|
|
 |
| AXION TECHNOLOGIES - REVIEWS |
| |
| Benchmarks-should i upgrade? |
| |
| By Raymond griffith |
| Tuesday,May 21,2002 |
| |
Upgrading your system or even building a new system is not a simple matter of picking the cheapest or most expensive parts. There are plenty of hardware review sites around that provide a staggering amount of information on all the latest and greatest hardware. Certainly worth your time to read and compare before spending your cash on components.
Most people want to read bench mark numbers, and graphs. as they are very easily understood. They are easily digestible, and offer a quick means of comparison. Some good sites for comparison are:
[H]ardOCP
Tom's Hardware
AnandTech
FiringSquad
Ace's Hardware
Many have Forums for posting questions, feedback and an occasional flame war, but are a good source for help as well. If you don't know or don't understand register and post in a forum you feel comfortable with.
Now onto benchmarks,
Benchmarks and their scores have become increasingly complex. There are benchmarking utilities for just about every type of system, and even sub-system specific tests widely available (as shareware). The majority of users are familiar enough with the likes of WinBench, Winstone, and Mad Onions offerings. Other popular utilities used like Sandra which is not strictly a benchmarking software offers many useful modules.
There are a couple things you should keep in mind regarding any bench mark. Most benching packages are Synthetic tests. Meaning they simulate conditions a system could encounter, and measure how fast. Efficiently a given computer handles the execution of code under those circumstances. These types of tests are useful when you wish to compare the performance in this simulation between two or more systems. As the conditions from software point of view are meant to be identical, leaving (in theory) only the hardware as the difference, and thus result should be an accurate comparison between hardware.
Of course there are many issues that can make a difference in scoring that are not considered. Certainly the OS should be the same as well as drivers being the most recent etc. This can become a quagmire of issues though.
Ever hear the saying "All News is bias"? Well, even the most impartial tester can become guilty of this. The manufacturers are certainly not immune. nVidia released a "new" detonator driver about the time ATI was making the Radeon 8500 available for reviewers that definitely increased performance over the drivers that were available at the time(story here). ATI even released a set of Drivers for their Radeon series that contained specific tweaks and a patch for QIII to improve scores under testing(another story). These are just 2 recent events.
I'm not knocking benchmarks, as they provide a great method for testing while minimizing variables. They also allow testing for stability, and comparisons between driver revisions etc. the issue is that they offer a glimpse as to what a system can do, and offer a ballpark measurement.
By far the most widely reported scores are obtained with Mad Onions 3D Mark software. And they have also released a wider based utility for testing the entire system called PC Mark 2002. Both are very easy to obtain, and to install/run (grab them here, and here). One popular aspect is the online comparisons, where you may opt to compare your system against other enthusiast systems, and compare scores/hardware.
Don't be frustrated when some gal/guy with the same system (or even slower clocked) completely demolishes your score! As I said earlier the bench marks are not the end all be all of a system. So you spent more money on components, and racked your brain to set up the killer rig only to be out scored by someone with the word H4x0r* in their signature, and their water cooled P4 running @ 3+GHz and 3-5k higher score!
*is not direct reference to anyone or group- whether you know what H4x0r means or not. And I am sure no animals were harmed in the benchmarking of their system(s)! *fingers crossed
The only thing that is important is if you are satisfied with the performance of your system. Do your programs/Apps run in a time frame you find acceptable? Is your system stable? If so, then the numbers generated by benchmarking means nothing other than bragging rights. There will always be a new video card or faster CPU just around the proverbial corner that can out perform what you have currently.
Now if your system does not perform in a satisfactory level, then you may want to look at how to best increase the performance. The most expensive is not always the best choice, and matching your components to each other is essential. I encourage everyone to visit some of the links, and try out some of shareware; if nothing else there is some great eye candy to behold.
Please take a look at the FAQ's on the sites, and check out the help files/documentation included with each software package. If you are unfamiliar with what some tests are actually reporting, most have a glossary to help explain terms used and so forth. I plan on continuing this article to cover (in general) some of the points that should be considered when building or upgrading, and possibly some tweaks to get more out of your system.
as always if you have a question give us a Call in Tech support @ 281-575-0004.
Raymond Griffith
^
|
|
|
|
|
|
