Overclocking Your P4 800FSB - Results
(Page 4 of 5 )
I'm sure that most of you would like to see what comes out of all of this work. Well, here are my system specs, and what I managed to get out of it. Depending on your hardware, you are going to get different results. But I'll toss in a few different levels, that are accomplishable with a lot of hardware out there. The only thing exotic about my system in particular is my water cooling setup, but that does not come into play here, since my motherboard is the limiting factor.
- Intel 2.4C 800FSB processor
- Soltek 86SPE-L "Springdale" motherboard
- Infineon PC3200 memory (2x512MB)
- ATi Radeon 8500LE 128MB video card
- 2x60GB IBM 120GXP
- Silicon Image Sil3112 PCI SATA RAID
- Enermax 430W PSU
- #Rotor copper water block
- Hayden 676 transmission oil cooler
- Swiftech MCP 300 pump
Now, for today's test suite
- PCMark2002
- 3DMark2001
- Comanche 4
- AquaMark2.3
- Unreal Tournament 2003
- Divx Encoding
Okay, first up to bat,
FutureMark's synthetic system benchmark, PCMark2002.
As you can see, the CPU speed scales linearly, as expected, since this is a synthetic (ie. theoretical) benchmark. The memory scores also increase across the board, even though at 275MHz FSB while using the 3:2 divider leaves the memory speed below that of the other two (200MHz).
Second on the list is
3DMark2001.
FutureMark originally planned this to be a video card benchmark, it is very system dependant as well. Here's what it has to say:
Again, we are seeing a fairly linear increase. Going from the stock 200FSB to 275 earns us an extra 1300 marks. Not too shabby, considering the video card didn't change whatsoever.
Comanche 4 isn't the greatest game out there. However, it has a nicely repeatable benchmark included in the demo. Running this at 640x480 to reduce the dependency on the video card, the average FPS should again scale with the CPU.
Not quite linear, but again showing an increase. This game is more bound by my crappy video card than anything else. Still, you can't complain about getting another 13FPS for free.
This is the previous version of
Aquamark, 2.3. The newer version, 3.0, doesn't show very good results for me, thanks to the lack of a DX9 based video card. However, this one is showing similar results to the above. Games, especially older ones, are not always going to be video card limited; increasing your system speed will get you more FPS, especially at lower resolutions.
Not much to say for these either. After making use of the
HardOCP benchmark tool for the Flyby's, and
UT2K3's own benchmark tool for the Botmatch, you can see the results for yourself. They mirror all of the ones achieved already.
Okay, last bench of the day. Since I use my computer more for this than anything else, it's an especially relevant benchmark for me. The sequence used was a conversion of the 3rd period of Game 8 from the '72 Summit Series. Any true hockey fan knows of Canada's victory over the USSR in this fantastic hockey exhibition, and Paul Henderson's famous winning goal. I merely timed how long it took to get to 25,000 frames, and calculated the frames per second average. Just like all the other benchmarks, this one was done three times at each speed, and that average is the one quoted in the graph.
As long as you have sufficient memory and a fast disk subsystem, this benchmark is completely CPU dependant. Overclocking allowed me to get better performance with my 2.4C than I would have achieved with Intel's previously most expensive processor, the Pentium 4 3.2GHz.
Next: Conclusion >>
More Hardware Guides Articles
More By DMOS
/ 68 
