[tmpilling]

This benchmarking tool "http://www.xbench.com is awesome! I ran it on my stock PB G4 1.67 (last release) with 512RAM (I know, I know) and I scored 48.06. I wasn't terribly overwhelmed, but compared to the average Al PB results on their site, it was alright. I then ran http://www.macupdate.com/info.php/id/11582 Onyx and it jumped up about 5%. I only ran the complete optimisation script and the maintenance script.

Was just wondering what other people jump in benchmark test might be after these operations? Or just your general benchmark score?

Care to share

[mduell]

Xbench is poo and gives wildly inconsistent results.

[MichiganRich]

How about Geekbench? I wish their results site was more helpful, as opposed to just listing all the results by time submitted. I'm always curious about being able to compare *ahem* Apples to Apples...

What's a good benchmark tool? Anyone?

[ghporter]

mduell, how did I know you'd post that?

I have used Xbench only because I wanted to have SOME measure of the difference between 1GB of RAM and 2GB when I upgraded my MBP, and frankly I'm cheap sometimes. I can "feel" a significant difference, but Xbench only gave me a few points higher in any of the tests. It did not, obviously, stress the system, like running XP under Parallels while doing something interesting (beyond surfing) in OS X. I used to wait a LONG time for Parallels to load XP-not anymore. So Mark, I agree that Xbench is poo. Any "decent" benchmarks that are free?

[tmpilling]

Well, disappointing to hear that people aren't huge fans of Xbench. I found it interesting. I'm interested to see what happens when I jam another stick of RAM into the bay. Considering that the results seem to be from a large number of systems and mine was close to the average for my system configuration I would have thought that it was somewhat accurate?

[ghporter]

A datum derived from a large number of data points is only worthwhile if it actually says something-and provides repeatable results. I just reran the whole suite and compared it to my first scores after upgrading from 1GB to 2GB of RAM. I got a composite score of 79.53 after the upgrade and 82.13 this morning. Interestingly, I got a composite score of 82.39 with my original configuration... So I got LOWER scores after adding RAM, but not the SAME lower scores on two different runs. This says that, as a measurement tool, Xbench is not very precise or accurate.

As I said earlier, it was free and there when I wanted to measure something, but it really didn't tell me a whole lot-go ahead and look at your scores after adding RAM and see how the details change. It's pretty to watch the video tests, but Xbench really disappoints me by no providing anything that would tell me what was really happening.

[mduell]

Originally Posted by ghporter 

mduell, how did I know you'd post that?

I have used Xbench only because I wanted to have SOME measure of the difference between 1GB of RAM and 2GB when I upgraded my MBP, and frankly I'm cheap sometimes. I can "feel" a significant difference, but Xbench only gave me a few points higher in any of the tests. It did not, obviously, stress the system, like running XP under Parallels while doing something interesting (beyond surfing) in OS X. I used to wait a LONG time for Parallels to load XP-not anymore. So Mark, I agree that Xbench is poo. Any "decent" benchmarks that are free?

Instead of going to the trouble of running Xbench, you could have just asked me and gotten a more accurate/meaningful result. Before your system scored a 5 and now it scores a 7. Congratulations on the upgrade!

If you want to benchmark your shiny new graphics card, go play your favorite game.
If you want to benchmark your shiny new hard drive, copy a big file from a ramdisk or to /dev/null.
If you want to benchmark your shiny new CPU, run whatever it is that you do that is CPU bound.
If you want to benchmark your shiny new RAM, look at the number of page ins/outs after a days work.