Archives
 
 
 
  Special
 
 
 
  About Us
 
 
 

Newsletter
Free E-mail Newsletter from BYTE.com

 
    
           
Visit the home page Browse the four-year online archive Download platform-neutral CPU/FPU benchmarks Find information for advertisers, authors, vendors, subscribers Request free information on products written about or advertised in BYTE Submit a press release, or scan recent announcements Talk with BYTE's staff and readers about products and technologies

ArticlesHow the Power Macs Measure Up


April 1994 / Cover Story / How the Power Macs Measure Up
Tom Thompson

While Apple's Power Macs represent a tremendous technical achievement in providing compatibility with the existing software base, the question everyone will ask is: How fast are they? Answering this question required running the BYTE low-level and application benchmarks (which test the 680x0 emulator), the BYTE native benchmarks (which test native performance), and two Windows applications (which check the 80x86 emulation). These results were on preliminary systems, and Apple plans a system software release this summer that will make more of the Toolbox code native.

Power Mac performance ranges from that of a IIci (which uses a 20-MHz 68030) to better than that of a Quadra 840AV (which uses a 40-MHz 68040), depending on how much native Toolbox code the application uses during the cours e of operations. The application benchmark results are skewed downward, because Excel and Mathcad, upon detecting the lack of an FPU (because the emulator reports that it's a 68020), did their own computations. This kept execution in the emulator, impairing performance. The word processing test, which makes heavy use of native QuickDraw code, shows better results: The low-end 6100/60 outgunned a Quadra 950 (a 33-MHz 68040 system), and the 8100/80 did better than the Quadra 840AV.

Native results showed that the 6100/60 and 7100/66 trailed the RISC systems BYTE tested (see "Windows on RISC" on page 109), but this was because they lacked the level 2 cache that most of the other RISC systems possessed. (An optional level 2 cache SIMM is available for these Power Macs.) The 8100/80, with a 256-KB level 2 cache as standard equipment, fared much better, posting an overall index close to that of the Sun SparcStation 10.

Windows emulation, supplied by Insignia Solutions, was on par with other RISC system s--slow. Excel 4.0 and Word for Windows 2.0 benchmarks indicate that even lacking the level 2 cache, the 6100/60 and 7100/66 were in the middle of the RISC pack on word processing tasks. The 6100/60 trailed the pack on spreadsheet operations, while the 7100/66 did as good as the DEC Alpha, and the 8100/80 was on par with the 200-MHz Carrera Cobra. Considering that the prices of the RISC pack range from $4000 for a basic system up to $25,000, the Power Macs do an impressive job.

Again, remember that performance is a moving target here. The Power Macs will get faster in the future as Mac applications are ported to native code, which eliminates the overhead of the emulator. Also, Apple plans to keep moving more of the Toolbox to native code, and these changes will be distributed periodically with system software releases.


PowerMac vs other RISC systems on Windows emulation



The Power Macs compare favorably with other RISC systems on the Windows emulation. Native performance wa
s lower, but the 6100/60 and 7100/66 lacked a level 2 cache, as did the IBM PowerStation 250 and HP 900 Series 700. The native benchmarks were compiled using a beta version of the MPW PowerPC compiler. On all results, a higher number indicates better performance.


MAC LOW-LEVEL AND APPLICATION INDEXES


                   QUADRA    POWER MAC   POWER MAC   POWER MAC
LOW LEVEL           840AV      6100/60     7100/66     8100/80


CPU                  7.11         2.96        3.34        4.08
FPU                 57.64        26.22       29.98       36.32
Disk                 2.85         2.51        2.53        3.28
Video                7.62         4.48        5.86        7.29
APPLICATION
Word Processing      2.61         2.25        2.45        3.68
DTP                  4.50         2.11        2.28        3.32
Database             6.28         3.84        4.21        5.98
Development          4.25         2.83        3.09        3.58
Graphics             4.57         2.50        2.6
3        3.37
Scientific          14.25         6.29        6.62        9.20
Spreadsheet         13.23         2.36        2.62        3.78
Overall             49.69        22.18       23.91       32.89


SPREADSHEET AND WORD PROCESSING PERFORMANCE INDEXES UNDER EMULATION


                               SPREADSHEET        WORD PROCESSING
IBM PowerStation 250 (Wabi)           0.14        0.14
DeskStation Tyne v4633v (Wabi)        0.21        0.21
Sun SparcStation 10 (SunPC)           0.21        0.25
Sun SparcStation 10 (Wabi)            0.08        0.19
HP 900 Series 700 (SoftWindows)       0.21        0.28
HP 900 Series 700                     0.21        0.47
DEC Alpha 200 AXP                     0.14        0.32
NEC RiscServer                        0.24        0.47
Carrera Cobra                         0.18        0.29
PowerMac 6100/60 (SoftWindows)        0.12        0.34
PowerMac 7100/66 (SoftWindows)        0.14        0.31
PowerMac 8100/80 (SoftWindows)        0.18        0.4
5


NATIVE PERFORMANCE INDEXES
Carrera Cobra                   3.61
DeskStation Tyne v4633v         2.88
NEC RiscServer                  2.73
IBM PowerStation 250            2.57
HP 900 Series 700               2.36
DEC Alpha                       2.22
Sun SparcStation 10             1.96
PowerMac 8100/80                1.90
PowerMac 7100/66                1.50
PowerMac 6100/60                1.37


Up to the Cover Story section contentsGo to previous article: Apple, IBM Bring PowerPC To The DesktopGo to next article: Open Firmware Provides CPU IndependenceSearchSend a comment on this articleSubscribe to BYTE or BYTE on CD-ROM  
Flexible C++
Matthew Wilson
My approach to software engineering is far more pragmatic than it is theoretical--and no language better exemplifies this than C++.

more...

BYTE Digest

BYTE Digest editors every month analyze and evaluate the best articles from Information Week, EE Times, Dr. Dobb's Journal, Network Computing, Sys Admin, and dozens of other CMP publications—bringing you critical news and information about wireless communication, computer security, software development, embedded systems, and more!

Find out more

BYTE.com Store

BYTE CD-ROM
NOW, on one CD-ROM, you can instantly access more than 8 years of BYTE.
 
The Best of BYTE Volume 1: Programming Languages
The Best of BYTE
Volume 1: Programming Languages
In this issue of Best of BYTE, we bring together some of the leading programming language designers and implementors...

Copyright © 2005 CMP Media LLC, Privacy Policy, Your California Privacy rights, Terms of Service
Site comments: webmaster@byte.com
SDMG Web Sites: BYTE.com, C/C++ Users Journal, Dr. Dobb's Journal, MSDN Magazine, New Architect, SD Expo, SD Magazine, Sys Admin, The Perl Journal, UnixReview.com, Windows Developer Network