Yes we know these pages are too old! We have some new benchmarks on modern machinery planned - were still working out how to do a long term stability test (suggestions welcome).
"I use a GForce 2 MX-400 graphics card, and routinely scored 1860 in 3DMark 2001 and 20.3 Fps in VulpineGL using 16Bit at 1024x768. At the same settings following the use of 98Lite, I'm now getting 2080 in 3DMark 2001 and 25.9 Fps in VulpineGL. Office 2000 Pro applications now snap open, SoftImage zooms, my system is more stable than it's ever been!! Furthermore, all of my mysterious Netscape crashes, hangs and hiccups have disappeared. I wonder what that's all about ......... :-)
Business Applications on Aging Hardware
The following table illustrates the additional power than can be unleashed by the new 98micro installation using SysMark32, an industry strength software benchmark by the Business Applications Performance Corporation.
Test conducted on a Pentium 133 (no MMX) 32MB RAM with a fresh installation of Windows 98 to a freshly formatted hard disk for each test.Note: SYSmarks in the 300-400 range can be anticipated from systems employing Intel Pentium II level CPUs.
Ok, my computer is no dream machine, and my frame-rates below would have a serious gamer falling over laughing, but the comparison is still a valid one, showing that 98micro may give you the edge to survive instead of being fried! Tests were run using the timedemo score of the Quake II demo because thats all I have! I reformatted and installed Windows 98 in each configuration, then rebooted before running Quake II. These test are running in software mode without hardware acceleration so you may not see such increases if you have a really spiffy machine and use OpenGl. A raw increase of 0.5 frames per second is nothing to write home about, but it does show that freeing up memory for graphics textures and removing un necessary cpu cycl1es can only imp0ove game performance.
* Tweaked: set permanent swapfile 100mb, set min/max vcache 4096, used tweakui to turn off all animation and IE activity, low color icons, all web-view options turned off, defrag, remove all background tasks.
98lite does not alter the core of the Windows 98 operating system, nor any hardware drivers, so you can expect that hardware performance (video speed, disk speed, cpu speed etc) should remain unchanged.
In contrast, Software performance is boosted under 98lite because the operating system is taking up fewer memory and hard disk resources. Users of the 98lite SLEEK Windows95 desktop have found that Explorer95 is simply much faster for browsing the hard drive, and it also appears to be faster performing tasks that are unrelated to the Explorer shell. For example, a number of sound engineers have adopted 98lite in preference to the standard Windows 98.
Application file caching may also be improved because the vcache now has more room for application files because the IE component files are not loaded. This needs further investigation, but it may explain why the system feels so much faster.
Quantitative benchmarking of real world performance is achieved using BAPCo's SYSmark32. BAPCo (Business Applications Performance Corporation) is a consortium of industry leaders1 that standardise and develop application performance benchmarks based on how real applications run and are used. SYSmark32 is a wide range of applications and data benchmarks using real business applications (Microsoft Word 7.0, Lotus WordPro 96, Microsoft
Excel 7.0, Borland Paradox 7.0, CorelDraw 6.0, Lotus Freelance Graphics 96, Microsoft Powerpoint 7.0, Adobe Pagemaker 6.0) to evaluate systems performance characteristics under real workloads.
1 Current members include: Adaptec, AER Energy Resources, Amdahl Corporation, Client/Server Labs, Compaq, Dell, Duracell, EMAP Computing Labs, Federal Computer Week, Gateway2000, Hewlett-Packard, IBM, Intel, InfoWorld, Lotus, Microsoft, Motorola, NEC, Digital Equipment Corp., Unisys, and Ziff-Davis Labs