Final Benchmarking Results

[madad2005]

Ok, here are the benchmarks from my old and now updated system. I was expecting a bit of a better result to be honest, but I have a feeling my power supply isn't handling the new graphics card (unlocked) too well; even though it is rated at 480W it only has 17A on the 12V rail which could be slightly overloading it. It could also be the CPU itself which is holding it back. If anyone knows better, please let me know. Anyway, here are my initial and current specs:

Initial:

sparkle 6600 256MB graphics card
P4 s478 3Ghz @ 3.1Ghz, 207Mhz
1GB (2x512MB) Elixir 2700 3-4-4-8
DiamondMax 9+ 120GB IDE ATA100
480W ThermalTake Butterfly PSU

Current:

xfx 6800xt 128MB graphics card (this thing is huge! Almost a foot long!!)
P4 s478 3Ghz @ 3.1Ghz, 207Mhz
2GB (2x1GB) Kingmax 3200 3-4-4-8
Soundblaster Audigy 2 ZS 7.1
DiamondMax 9+ 120GB IDE ATA100
480W ThermalTake Butterfly PSU

Results-

6600:

Aquamark3: 33000; 3DMark03: ~5400; 3DMark05: ~1700

6800xt - stock settings (8ps,4vp@325/700):

Aquamark3: 45375; 3DMark03: 7387; 3DMark05: 2785

6800xt - 12ps,4vp (unlocked +4ps):

Aquamark3: 50104; 3DMark03: 8779; 3DMark05: 3138

6800xt - 16ps,4vp (unlocked +8ps):

Aquamark3: 52029; 3DMark03: 9417; 3DMark05: 3357

6800xt - 16ps,5vp (unlocked +8ps,+1vs):

Aquamark3: 53139; 3DMark03: 9689; 3DMark05: 3602

6800xt - 16ps,5vp@355/768 (unlocked and overclocked):

Aquamark3: 53750

I turned my settings back to the stock 325/700 after that test as there wasn't much performance improvement. One of the vertex shaders was damaged, which is why this card wasn't selling as a 6800. The improvement in gaming performance is quite noticeable: I can play Guild Wars at 1280x1024 with maximum detail settings and maximum antialiasing and the frame rates are at around 50-90. However, I have still to test it out in a proper battle scene. This should set me up nicely for the next year and for Oblivion (when it desides to come out!).

[madad2005]

Sorry.. done it again!

[saphalline]

As fun as it is to run 3DMark, the performance difference between these synthetic benchmarks and actual games is increasing. 3DMark is no longer an accurate measure for real-world gaming performance, as you can see in your numbers. The amount of performance increase you'll see with double the pipes and with a slight overclock will be much more profound than 3DMark can show you. Especially at the higher resolutions with AA & AS turned on, the extra MHz will help a lot.

[madad2005]

Yeah, that is what I've found so far. I can play Guild Wars at 1280x1024 with 4x AA and details settings at max and my fps are between 35-70. This, compared to my 6600 performance, is excellent!

I finally applied my Zalman vf700-cu to my 6800 and I saw a temperature drop of 13C (from 60-63 idle to 47-51 idle). Pretty impressive and it is much quieter too (18db), although the GPU was a mess since whatever built the card put thermal grease everywhere - took a while to clean it up. The new cooler allowed me to get a maximum overclock on my card of 392Mhz (GPU) and 785Mhz (RAM) which gave a 3DMark05 score of 4289 and a 3DMark03 score of 11290. However, I've put my settings back down to the stock since I don't need a card that fast yet.

Anyway, I am very pleased with my purchase.