Forum Wars Journal Part 2: “Safe” Scores

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

With stock scores in the rear-view mirror and the first weekend of Forum Wars coming to a close, most of the safe runs have been done. Well, I say safe, what this usually means is just slow enough to keep things from making a mess all over the room with hot electronics. Igniting various fabrics and hairs is a possibility as well, if you don’t know how to insulate or spend more than $20 on a power supply.

Generally the best thing to do at this point in the wars is to get a feel for your hardware. It’s a bit tough to get this done with two rigs in a single weekend, but it’s pretty easy with one rig. I started out running the 3D benchmarks on rig #1 (Q6600 and GTX 260, for those of you who skimmed over that bit in the first article). Since I wasn’t doing anything to cool down the air like opening windows or putting the computer outside, I decided to stick with around 3.7GHz (420×9) which is only slightly higher than my daily speed of 3.6 (400×9) and at the same voltage of 1.5V. The goal here is to get a feel of my GTX 260, which I’ve yet to do any benching on, and 3dmark 03 and 05.

I started out with 05, making scores of around 22K. Then I started working the GPU clocks up to what would eventually be roughly 750/1500/1100. After some tweaking, the final score I got was 24323. I then moved on to 03, which was giving me some problems mainly because I quickly got used to running my 4870 with it. The GTX 260 takes a huge hit in game test 4, and it just drives me nuts every time seeing the frame rate that low. I believe my final scores were around 55-60K, but I was apparently so flustered that I decided to swap to rig #2 at the end.

I forgot to do 2-D benches, might I add. Remember what I said about mistakes? There you go.

Moving on to rig #2, it was time for something to break. Luckily, it wasn’t anything serious. The nVidia and ATI drivers on Vista decided to not get along and after a few hours of trying to get drivers for the 4870 installed, I just gave up and went to 2-D. I’ll have to re-install the Vista partition sometime this week and do all of my rig #2 3-D benchmarks next weekend, maybe sooner.

I don’t have much experience with this E3200, so I decided to boot at 350×12 at 1.5V to see if it would run (and what clock speed it was for that matter. I don’t like math without a calculator, alright?) Turns out it was 4.2GHz, and quite stable. I then turned my attention to memory speeds, settling on a 2:3 divider, taking my RAM up to 1050MHz, or 525MHz actual speed. The first SuperPi runs were around 16s thanks to the E3200’s astounding and complete lack of L2 cache. I mean seriously, even if it is a Celeron, give it at least 2M! After some tweaking and taking the FSB up to 360, I ended up with a time around 14.5s. Some later testing would reveal that this chip has a nasty FSB wall somewhere right above 360mhz, so 360×12 (4.32GHz) is probably the highest this is going to go before things get chilly. I then moved on to WPrime, but it appears that somehow my screenshot of it got overwritten. How? Stupid screenshot program I’m using. It’s better than MS Paint though. That aside, this chip is scary stable right up until the FSB wall, in fact those WPrime screenshots got overwritten by SuperPi 32M screenshots. Too bad we don’t use that in Forum Wars any more.

Now, some of you clever people out there might be going “You forgot to run Geekbench, you idiot!” Well, for those people, I did run it on my E3200. Here is the proof, a single run at 4.2GHz. Why only one? Well to put it in ways that won’t keep this article from getting published, I (and most everyone else I’ve talked to on the benching team) don’t really like Geekbench. I just find a stable speed, run it, and be done with it. The less time you spend running Geekbench, the more time you can tweak the rest of your scores. If anyone was in chat watching moocow or Sam__ bench this weekend, then you know that they have many other names for Geekbench, and out of all of them there are a grand total of zero I can post here without getting in trouble.

That’s basically all of the work I did this weekend. Although I managed to not have scores for a handful of benches, I do know where most of my hardware maxes out now, so I know exactly where to start when things get cold next week. If I get really productive I’ll fill in all of the holes in my safe scores before next weekend, otherwise I’m gonna have to break out the caffeinated substances. Things really start going downhill when you bring caffeine into the mix, I had a friend one time that… Oh look at the word count. See you guys next article!


Loading new replies...