Need help, w/ building my computer.

madmag9999 · Post by **madmag9999** » Wed Apr 07, 2004 11:21 am

o i though what u were saying was that toms hardware and sites like that lie on the benchmakrs they get not that the programs suck. guess ill just compare encodeing speed and game smoothness

dwchang · Post by **dwchang** » Wed Apr 07, 2004 11:38 am

madmag9999 wrote:o i though what u were saying was that toms hardware and sites like that lie on the benchmakrs they get not that the programs suck. guess ill just compare encodeing speed and game smoothness

Well yes Tom's Hardware is an Intel fanboy, but this is more demonstrated in his editorials and what he chooses to report. He uses the same (flawed) benchmarks tools as everyone else, but obviously emphasizes certain things and so on.

What I was getting at was that Intel physically OWNS one of the major benchmarks and has a committee made up of all Intel people who decide what is run. That seems hardly fair at all and yet many people rely on such results. It makes sense if you didn't know it was owned by Intel.

I also wanna warn you that when benchmarking, you need to have as few variables as possible. In your case, you may have two computers with diff. chips, but do they have the identical memory? HD? Graphics Card? etc.? These are all things that can affect performance and benchmarks TRY to keep things as fair and uniform as possible. That way you know what is creating the performance differences.

madmag9999 · Post by **madmag9999** » Wed Apr 07, 2004 1:18 pm

well the systems are not idenitcal but i can make them close to identical but i guess even that wont give me a true benchmark. so there seems to be no way to aquier true benchmarks becouse even if everything in the system is the same the motherboard is still diff so that can throw off the benchmarks cant it

dwchang · Post by **dwchang** » Wed Apr 07, 2004 7:51 pm

madmag9999 wrote:well the systems are not idenitcal but i can make them close to identical but i guess even that wont give me a true benchmark. so there seems to be no way to aquier true benchmarks becouse even if everything in the system is the same the motherboard is still diff so that can throw off the benchmarks cant it

Let's just say the words "true" and "benchmark" don't belong together. We can have pretty dern close representations, but there is no such thing as a 100% accurate benchmark.

The best I can do is run simulations on my design files, but that's obviously only one side. I guess we could simulate code on the two designs here at work, but obviously this is very unrealistic to a consumer and more importantly, it's a simulation. The only reason it might have some validity is b/c the "systems" would be the same...i.e. non-existant.

Kalium · Post by **Kalium** » Wed Apr 07, 2004 10:11 pm

There are three kinds of lies: lies, damn lies, and benchmarks.