 3.11*, Windows 3.1*, DOS. onboard, high-speed secondary cache advanced processor provide bandwidth needed computer-intensive applications. Glueless 4-way delivers superior performance provides cost-effective upgrade path. Support enhanced data integrity reliability features mission-critical applications: (Error Checking Correction), Fault Analysis Recovery, Functional Redundancy Checking. Dynamic Execution:
Multiple Branch prediction allows many branches outstanding Dataflow analysis enables out-of-order execution over tens instructions Speculative Execution enables many instructions executed beyond program counter
Headroom performance growth through process improvement. Pentium processor manufactured using well-proven 0.35 process.
Pentium® Processor Workstation Performance Brief
Dynamic Execution involves optimally adjusting instruction execution predicting program flow, analyzing program's dataflow graph choose best order execute instructions, then speculatively executing instructions preferred order. Branch prediction concept found many mainframe high-speed microprocessor architectures. allows processor decode instructions beyond branches keep instruction pipeline full. Pentium processor, instruction fetch/decode unit uses highly optimized branch prediction algorithm predict direction instruction stream through multiple levels branches, procedure calls, returns. Pipe-lined machines must fetch next instruction before they have completely executed previous instruction. previous instruction branch, then next-instruction fetch could have been wrong place. Branch prediction technique that attempts infer correct next instruction address, knowing only current one. Typically uses Branch Target Buffer (BTB), small separate memory that watches instruction cache index attempts predict which index should accessed next, based branch history. Dynamic data flow analysis involves real-time analysis flow data through processor determine data register dependencies detect opportunities out-oforder instruction execution. Pentium processor dispatch/execute unit simultaneously monitor many instructions execute these instructions order that optimizes processor's multiple execution units, while maintaining integrity data being operated This out-of-order execution keeps execution units busy even when cache misses data dependencies among instructions occur. Speculative execution refers processor's ability execute instructions ahead program counter ultimately commit results order original instruction stream. make speculative execution possible, Pentium processor microarchitecture decouples dispatching executing instructions from commitment results. processor's dispatch/execute unit uses data-flow analysis execute available instructions instruction pool store results temporary registers. retirement unit then linearly searches instruction pool completed instructions that longer have data dependencies found, retirement unit commits results these instructions memory and/or Intel Architecture registers (the processor's eight general-purpose registers eight floating-point registers) order they were originally issued retires instructions from instruction pool. Through deep branch prediction, dynamic data-flow analysis, speculative execution, Dynamic Execution removes constraint linear instruction sequencing between traditional fetch execute phases instruction execution. allows instructions decoded deep into multi-level branches keep instruction pipeline full. promotes out-oforder instruction execution keep processor's instruction execution units running full capacity. Finally, commits results executed instructions original program order maintain data integrity program coherency. summary, Dynamic Execution technology optimally adjusts instruction execution predicting program flow analyzing program's dataflow graph choose best order execute instructions, then having ability speculatively execute instructions preferred order. Pentium processor dynamically adjusts work, defined incoming instruction stream, minimize overall execution time.
SPEC
SPEC95* software benchmark product produced Standard Performance Evaluation Corp. (SPEC), non-profit group computer vendors, system integrators, universities, research organizations, publishers consultants throughout world. provides measures performance comparing compute-intensive workloads different computer systems. SPEC95 consists suites benchmarks: CINT95* measuring comparing
Pentium® Processor Workstation Performance Brief
compute-intensive integer performance, CFP95* measuring comparing computeintensive floating-point performance. suites provide component-level benchmarks that measure performance computer's processor, memory architecture compiler. SPEC benchmarks selected from existing application benchmark source code running across multiple platforms. Each benchmark tested different platforms obtain fair performance results across competing hardware software systems. SPEC95 third major version SPEC benchmark suites, which 1989 became first widely accepted standard comparing compute-intensive performance across various architectures. release replaces SPEC92*, which been phased out. SPEC longer publishes SPEC92 results discontinued selling benchmark suite. Performance results from SPEC95 cannot compared those from SPEC92, since benchmarks have been added existing ones changed. CINT95 suite, written language, contains eight CPU-intensive integer benchmarks. measures calculates following metrics:
SPECint95 geometric mean eight normalized ratios (one each integer benchmark) when compiled with aggressive optimization each benchmark. SPECint_base95* geometric mean eight normalized ratios when compiled with conservative optimization each benchmark.
CFP95 suite, written FORTRAN* language, contains CPU-intensive floating-point benchmarks. measures calculates following metrics:
SPECfp95 geometric mean normalized ratios (one each floating-point benchmark) when compiled with aggressive optimization each benchmark. SPECfp_base95* geometric mean normalized ratios when compiled with conservative optimization each benchmark.
following figures, Intel Pentium processor performance compared with other non-Intel processors. Figures show SPECint95 SPECfp95 performance Pentium processors. following figures show, Pentium processorbased system from Intergraph Computer Systems delivers comparable performance with today's highest rated non-Intel processor-based systems.
SPEC result 10.00 12.00 14.00 0.00 9.77 12.30 9.45
9000 C160* Digital AlphaStation 500/400
8.38 9.82 8.36 9.78 8.04 8.04 7.22 7.22 6.90 7.72 5.58 6.26 4.55 4.72
2.00 10.40
4.00
6.00
8.00
Digital AlphaStation 500/333* Digital AlphaStation 5/333* Intergraph TDZ300/200* Intergraph TDZ300/180* Ultra Series 1200* Ultra model 170* RISC System/6000 43P-133* 9000 C110*
4.41 4.41 4.31
Digital AlphaStation 255/300*
Results Intel processors have been submitted SPEC publication; others obtained from SPEC website, August 1996.
Figure SPECint_base95* SPECint95* Benchmarks
5.23 4.00 INDY 4.10 R5000SC/180MHz*
9000 C100*
SPECint95*
SPECint_base95*
3.67 3.67
Pentium® Processor Workstation Performance Brief
Pentium® Processor Workstation Performance Brief
18.00 16.00 SPECfp_base95* 14.00 SPECfp95* 12.00 SPEC result 10.00 8.00 6.00 4.00
15.20
16.30 13.40 14.10 12.50 10.20 9.06 7.45 6.20 6.41 5.81
5.76 4.71 4.78 4.72
INDY R5000SC/180MHz*
12.60
12.80
11.80
9.61
8.45
7.45
6.20
5.82
5.38
5.25
4.73
Intergraph TDZ300/200*
Intergraph TDZ300/180*
RISC System/6000 300/3CT*
Results Intel processors have been submitted SPEC publication; others obtained from SPEC website, August 1996.
Figure SPECfp_base95* SPECfp95* Benchmarks
SPEC organization also publishes results throughput (rate) measurement. benchmark numbers that represent throughput known SPECint_rate95*, SPECfp_rate95*, SPECint_rate_base95* SPECfp_rate_base95*. With this method, called "homogeneous capacity method," several copies given benchmark executed. This method particularly suitable multiprocessor systems. results express many jobs particular type (characterized particular benchmark) executed given time. SPEC rates therefore characterize capacity system compute-intensive jobs similar characteristics. Because different units, values SPECint95/SPECfp95 SPECrate_int95/SPECrate_fp95 cannot compared directly. SPEC_rate results shown Intel Pentium processor-based system several today's leading non-Intel processor-based workstations.
RISC System/6000 43P-133*
SPARCstation Model 151*
Digital AlphaStation 5/333*
Digital AlphaStation 500/400*
Digital AlphaStation 500/333*
Digital AlphaStation 255/300*
9000 C160*
9000 C110*
9000 C100*
0.00
Ultra model 170*
3.59
2.00
3.76
Pentium® Processor Workstation Performance Brief
120.00 SPECint_rate_base95* 100.00 SPECint_rate95* 80.00 SPEC result
60.00
40.00 88.00 111.00 85.00 93.60 75.50 88.40 75.20 88.00 72.40 72.40 64.90 64.90 38.80 47.10 39.70 39.70
9000 C110*
20.00
Digital AlphaStation 500/333*
Digital AlphaStation 500/400*
Digital AlphaStation 5/333*
Digital AlphaStation 255/300*
0.00
9000 C160*
Results Intel processors have been submitted SPEC publication; others obtained from SPEC website, August 1996.
Figure SPECint_rate_base95* SPECint_rate95* Benchmarks
9000 C100*
Intergraph TDZ300/200*
Intergraph TDZ300/180*
33.00 33.00
Pentium® Processor Workstation Performance Brief
160.00 SPECfp_base_rate95* 140.00 SPECfp_rate95 120.00 SPEC result 100.00 80.00 60.00 147.00 120.00 127.00 113.00 67.10 40.00 136.00 20.00 0.00
57.70 55.80 52.30 113.00 115.00 106.00 67.10 52.40 55.80 48.40
47.20
Intergraph TDZ300/180*
Digital AlphaStation 5/333*
Digital AlphaStation 500/400*
Digital AlphaStation 500/333*
Results Intel processors have been submitted SPEC publication; others obtained from SPEC website, August 1996.
Figure SPECfp_base_rate95* SPECfp_rate95* Benchmarks
APPLICATION PERFORMANCE BENCHMARKS
this section, present Pentium processor performance Windows application-based benchmarks such MicroStation* Pro/E* MSC/NASTRAN*.
SYSmark* Windows
SYSmark* Windows suite application software associated benchmark scripts that have been developed Business Applications Performance Corporation (BAPCo) order provide tool accurate realistic measurement system performance personal computers running popular business-oriented applications Windows operating environment. scripts developed reflect usage patterns users business-oriented environment. Workloads SYSmark Windows were developed based BAPCo's standardized practice surveying users determine they exercise popular applications day-to-day work. applications selected testing able across three popular architectures (Intel, MIPS Alpha). SYSmark Windows generate performance metrics composite different applications specific application, such word processing spreadsheets.
Digital AlphaStation 255/300*
9000 C160*
9000 C110*
9000 C100*
Intergraph TDZ300/200*
51.80
Pentium® Processor Workstation Performance Brief
WORD PROCESSING
Microsoft Word* (native 32-bit architectures) Microsoft Excel* (native 32-bit architectures) Welcom Software Technology Texim Project* 2.0e (native 32-bit architectures) Orcad MaxEDA* (PCB design tool) (native 32-bit architectures) Microsoft PowerPoint* (16-bit Windows emulation)
SPREADSHEET
PROJECT MANAGEMENT
COMPUTER-AIDED DESIGN
PRESENTATION GRAPHICS
Figure illustrates SYSmark Windows Performance Intel Pentium processor.
Pentium® Processor Workstation Performance Brief
SYSmark result
SYSmark* Windows
AlphaStation 5/266*
Intergraph TD-400* (dual PPro MHz)
Vectra* 6/200 (dual PPro MHz)
Intergraph TDZ300 GLZ1T* (PPro 180)
Intergraph TDZ300 GLZ1T*-modified (PPro MHz)
Results systems obtained from BAPCo website June 1996.
Figure SYSmark* Windows Benchmark
LINPACK* 100x100 Benchmark
LINPACK* linear equation solver written FORTRAN. LINPACK program consists floating-point addition multiplication matrices. 100x100 LINPACK solves 100x100 matrix simultaneous linear equations. Source code changes allowed that results used evaluate compiler's ability optimize target system. LINPACK benchmark measures execution rate MFLOPS (millions floating-point operations second). When running, benchmark depends memory-bandwidth gives little weight I/O. Therefore, when LINPACK data fits into system cache, performance higher. Figure shows LINPACK performance Pentium processor-based systems other non-Intel processor-based systems.
AlphaStation 5/300*
Pentium® Processor Workstation Performance Brief
160.00 140.00 120.00 100.00 Mflops 80.00 60.00 40.00 156.30 153.00
9000 C160* 140.00
LINPACK* 100x100 double precision
133.60 67.00
Intergraph 62.00 TDZ300/180*
53.30
27.80
0.00
RISC System/6000 42W/T* 22.40
20.00
15.10
Digital AlphaStation 5/333*
Digital AlphaStation 500/333*
Digital AlphaStation 255/300*
RISC System/6000 300/3CT*
RISC System/6000 43P-133*
Results non-Intel processors obtained from each manufacturer's websites Jack Dongarra's report through June 1996.
Figure LINPACK* MFLOPS Performance
MicroStation* Performance Benchmark
This benchmark replaces CADbench* which fully-automated benchmark that with MicroStation*. benchmark contains tests, which nine files. basic operation consists copying each design file temporary directory before doing manipulations, recording times each test, then deleting newly-created design file. copying files included times that reported. tests grouped into three sets tests better test certain aspects system. Within Group three tests form heavy video test. Results yield very good representations power processor video subsystem. Within Group tests that disk subsystem test results. Group switches processor must good floating-point calculations. this test MicroStation Performance Benchmark seven times each different configuration with warm boot between each run. warm boot used reset operating system "states" same value prior each run. Results were then averaged standard deviation calculated. When possible, exact same combination drives loaded, memory use, MicroStation setup resolution monitors used. benchmark Pentium processor-based systems DEC* Alpha-based systems. chart shown this report contains only those systems containing Intel Pentium processor using memory compared with systems using Alpha processor running both MHz, also with memory. actual benchmark data article references other systems well, these were chosen represent highest available performance available Pentium processor
RISC System/6000 200/25S*
Intergraph TDZ300/200*
Pentium® Processor Workstation Performance Brief
consistent memory across platforms. Alpha system used memory. This particular system chosen since amount memory affect time result. Symbion Pro* Pro* used Intel Pentium processor. Figure shows results MicroStation benchmark from MicroStation Manager magazine comparing Pentium processor-based systems with non-Intel processor-based systems.
MicroStation* Performance Benchmark
Time Seconds
127.91
125.61
118.58
104.6
Pro* MiniTower Pentium Symbion Pro* Pentium DeskStation Raptor3* Alpha* Predator 4-300* Alpha* Symbion 164* Alpha*
Results from MicroStation* Manager Magazine, 1996
Figure MicroStation* Performance Benchmark.
MSC/NASTRAN*
MSC/NASTRAN*, principal product MacNeal Schwendler Corporation, industry's leading Finite Element Analysis (FEA) program. MSC/NASTRAN offers wide variety analysis types, including linear statics, normal modes, buckling, heat transfer, frequency response, transient response, random response, response spectrum analysis, aeroelasticity. Virtually material type modeled, including composites hyper elastic materials.
107.4
Pentium® Processor Workstation Performance Brief
Sparse matrix numerical methods, incorporated every analysis type, greatly increase solution speed reduce amount disk space, making numerical processing fast efficient. following benchmarks from MSC/NASTRAN suite demonstrate that version code, running Intel Pentium processor, provides type performance that users have come expect from technical workstations. MSC/NASTRAN Version 68.2 been ported Windows environment running Intel platforms. This released product results presented show that Intel platforms, running Windows represent good choice MSC/NASTRAN.
POWER TRAIN VIBRATIONAL MODEL
vibration analysis power train model with approximately 60,000 degrees freedom been variety platforms comparative reference. order find natural frequencies range MSC/NASTRAN performed decompositions solves, extracted modes. Figure shows MSC/NASTRAN performance Pentium processor-based systems what might considered typical workstation configuration with memory less. Figure shows MSC/NASTRAN performance larger memory configurations. These include Cray supercomputer workstations configured with least memory. This chart included reference only represent configurations that definitely higher cost than those focused this performance brief.
2500
MSC/NASTRAN* Power Train Vibrational Model 256MB
2000 Time Seconds
1050.8
969.3
1500
1012.0
1338.1
1000
Intergraph TDZ400* Celebris XL6200* NeTpower Calisto* RS6000/375*
Results non-Intel processors provided MacNeal Schwendler Corporation.
Figure MSC/NASTRAN* Power Train Vibrational Model Performance
1098.9
639.2
735.7
828.0
Pentium® Processor Workstation Performance Brief
1200.00 1000.00 Time Seconds
MSC/NASTRAN* Power Train Vibrational Model 256MB
120.00
146.8 800.00 600.00 400.00 360.00 913.00 641.2 184.00
119.5
200.00 0.00
CRAY*
9000/819*
Intergraph TDZ400*
Celebris XL6200*
Results non-Intel processors provided MacNeal Schwendler Corporation.
Figure MSC/NASTRAN* Power Train Vibrational Model Performance
BCELL*
BCELL* benchmark been used many past versions MSC/NASTRAN. This benchmark represents overall time seconds calculations involved. model BCELL benchmark cube solids with plate each surface each solid. tests represent ~10000 ~22000 degrees freedom. These models were chosen upper bound resource requirements. runs were done their facility except result Celebris XL6200* system. This done Workstation Technology Intel. Figures show performance Intel Pentium processor-based systems compared with non-Intel processor-based systems BCELL benchmark using ~10000 ~22000 degrees freedom.
733.5
Time (secs)
RS6000 590*
90.00
Celebris XL6200*
166.60 180.00 180.00 190.00 200.00 205.00
9000/810* R8000* 9000/770* RS6000 980/98B* Alpha* 3000/800 Alpha* 2100 4/275 9000/735* RS6000 580* R4400* Alpha 3000/500* Ultra Sparc* R4000*
BCELL* Degrees Freedom ~10000
210.00 210.00 250.00 300.00 310.00 400.00 420.00
Results non-Intel processors provided MacNeal Schwendler Corporation.
Figure MSC/NASTRAN* BCELL* Model Performance ~10000 Degrees Freedom
Pentium® Processor Workstation Performance Brief
Pentium® Processor Workstation Performance Brief
BCELL* Degrees Freedom ~10000
Time (secs) 166.60 180.00 180.00 190.00 200.00 205.00
Alpha* 2100 4/275 210.00
210.00 250.00 300.00 310.00 400.00
Ultra Sparc*
RS6000 590*
RS6000 580*
Alpha* 3000/800
Alpha 3000/500*
Celebris XL6200*
9000/810*
9000/770*
9000/735*
RS6000 980/98B*
R8000*
R4400*
Results non-Intel processors provided MacNeal Schwendler Corporation.
Figure MSC/NASTRAN* BCELL* Model Performance ~22000 Degrees Freedom
ANSYS*
ANSYS* most widely used Finite Element Analysis packages world today. From nonlinear linear, structural, thermal, piezoelectrics acoustics analysis optional modules electromagnetic field computational fluid dynamics studies, ANSYS program provides tools meet almost engineering need. ANSYS designed benchmark programs, known benchmarks, which have data that demonstrates well large ANSYS problems perform computer systems. data contains five static analysis examples using ANSYS frontal solver. These examples were deliberately with large wavefront stress computer systems. analysis examples represent cantilevered plate with element through thickness. mesh size varied different cases. force loading applied plate free end. ANSYS solid isoparametric element, SOLID45, used these static examples.
Disk Requirements (MB)
Name
Elements 1,000 2,000 3,000 4,000 6,000
Nodes 2,222 4,422 6,622 8,822 12,642
Active Degrees Freedom 6,060 12,060 18,060 24,060 36,120
Maximum Wavefront 1,218 1,818 2,418 1,818
Wavefront 573.3 1,131.9 1,690.5 2,249.0 1,754.1
R4000*
420.00
90.00
Pentium® Processor Workstation Performance Brief
Following results obtained four commercially available Pentium processor workstations. These machines come with variety memory architectures, disk drives number processors. reported Intel results from ANSYS Version 5.3, released July 1996.
Time Calisto* Celebris* TDZ-400* Vectra* 2500 40.749 38.075 34.141 38.906 257.941 235.208 207.609 247.625 865.795 900.945 709.391 861.25 2035.537 1975.661 1512.125 1866.125 1767.101 1570.158 1403.813 1659.672
Time
2000
Seconds
1500
Calisto* Celebris* Intergraph TDZ-400*
1000
Vectra*
Figure ANSYS* Performance
Comparative results elapsed time less reliable. current routine reporting elapsed time ANSYS reports even seconds. Further, elapsed time numbers, because reliance ANSYS reading writing scratch files, depend upon structure fullness file system plus other system network activity. fact, consecutive runs same machine show deviations elapsed time several percent.
Time Calisto* Celebris* TDZ-400* Vectra* 1022 2267 2205 1814 2113 2119 1877 1790 1983
Pentium® Processor Workstation Performance Brief
2500
Elapsed Time
2000
Time (secs)
Calisto*
1500
Celebris*
Intergraph TDZ-400*
1000
Vectra*
Figure ANSYS* Performance
Following chart ANSYS performance compared with several other available workstations. Intel processor-based machine Intergraph TDZ400* Pentium processor-based system with memory. results shown chart compared results other non-Intel processor-based systems running Release 5.2. ANSYS published numbers Release large scale benchmark. These numbers reference only.
Pentium® Processor Workstation Performance Brief
3500 ANSYS* Large Scale Benchmarks 3000 2500 Time (secs) 2000 1500 1000
Alpha* Server 2100 4/275
Alpha* Station 5/333
Alpha* Station 500/333
UltraSPARC 170*
ANSYS* Version results non-Intel based systems obtained from ANSYS* website dated June 1996.
Figure ANSYS* 5.3/Intergraph ANSYS* 5.2/non-Intel systems
Parametric Technology Corporation Pro/ENGINEER*
Parametric Technology Corporation's Pro/ENGINEER* suite integrated software products that automate mechanical design-through-manufacturing process. This unique, fully-associative suite mechanical design automation software includes application-specific products that address complete spectrum product-development activities. Within Pro/ENGINEER family, more than modules address every aspect product development: design, drafting, manufacturing. addition, specialized tools meet your requirements advanced surfacing, assembly management, design optimization, cable pipe routing, design, data management, more. Since applications fully associative, engineering teams work parallel develop product designs their manufacturing processes concurrently. Systems Group Texas Instruments developed benchmark Parametric Technology Corporation's Pro/ENGINEER. This benchmark been made publicly available Pro/E: Magazine. results their benchmarking been published three issues date, most recently March/April 1996 edition. results published were obtained using Release Pro/ENGINEER. machine tested with beta version Release Texas Instruments group developed Pro/ENGINEER benchmark test overall system performance complex assembly models assembly drawings. Texas Instruments engineers modeled real parts assemblies create accurate manufacturing design work environment. models represented wide range sizes complexities, while benchmark tested many advanced capabilities Pro/ENGINEER such holes, chamfers, rounds, attachment hardware. goal stress single task, test cross section tasks operations that make true work environment. result, benchmark measured tasks including those only performed occasionally such plotting
Intergraph TDZ-400* 200MHz
Power Indigo2 R8000*
715/100 128MB
712/100* 128MB
J210
Pentium® Processor Workstation Performance Brief
PostScript* files creating exploded views, well more frequently executed tasks such model spinning, panning zooming, rotation. benchmark's developers determined operations well relative weighting those operations observing engineers' usage patterns over period several weeks. Evaluation output files known "trail files" determination weighting factors. These trail files contain actual commands chosen during time design engineer using Pro/ENGINEER. test exercises graphics capabilities well computational capabilities platform being tested. also contains some occasionally performed tasks, such generating shaded color PostScript files measuring mass properties. Weighting factors were determined represent often user might these commands single eight-hour day. Adequate memory used each machine ensure there enough avoid excessive paging disk. requirement this benchmark results presented here Release Pro/ENGINEER. result beta release Release while NeTpower result production release Pro/ENGINEER Release Release Pro/ENGINEER scheduled deployment later 1996 preliminary results Pentium processor-based platforms running Windows demonstrating improved performance. Texas Instruments Pro/E benchmark lower weighted score better.
Pentium® Processor Workstation Performance Brief
Pro/Engineer*
Weighted Total Time
Indigo2* Solid IMPACT NeTpower Calisto* 200MHz Pentium
result from Pro/E: Magazine January/February 1996. NeTpower result generated NeTpower April
Figure Pro/ENGINEER* Performance
Simulators
benchmark report from Solutions provides detailed comparisons execution speeds memory requirements variety simulators. intent benchmark, which focuses performance rather than ease-of-use issues, help users choose simulators while minimizing amount evaluation needed. benchmark exercise included VHDL simulators from Cadence, Ikos, Mentor Graphics, Synopsis Veda Design Automation, along with Verilog simulators from Fintronic, Mentor Graphics Intergraph. encompassed some benchmark circuits ranging 3200 gates gate level 48,000 lines code register-transfer level (RTL). This benchmark exercise carried over nine-month period. While independent evaluators from Solutions tests, vendors provided information about best options switches various circuits, vendor representatives monitored some tests. following charts lower score better performance.
Pentium® Processor Workstation Performance Brief
Raynet-gate TI-Mult TI-Mult with unit timing Total Elapsed Time Raynet-gate
Sparc UltraSparc 167* Pentium® Alpha*
Benchmark results available from Solutions Limited Gale Moor Avenue Gosport PO12 2SJ, Results appearing here from EETimes, June 1996 edition.
Figure Raynet-gate TI-Mult
Pentium® Processor Workstation Performance Brief
CPU-Mix DMult Dmult with unit timing Total Elapsed Time CPU-Mix with unit timing
Sparc
UltraSparc 167* Pentium Alpha*
Benchmark results available from Solutions Limited Gale Moor Avenue Gosport PO12 2SJ, Results appearing here from EETimes, June 1996 edition
Figure CPU-Mix Dmult
GRAPHICS PERFORMANCE SUMMARY
Viewperf*
Viewperf* benchmark which measures rendering performance systems running under OpenGL* which written maintained OpenGL Performance Characterization (OPC) group. organization began 1993 ad-hoc project group aimed establishing graphics performance benchmarks systems running under OpenGL application programming interface (API). group joined committee summer 1994. Viewperf measures mixture graphics primitives, drawn with different mixtures primitive attributes. Results independent tests then combined into weighted average create composite score test. Because this, almost impossible project Viewperf score accelerator, must measured. Improving aspect performance card will cause score increase. doubling performance triangles won't necessarily translate doubling performance Viewperf scores. Adding advanced features accelerator like anti-aliasing hardware support texture acceleration, other hand, could cause score double even shaded triangle rate held constant. Since applications actually these advanced features, this still represents growth performance application increase Viewperf metric correct indicator. project group worked with Independent Software Vendors (ISVs) obtain tests, data sets weights that constitute what called Viewset. Each Viewset represents
Pentium® Processor Workstation Performance Brief
graphics rendering portion actual application. Currently, Viewperf5* comprises five standard viewsets: Parametric Technology's CDRS*, which contains seven different Viewperf tests, modeling rendering application computer-aided industrial design. IBM's Data Explorer* (DX), which different tests, visualization application. Intergraph's Design Review*, which different tests, computer model review package. Alias/Wavefront's Advanced Visualizer*, with tests, animation application. Lightscape Technology's Lightscape Visualization System*, with tests, radiosity visualization application.
five Viewsets represent relatively high-end applications. These types applications typically render large data sets. They almost always include lighting, smooth shading, blending, line anti-aliasing, z-buffering, some texture mapping. ISVs that develop Viewsets have provided percentage weights each test which performance number reported. ISVs have defined these percentages indicate relative importance test within overall application. Viewperf offers following characteristics:
provides single-source code apples-to-apples comparison performance tuning across different hardware platforms. runs multiple operating systems, including OS/2, UNIX Windows runs across different processors, including Alpha, Intel, MIPS PowerPC. runs multiple windowing environments, including Presentation Manager*, Windows Windows. encompasses wide variety OpenGL features rendering techniques. easily accessible through project subcommittee, through OpenGL sample disk distribution.
Several factors make Viewperf unique from other benchmarks:
uses datasets that designed used real applications. uses rendering parameters models selected ISVs graphics users. produces numbers based frames second, measurement with which users readily identify. provides number each rendering path using data set.
Viewperf measures performance following entities: primitives, including points, lines, line strip, line loop, triangles, triangle strip, triangle fan, quads polygons. Attributes vertex, primitive frame. Lighting Texture mapping Alpha blending Fogging Anti-aliasing Depth buffering
Pentium® Processor Workstation Performance Brief
Viewperf CDRS* Viewset
CDRS Parametric Technology's modeling rendering software Computer-Aided Industrial Design (CAID). used create concept models automobile exteriors interiors, other vehicles, consumer electronics, appliances other products that have challenging free-form shapes. users CDRS typically creative designers with titles such automotive designer, products designer industrial designer. There seven tests specified that represent different types operations performed within CDRS. Five tests triangle strip data from lawnmower model created using CDRS. other tests show representation lawnmower wireframe. tests weighted show following CDRS functionality:
Table Summary CDRS* Tests1
Test Weight CDRS Functionality Represented Vectors used designing model. Represents most design work done CDRS. Anti-aliasing turned allow designer cleaner version model. Surfaces shown polygons, with single surface color. Surfaces grouped with different colors group. Textures added groups polygons. Texture used evaluate surface quality. Color added vertex show curvature surface. Same Test Number without anti-aliasing.
Note results obtained from SPEC/GPC/OPC page containing results Viewperf5. Results obtained before August 1996.
Pentium® Processor Workstation Performance Brief
35.00
30.00
CDRS-03* Viewset
25.00 Mean Composite Score
20.00
15.00
10.00
RISC System/6000 Model 42T* 12.05 POWER GXT500*
RISC System/6000 Model 43P* 4.26 POWER GXT250P* RISC System/6000 Model 43P* 4.20 POWER GXT150P*
5.00 32.09 27.88
RISC System/6000 Model 3BT* 23.52 POWER GXT1000* Model RISC System/6000 Model 43P* 23.47 POWER GXT1000* Model
23.24
13.91
13.11
11.58
AccelGraphics Pro/T* (200 MHz) (HP) 11.03
9.90
9.48
7.90
4.28
4.05
3.61
RISC System/6000 Model 43P* 3.57 POWER GXT250P*
Intergraph TDZ310 MHz* RealiZm Z13-T*
Indigo2* R4400 Solid IMPACT*
Dell Dimension XPS* 200n V192*
PC360 PentiumPro Intergraph Intense 3D-T*
Dell Dimension XPS* 200n Oxygen 102*
Intergraph TDZ310* RealiZm Z13-GT*
Indigo2* R4400 High IMPACT*
PC360 PentiumPro Matrox Millenium* bit)
PC360 PentiumPro Matrox Millenium* bit)
results obtained from SPEC/OPC page containing results Viewperf5*. Results obtained August 1996. Those appearing have been submitted inclusion.
Figure CDRS* Composite Scores
Viewperf Design Review* Viewset
Design Review* computer model review package specifically tailored plant design models consisting piping, equipment structural elements such I-beams, HVAC ducting, electrical raceways. allows flexible viewing manipulation model helping design team visually track progress, identify interference, locate components, facilitate project approvals presenting clear presentations that technical non-technical audiences understand. construction site, Design Review display construction status sequencing through vivid graphics that complement blueprints. After construction complete, Design Review continues valuable tool planning retrofits maintenance. Design Review multi-threaded application that available both UNIX Windows model this Viewset subset plant model made GYDA offshore production platform located North southwest coast Norway. special thanks goes British Petroleum, which given subcommittee permission geometric data sample data this Viewset. this data restricted this Viewset.
PC360 PentiumPro Matrox Millenium* bit)
0.00
RISC System/6000 Model 43P* 3.21 POWER GXT150P*
Intergraph TDZ300 Pentium GLZ1T*
Intergraph TDZ300 Pentium GLZ1
Pentium® Processor Workstation Performance Brief
Design Review works from memory-resident representation model that composed high-order objects such pipes, elbows valves, I-beams. During plant walk-through, each view rendered transforming these high-order objects triangle strips line strips. Tolerancing each object done dynamically only triangles that front facing generated. This apparent Viewset model rotated. Most Design Review models greater than stored high-order objects. this reason, benefit dynamic tolerancing face culling, display lists used. There tests specified Viewset that represent most common operations performed Design Review. These tests follows:
Table Summary Design Review* Tests1
Test Weight Functionality Represented Walk-through rendering curved surfaces. Each curved object rendered triangle mesh, depth-buffered, smooth-shaded, with light different color object. Walk-through rendering flat surfaces. This treated different test than Number because normals sent facet flat shade model used. Walk-throughs require many objects clipped. This same Number except clipping forced. Again, clipping test Number easily spot rendered objects within complex model, objects identified rendered solid rest view rendered wireframe (line strips). line strips depth-buffered, flat-shaded, unlit. Colors sent primitive. additional help visual identification location objects, model have "screen door" transparency applied. This adds polygon stippling test Number other views present screen help user select model orientation. These views display position orientation viewer. wireframe, orthographic projection model used. Depth buffering used, multithreading cannot used order preserve draw order. more realism, objects model textured. Decal texturing with linear blending mipmaps used. Same test Number except clipping forced. Same test Number except clipping forced.
Note
results obtained from SPEC/GPC/OPC page containing results Viewperf5. Results obtained before August 1996.
Pentium® Processor Workstation Performance Brief
6.00
5.00
Mean Composite Score
4.00
DRV-04 Viewset
3.00
2.00
1.00
0.85 0.71 0.70 0.68 0.65
RISC System/6000 Model 43P* POWER GXT250P* RISC System/6000 Model 43P* 100MHz POWER GXT250P* PC360 Matrox Millenium* bit) PC360 Matrox Millenium* bit) Dell Dimension 200n V192*
0.63 0.59
RISC System/6000 Model 43P* POWER GXT150P* RISC System/6000 Model 43P* 100MHz POWER GXT150P* PC360 Matrox Millenium* bit)
5.14
4.52
4.41
4.13
3.34
3.32
3.30
3.29
2.85
2.56
2.39
RISC System/6000 1.35 Model 42T* POWER GXT500*
0.93
RISC System/6000 Model 43P* POWER GXT1000* Model
Indigo2* R4400 Solid IMPACT*
Indigo2 IMPACT 10000* Solid IMPACT*
Intergraph TDZ310* RealiZm Z13-GT*
Intergraph TDZ410* RealiZm Z13-T*
Intergraph TDZ410* RealiZm Z25-GT*
Intergraph TDZ310* RealiZm Z13-T*
0.00
Intergraph TDZ400* GLZ1T*
results obtained from SPEC/OPC page containing results Viewperf5*. Results obtained August 1996. Those appearing have been submitted inclusion.
Figure Design Review* Composite Scores
Viewperf Data Explorer* Viewset
Visualization Data Explorer* (DX) general-purpose software package scientific data visualization analysis. employs data-flow driven client-server execution model currently available UNIX workstations from Silicon Graphics, IBM, Sun, Hewlett-Packard Digital Equipment. OpenGL port Data Explorer completed with recent release 2.1. tests visualize particle traces through vector flow field. width each tube represents magnitude velocity vector that location. Data such this might result from simulations fluid flow through constriction. object represented contains about 1,000 triangle meshes containing approximately vertices each. This medium-sized data tests assume z-buffering with light addition specification color every vertex. Triangle meshes primary primitives this Viewset. While Data Explorer allows many other modes interaction, these assumptions cover majority user interaction.
Dell Dimension XPS* 200n Oxygen 102*
RISC System/6000 Model 3BT* POWER GXT1000* Model
Indigo2* R4400 High IMPACT*
PC360 Pentium 200MHz Intergraph Intense 3D-T*
0.45
Mean Composite Score
Test
Indigo2* R4400 7.33 Solid IMPACT*
Weight
2.5%
2.5%
Indigo2* R4400 7.32 High IMPACT* RISC System/6000 Model 3BT* POWER GXT1000* Model
RISC System/6000 Model 43P* POWER 5.88 GXT1000* Model
Intergraph TDZ310* RealiZm Z13-G*
TMESH's clipped
LINE's display listed
Intergraph TDZ310* 3.24 RealiZm Z13* Intergraph TDZ300* GLZ1*
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
6.49
5.75
TMESH's display listed
LINE's immediate mode
3.15 3.12
POINT's immediate mode
TMESH's immediate mode
TMESH's with polygon stippling
TMESH's list with facet normals
TMESH's with sided lighting
Functionality Represented
PC360 PentiumPro Intergraph Intense 3D-T*
POINT's direct rendering display listed
RISC System/6000 Model 42T* POWER GXT500*2.88 Dell Dimension XPS* 200n V192*
2.44 2.37
Table Summary Data Explorer* Tests1
Note results obtained from SPEC/GPC/OPC page containing results Viewperf5*. Results obtained before August 1996.
results obtained from SPEC/OPC page containing results Viewperf5*. Results obtained August 1996. Those appearing have been submitted inclusion. Dell Dimension XPS* 200n Oxygen 102*
Pentium® Processor Workstation Performance Brief
Figure Data Explorer* Composite Scores
DX-03 Viewset
RISC System/6000 Model 43P* POWER 1.49 GXT250P* PC360 PentiumPro 1.27 Matrox Millenium bit) PC360 PentiumPro Matrox Millenium* 1.26 bit) PC360 PentiumPro Matrox Millenium* 1.22 bit) RISC System/6000 Model 43P* POWER 1.15 GXT250P* RISC System/6000 Model 43P* POWER 0.90 GXT150P* RISC System/6000 Model 43P* POWER 0.73 GXT150P*
Pentium® Processor Workstation Performance Brief
Advanced Visualizer* Viewset (AWadvs-01)
Advanced Visualizer* from Alias/Wavefront integrated workstation-based animation system that offers comprehensive tools modeling, animation, rendering, image composition, video output. Advanced Visualizer provides:
Geometric, analysis, motion data importation from wide range CAD, dynamics structural systems. Automatic object simplification switching tools working with ultra-large production data sets. Motion unlimited number object, cameras, lights with Wavefront SmartCurve* editing technique. Interactive test rendering high-quality, free-form surface rendering. Software rotoscoping matching computer animation with live action background footage. Realistic imaging effects, including soft shadows, reflection, refraction, textures, displacement maps, using Wavefront's fast hybrid scanline/raytracing renderer. Interactive image layering output video with Wavefront's Recording Composer*. Powerful scripting customization tools, open file formats, user-defined interfaces.
operations within Advanced Visualizer performed immediate mode with doublebuffered windows. There four basic modes operation within Advanced Visualizer: material shading (textured, z-buffered, backface-culled, local lights)
perspective, trilinear mipmapped, modulated (41.8%) perspective, nearest, modulated (10.45%) ortho, trilinear mipmapped, modulated (2.2%)
ortho, nearest, modulated (0.55%) wireframe z-buffering, lighting)
perspective (28.5%)
ortho (1.5%) smooth shading (z-buffered, backface-culled, local lights)
perspective (9.5%)
ortho (0.5%) flat shading (z-buffered, backface-culled, local lights)
perspective (4.75%) ortho (0.25%)
Pentium® Processor Workstation Performance Brief
These tests specified Viewset that represent most common operations performed Advanced Visualizer. tests follows:
Table Summary Advanced Visualizer* Tests1
Test Weight 41.8% 28.5% 10.45% 9.5% 4.75% 2.2% 1.5% .55% .25% Advanced Visualizer* Functionality Represented Material shading polygonal animation model with highest interactive image fidelity perspective projection Wireframe rendering polygonal animation model with perspective projection Material shading polygonal animation model with lowest interactive image fidelity perspective projection Smooth shading polygonal animation model with perspective projection Flat shading polygonal animation model with perspective projection Material shading polygonal animation model with highest interactive image fidelity orthogonal projection Wireframe rendering polygonal animation model with orthogonal projection Material shading polygonal animation model with lowest interactive image fidelity orthogonal projection Smooth shading polygonal animation model with orthogonal projection Flat shading polygonal animation model with orthogonal projection
Note results obtained from SPEC/GPC/OPC page containing results Viewperf5*. Results obtained before August 1996.
12.00
10.00
Mean Composite Score
AWadvs-01 Viewset
8.00
6.00
4.00
2.47 1.52
RISC System/6000 Model 42T* POWER GXT500* PC360* Matrox Millenium* bit)
2.00 10.57 8.67
RISC System/6000 3BT* POWER GXT1000* 6.34 Model RISC System/6000 43P* POWER GXT1000* 6.29 Model (133 MHz)
Dell Dimension XPS* 1.51 200n V192*
PC360* Matrox Millenium* 1.45 bit)
PC360* Matrox Millenium* 1.42 bit)
4.01
Intergraph TDZ310* 4.01 RealiZm Z13-T*
Intergraph TDZ300* 3.91 GLZ1T* (200 MHz)
Indigo2* R4400 High IMPACT* (200 MHz)
Intergraph TDZ310* RealiZm Z13-GT*
results obtained from SPEC/OPC page containing results Viewperf5*. Results obtained August 1996. Those appearing have been submitted inclusion.
Figure AWadvs-01* Composite Numbers
PC360* Pentium Intergraph Intense 3D-T*
0.00
Dell Dimension XPS* 2.53 200n Oxygen 102*
Pentium® Processor Workstation Performance Brief
Lightscape* Viewset (Light-01)
Lightscape Visualization System* from Lightscape Technologies, Inc. represents generation computer graphics technology that combines proprietary radiosity algorithms with physically-based lighting interface.
LIGHTING
most significant feature Lightscape ability accurately simulate global illumination effects. system contains integrated visualization components. primary component utilizes progressive radiosity techniques generates view-independent simulations diffuse light propagation within environment. Subtle significant effects captured, including indirect illumination, soft shadows, color bleeding between surfaces. post process using tracing techniques adds specular highlights, reflection, transparency effects specific views radiosity solution.
INTERACTIVITY
Most rendering programs calculate shading surfaces time image generated Lightscape's radiosity component precalculates diffuse energy distribution environment stores lighting distribution part model. resulting lighting "mesh" then rapidly displayed. Using OpenGL display routines, Lightscape takes full advantage advanced graphics capabilities Pentium processor-based systems with OpenGL-compliant graphic acceleration boards. Lightscape allows interactively move through fully simulated environments.
PROGRESSIVE REFINEMENT
Lightscape utilizes progressive refinement radiosity algorithm that produces useful visual results almost immediately upon processing. quality visualization improves process continues. this way, user total control over quality (vs. time) desired perform given task. point solution process, users alter characteristic light source surface material system will rapidly compensate display results without need restarting solution. This flexibility performance allow users rapidly test various lighting material combinations obtain precisely visual effect desired. There four tests specified Viewset that represent most common operations performed Lightscape Visualization System. four tests follows:
Table Summary Lightscape* Tests1
Test
Note
Weight
Lightscape* Functionality Represented Walk-through wireframe rendering "Cornell Box" model using line loops with colors supplied vertex Full-screen walk-through solid rendering "Cornell Box" model using smooth-shaded z-buffered quads with colors supplied vertex Walk-through wireframe rendering 750K-quad Parliament Building model using line loops with colors supplied vertex Full-screen walk-through solid rendering 750K-quad Parliament Building model using smooth-shaded z-buffered quads with colors supplied vertex
results obtained from SPEC/GPC/OPC page containing results Viewperf5*. Results obtained before August 1996.
Pentium® Processor Workstation Performance Brief
0.90 0.80 Light-01 Viewset 0.70 Mean Composite Score 0.60 0.50 0.40 0.30 0.20 0.87 0.61 0.52 0.50 0.48 0.47 0.32 0.10 0.00
Indigo2* R4400 Solid IMPACT* (250 MHz)
0.31
RISC System/6000 Model 42T* POWER GXT500*
results obtained from SPEC/OPC page containing results Viewperf5*. Results obtained August 1996. Those appearing have been submitted inclusion.
Figure Light-01* Composite Scores
SUMMARY
Pentium processor family offers superior performance high-end entry-level workstations. Pentium processor systems listed this performance brief priced under $25,000. performance results show that these systems compare very favorably higher-priced workstations from non-Intel architecture systems. With SPECint95 rating 8.71 SPECfp95 rating 6.68, Pentium processor fastest industry. even higher performance, Pentium processor offers 4-way glueless multiprocessing. Superior performance, coupled with availability advanced operating systems peripherals, makes Pentium processor ideal platform most demanding workstation applications. particular, Pentium processor performs extremely well CAD, scientific modeling analysis, other computationally intensive tasks. With availability workstation applications such MSC/NASTRAN, MicroStation Pro/ENGINEER Intel Architecture, Pentium processor platform choice mechanical users performance graphics related benchmarks demonstrates that Pentium processor provides excellent cost/performance choice digital content creation market well. outstanding integer performance comprehensive library tools available also make Pentium processor ideal platform software development environment.
Dell Dimension XPS* 200n Oxygen 102*
Intergraph TDZ310* RealiZm Z13-G*
Intergraph TDZ310* RealiZm Z13*
PC360 Pentium Intergraph Intense 3D-T*
RISC System/6000 Model 3BT* POWER GXT1000* Model
RISC System/6000 Model 43P* POWER GXT1000* Model (133 MHz)
Pentium® Processor Workstation Performance Brief
APPENDIX TEST CONFIGURATIONS
Note
Specific Non-Intel system configuration information obtained from various website URL's referenced and/or other references noted. Celebris XL6200* Vectra* XU/200 Intel 82450 Orion Windows Workstation 3.51 (Build 1057) Adaptec 7880* SCSI Seagate ST32550N* 49.4 (burst) 4.17 AccelGraphics 3Dpro* 1024x768x15 Matrox Millennium* 1280x1024x24 Elsa* 1152x800x15 3DLabs Racer/TX* 1152x800x15
Pentium Processor Number processors Memory Primary Cache Secondary Cache Motherboard Chipset, Stepping Operating System
Intel 82450 Orion Windows Workstation 3.51 (Build 1057)
Hard Disk Controller Hard Disk Internal transfer rate (Mbit/sec) External transfer rate (MB/sec) (R/W) Seek (msec) Latency (msec) Graphics accelerator Tested Resolution (pixels)
SDMS SCSI Seagate ST15150N* 47.5 (burst) 4.17 AccelGraphics AG/300* 1280x1024x16
Intergraph TDZ-400* Pentium Processor Number processors Memory Primary Cache Secondary Cache Motherboard Chipset, Stepping Operating System
NeTpower Calisto* Intel 82450 Orion Windows Workstation 3.51 (Build 1057)
Intel 82450 Orion Windows Workstation 3.51 (Build 1057)
Hard Disk Controller Hard Disk Internal transfer rate (Mbit/sec) External transfer rate (MB/sec) (R/W) Seek (msec) Latency (msec) Graphics accelerator Tested Resolution (pixels)
Adaptec 7850* SCSI Seagate ST31250* 49.4 (burst) 4.17 GLZ2* 1280x1024x24
Adaptec 2940* SCSI Seagate ST32550N* 49.4 (burst) 4.17 AccelGraphics AG/300* 1280x1024x16 Elite 1152x800x15
Pentium® Processor Workstation Performance Brief
NeTpower Calisto* Pentium Processor Number processors Memory Primary Cache Secondary Cache Motherboard Chipset, Stepping Operating System Swap Space Hard Disk Controller Hard Disk Internal transfer rate (Mbit/sec) External transfer rate (MB/sec) (R/W) Seek (msec) Latency (msec) Pro/E release build Graphics accelerator Tested Resolution (pixels)
Intel 82450 Orion Windows Workstation 3.51 (Build 1057) 400MB Adaptec 2940* SCSI Seagate ST32550N* 49.4 (burst) 4.17 16/9610 Elite2* 1024x768x256
following lists configurations used testing cards running Viewperf tests: Intergraph Intense 3D-T* runs were 360* Pentium processor memory disk system configuration Dynamic Pictures* runs was: Dell Pro/200* Pentium processor memory disk AccelGraphics AccelPro Vectra* XU/200 Pentium processor memory disk
Pentium® Processor Workstation Performance Brief
Website References
This list shows URL's where data available that contained this document. also shows date last update available webpage. Individual vendor websites: (July 1996) (June 1996) (July 1996) (July 1996) (April 1996) (June 1996) (June 1996) SPEC results available (July 1996) BAPCo SYSmark results available (July 1996) Viewperf results available newsletter following URL: (June 1996) LINPACK results available ANSYS Large Scale Benchmark results available http://www.ansys.com/ (June 1996) Follow product pointer reach benchmark results. MicroStation Benchmark results: Taken from MicroStation Manager magazine, 1996 issue. Configuration information available this article well test results reported here.

September 1996 Order Number: 242999-001 Pentium® Processor Workst

Top Searches for this datasheet

Other recent searches