SPONSORED LINKS
On Mon, 20 Oct 1997, Paul Khoury wrote: > Sounds like the PEP system. I wish that there was a project > for a suite of free benchmarks for all OSes which could be standard, > though. Benchmarks themselves are somewhat synthetic approximations of true performance. Short of running the actual code you care most about the best they can do is give you a rough approximation of performance. The bulk of the market seems to care most about how well common Windows apps perform, so benchmarks like Winbench use large chunks of actual application code and time common activities like text scrolling and database sort and junk like that. While a benchmark like that can give me a rough idea of comparative performace between systems it doesn't tell me much about how my own personal "most used apps" will perform. These common Windows benchmarks give CPU's like the Cyrix 6x86MX very high performance numbers. This is all well and good for people that use those applications, but for the apps I use the 6x86MX is one of the slowest CPU's [on a rated Pxx spec, or actual clock per clock basis] I could choose. I care most about game performance, and anything above a P90 or so is fine for the apps I use. That's why I use the types of software I run to generate my own numbers that reflect my own needs in a system. ============================================================================ ==== Steve Parker ==== San Luis Obispo, CA ==== Multi-OS & Multitasking ==== ============================================================================