I enjoy benchmarking my PC and overclocking it because it actually allows me to get more out of my money and enjoy the games I like to play without having to fork thousands or be satisfied with mediocre performance.
Originally that is what it was all about for me with my old computer, but in my effort to get the most bang for my buck I think I selected a GPU from a maker that wasn't rated as well as the same GPU from other makers. That aside, this logic is perfect if you are trying to grab a model from a few classes below the top of the line and still want it to perform like a leader of the pack. I am sure if in the first place I had done more research on the exact model of my old GFX card, I would have selected an analog from another maker and been just fine with it and tweaking it would have gave me more positive effects than negative effects.
Seeing that you like to tweak stuff, you might be one with a solid answer to this...
I have noticed through testing many different pieces of hardware (not benching but literally real world usage testing) that cards that come stock overclocked or superclocked seem to fail more than stock cards that users tweaks themselves. Have you had similar experiences?
I'd say the paid version of 3DMark is worth it, I like it, but it's been about ~2+ years since I've used Heaven Benchmark, so can't really comment much on it -- would definitely say try the free version first.
I think 3d mark is getting put on my tab tbh. The way the result data is displayed and easily available is a gigantic plus one and having a database of other results to compare it against also puts another good aspect to my point of view of things. I expect by next weekend to make my purchases for my testing purposes.
On another note... I tried out heaven's Benchmark and REALLY wonder why these people make demos instead of games...seriously, I actually found myself sitting there thinking "
dude I can jump on this windmill blade and get to the roof and lay down with the sniper" and then found me reminding myself saying "
no Ant you can't play COD on this map ever, sorry".
I tested it against the DirectX 11 and OpenGL APIs with maxed out settings for everything and it looked stupendously awesome on both. Best of all my system didn't fail.
I also like how their free trial doesn't feel like it has any limitations though the paid versions appear to add a lot of useful functionality for people who want to take the testing to another level
(batch testing seems pretty damn useful if you ask me and so does the report generation for people who like to crunch numbers)
I got so caught up exploring the environment and playing with the settings that I literally didn't click on the benchmark function until the 3rd or 4th time I started it up the title.
<--doh doh dohhh Needless to say, without installing a full game it gave a very good idea of what my system can do as far as video is concerned before it bogs down or starts tearing or whatever other problems come with bad configurations.
If your after banchmarks, the 3dmark is the only real tool. For burning in, starcraft II will realy hammer your CPU in a big multiplayer game, and metro 2033 will bring any PC to its knees on max graphics settings.
Still haven't run any games yet on this...I know for a fact that SCII is on my list of to do's but I don't think I have ever tried or even heard of for that matter metro2033
(don't quote me on that though...I forget about loads of crap). When I get to the point where I am running games on the rig I may throw that in the list as well.
) but hey for every blown up computer there is a new one to be built 
