Upgrade Results...

[HaMMerHeD]

So, I got my parts in and installed them. Before I took the old kit apart though, I ran 3dmark03, 05, and 06, and I ran it again as soon as I got the new stuff put together, and windows installed onto it. I know 3dmark isn't a popular benchmarking utility on the [H], but it's all I had, so bugger off if you don't like it.

Anyways, here are the old specs, as they relate to 3dmark performance:
AMD Athlon64 3000+ (Newcastle, S754, 2.0ghz, 512KB L2 cache)
Asus K8V Deluxe (VIA K8T800)
MSI nBox GeForce FX 5900 Ultra (256mb AGP)
1GB Corsair XMS DDR400 (2x512mb twinx)

That rig, untweaked, most everything set to default, got the following scores:
3dmark03: 5885
3dmark05: 1237
3dmark06: 387

Here is the new configuration, as it relates to 3dmark performance:
AMD Athlon64 3700+ (San Diego, 2.2ghz 1mb L2 cache)
Asus A8N-SLI
BFGTech GeForce 7900GT (256mb PCIe)
3gb Corsair XMS DDR400 (2x1gb twinx, 2x512mb twinx)

The new rig, untweaked, on a fresh install, got the following scores:
3dmark03: 16521
3dmark05: 8029
3dmark06: 3994


Not a bad improvement, if I do say so myself. The system does seem to boot quite a bit faster. I finds that somewhat odd, since I am using the same hard hard drive to boot from as I was before (74gb Raptor).

Anyways, for those of you considering an upgrade, these are the results I got.

 

[st4rk]

It would be interesting to see how it performs with the newcastle and the 7900gt.

 

[Volkswagen]

the reason ur new system scores so much more is because of that 7900GT I am sure that the processor is a little faster but the 7900GT is what makes your 3dmark scores......

 

[HaMMerHeD]

The 3dmark scores are certainly due to the gpu, but there are a number of other observable speed increases that have more to do with the processor. After all, going from a 2.0ghz socket 754 newcastle to a 2.2ghz socket 939 san diego is more than just the 200mhz (221 actually) increase. The newcastle does not have a dual channel memory controller. The san diego does. The newcastle has a 1600mhz Hypertransport bus. The san diego has a 2000mhz hypertransport bus. The newcastle has 512kb of L2 cache memory, and the san diego has 1mb of L2 cache. The newcastle is based on a 130nm wafer, and the san diego is a 90nm part. While the manufacturing process is not necessarily indicative of better performance, it is generally true that an updated core is more efficient, and therefore performs better.

While the old-school CLOCK-IS-EVERYTHING people might consider a 200mhz upgrade not worthwhile, there are other differences between the 2 processors that affect performance.