*sigh* my turn to be concerned with '03 score!

[squidman]

Howdy!
Yeah is this good, considering the config below? Im surprised, that i get less than 2k m/texels in multi-texture test. On other 3d marks (2000, and 2001SE) i get above 2k. Odd.
'03 score
2001SE score

Thanks!

 

[selfbuilt]

Your links just take people to their own Futuremark scores, if they've registered. You'll have to post your results directly.

 

[vss1980]

Better to post the compare links given with each project you have on the ORB.

As for the scores, in 3DMark2001 and 2003 I get near identical scores for the fill-rate of my card (Radeon 9000 Pro). In 3DMark2000 it scores higher... I'm pretty sure the multi-texture fill-rate test in 3DMark2001 and higher is more stressful than the previous versions as I recall reading that it uses more layers but can't rightly recall.

 

[squidman]

I got =4200 in '03, and =14000 in '01.

 

[policy11]

um..wtf? you're worried about a 4200 3dmark2k3 score?

 

[Harabecw]

03 also gives a lower multi fillrate count than 01 for me. don't sweat it.

 

[selfbuilt]

Your scores are reasonable. I notice, by the way, that you seem to be o/cing a 2600+ 333FSB processor to 400Mhz (200*11?), but are only running PC2700 (333Mhz) ram. If you had to lower the memory timings to run it sync at 400, or if are running it async at 333, you are going to take a major performance hit that is offseting most of your CPU performance gain. You would be better off running the CPU at 333MHz, at a higher clock multiple (e.g. 13 or 13.5) running sync with ram. That will help your 3DMark 2001 scores and actual game performance more than the 400FSB with underperforming ram (I've run the stats both ways, so I know).

 

[MercenaryForHire]

/brings out the Big Stick of no0blAr +2000

2001 is not 2003.

/whack whack whack

Do not directly compare benchmark results.

/whack whack whack



- M4H

 

[Harabecw]

I want my 40000000000000000 3dmark2000 marks.

 

[squidman]

Originally posted by: selfbuilt
Your scores are reasonable. I notice, by the way, that you seem to be o/cing a 2600+ 333FSB processor to 400Mhz (200*11?), but are only running PC2700 (333Mhz) ram. If you had to lower the memory timings to run it sync at 400, or if are running it async at 333, you are going to take a major performance hit that is offseting most of your CPU performance gain. You would be better off running the CPU at 333MHz, at a higher clock multiple (e.g. 13 or 13.5) running sync with ram. That will help your 3DMark 2001 scores and actual game performance more than the 400FSB with underperforming ram (I've run the stats both ways, so I know).

Right now it runs at 175. My memory has troubles with 384kb blocks. compare. In this benchie, i just overclocked FSB to 195, with cpu at 11x, to give memory more juice. It ran at 9-3-3-3. Im planning on sticking a 512 Corsair XMS module in the slot 3 of me nForce, giving it a separate memory controller, thus hoping that Corsair would compensate for 384kb block hardship. Even if it means that the Corsair'd have to work at 8-3-3-3.

 

[selfbuilt]

In my experience, those sorts of memory timings are likely to limit your 3DMark01 and Unreal 2K3 benchies considerably. Although everyone gets bent out of shape about bandwidth, latency is what really seems to affect performance (at least on AMD systems). Try running something at least 7-3-3-2.5 (or 6-3-2-2.5 which I do easily with my PC2700) at the 166MHz bus, with the highest CPU multiple you can handle. I think you'll find it makes a measurable difference on your benchies and game performance.

 

[SickBeast]

Originally posted by: selfbuilt
In my experience, those sorts of memory timings are likely to limit your 3DMark01 and Unreal 2K3 benchies considerably. Although everyone gets bent out of shape about bandwidth, latency is what really seems to affect performance (at least on AMD systems). Try running something at least 7-3-3-2.5 (or 6-3-2-2.5 which I do easily with my PC2700) at the 166MHz bus, with the highest CPU multiple you can handle. I think you'll find it makes a measurable difference on your benchies and game performance.

Is that better than using the "Aggressive" setting on my Asus A7N-8X?

I have found that overclocking memory can very easily lead to instability and can even cause your system to not even post. What I'm wondering is if my mobo automatcially chooses a relatively good set of memory timings for me.

 

[squidman]

Ooh, i just found a "golden middle" - the timing at which the system is superstable, and really fast! here. The 3dmark didnt change significantly: here. But this is MORE than makes me happy: i AM runnig stock. I was probably too worried about it for no reason.
THanks y'all!

 

[Genx87]

I cant believe people get all worked up over thier 3dmark03 scores.

 

[selfbuilt]

Originally posted by: SickBeast
Is that better than using the "Aggressive" setting on my Asus A7N-8X?
I have found that overclocking memory can very easily lead to instability and can even cause your system to not even post. What I'm wondering is if my mobo automatcially chooses a relatively good set of memory timings for me.

Using Aggressive is likely fine, if you are not o/c the system. I've manually set mine to 6-2-3-2.5 and it is perfectly stable, and gives small but significantly better gaming scores over Agressive. The problem is that people seem perfectly happy to lower the mem timings in order to increase the FSB (i.e. sacrifice latency for bandwidth). However, in my tests, increasing bandwidth has a fairly neglible benefit in gaming whereas increasing latency signficant impairs it!

From experience, I know that if you increase bandwith at the expense of latency, you can improve synthetic cpu/mem benchmarks (e.g. Sandra, PCMark) while decreasing actual gaming performance! Try running the UT 2K3 benchmark ... you might be surprised at the results of that "optimal" configuration!