A BOBMark is the ratio of your chip's FPU speed to a "standard" FPU speed-- an AMD K6-2 300. If your chip has a BOBMark rating of 1.5 for example, it means that your chip is 1.5x faster than an AMD K6-2 at FPU operations. Likewise, if your chip has a BOBMark rating of .75, it means that your chip is .75x slower than an AMD K6-2 300.

But even if your don't care about an AMD 300, the tests are still relevant. For example: Take a Celeron 400 overclocked at 6.0 x 83 and 6.0 x 75. The BOBMark scores for each configuration are 2.94, and 2.64, respectively. If you divide the higher by the lower, you get about 1.109, which means the 6.0 x 83 configuration is about 11% faster than the 6.0 x 75 configuration.

The most important thing to note about the BOBMark is that the results are linear. Most other benchmarks (I won't mention any names) calculate results using combinations of different benchmarks, and then dump them into strange formulas that skew the data.

What does "BOBMark" stand for?

"Bob's semi-Objective BenchMark". I call it semi-objective because it only tests FPU speed and a limited set of the 386 instruction set.

What exactly is the algorhithm used by BOBMark?

BOBMark performs 8 different FPU operations on 4 random 32bit floating point values-- 900 million times. It then times how long it takes for your CPU to perform these operations, and takes the ratio of the your time to the time it took my AMD K6-2 300 to perform the same operations.

Why does BOBMark only test FPU speed?

My personal belief is that FPU's will always be with us, atleast for a couple more decades. :-) x86 engines, etc will eventually be deprecated. FPU's are also what really control the speed of number crunching programs, graphics programs, and video games, and are a little bit more telling about a chip's true performance. AMD has recently surpassed Intel in the speed of their x86 operations, (beating them at their own game-- I love it!) but AMD still lags behind in raw FPU speed. Benchmarks that test x86 speed can be very misleading.

Why an AMD chip as the base case?

Convience. The computer I used to write BOBMark was an AMD based system, so it made sense to compare the computer to itself while I was testing. :-) I also think that there's a little bit too much public perception that Intel makes superior chips, and I'd like to help change that. "Back in the day" Intel was a superior chip, but that's quickly changing..

Can you post your source code or make it available somehow?

This is a tricky question. I have no problem posting the source code, but I'm afraid that people will recompile the source on another platform.. which is what they will most likely do. :-) The problem with recompiling it on another platform is that there's no guarantee that the code generated is going to be the same. Other compilers may insert certain optimizations or timing features that would make the results useless. I'm sorry, but for now, the answer is no. :-(

I ran BOBMark on my computer and the results came out lower/higher than whats on your page?!

First: Did you remember to run BOBMark in DOS? Second: All chips are different. For this reason, the results that appear on the main page are averages of all of the submissions I received for the particular chip. If yours is different, submit it!

I ran BOBMark on my computer and it rebooted/locked/crashed/etc in the middle of the benchmark?!

If you're not the type that regularly pushes the envelope on your computer it could be possible that your chip overheated while performing the tests. This actually happened to me on my AMD 233 that was poorly heatsinked.

I saw on your page that if I overclock my chip I'll get a higher BOBMark. Should I overclock?

To overclock or not to overclock-- that is the question. :-) Let me just say this: If you have to ask this question you probably shouldn't. Overclocking is a potentially hazardous (for your chip) thing to do and should be left to individuals that know what they're doing. I, of course, can not be held responsible for anything bad that happens as a result of your overclocking.

Any plans for the future?