Old vs New

[PoopyPants]

ok here is the thing. my old system got way better benching and so far gaming scores, sometimes. and for the life of me i cant figure out why..

Old Config
AMD64 3200+ CG Clawhammer skt754 1meg L2 @ 2.5ghz
DFI LP UT 250GB
1gig OCZ pc4000 2.5-4-3-10 @ 260 SINGLE channel
BFG 6800 GT 440/1200
SB Audigy 2 ZS
19" CRT Monitor Vert Sync OFF

New Config
AMD64 3200+ Venice @ 2.65 skt939 512k L2
MSI NEO2 Platinum
1 gig Mushkin Redline PC4000 @ 265 2-2-2-5 DUAL CHANNEL
eVGA 6800U @ 425/1200 or 440/1200
sblive
Dell 2005FPW Vert Sync OFF

ok my old system would run 6300 in 3dmark 05 and 13k+ in '03 and 27k+ in '01

the new system REFUSES to go any higher than the following settings.
3dmark '05 - 56xx
3dmark '03 - 13,5xx
3dmark '01 - 25,6xx


now first of all i should be getting better scores in everything. Everything about this system just screams and runs right over that clawhammer system, atleast on paper.

basically what im looking at after much testing is something within my current config that is actually stopping the system from getting higher scores and yet i cant figure out what it is.
The Venice just tromps the Clawhammer -
The mushkin just tromps the OCZ and i mean big time
The DFI is a better board than the MSI but one is 754 one is 939
The DFI is a single channel ram board and the MSI is dual channel

The one and only factor at this time that i can see as being the limiting factor is the AGP bus on the MSI board... something about it is... simply put,,, BAD or so it seems.
Ive had another board ages ago, an Intel 850 board that had a bad AGP bus that made games and benches run so horribly it was just petrifying.

what confuses me more than anything is why 3dmark '01 isnt showing a massive increase because of my ram running 265 2-2-2-5 DUAL channel vs 260 2.5-4-3-10 Single channel. and this Venice cpu running 2.65 or 2.7 vs a CG Clawhammer 754 running 2.5ghz

The other issue that baffles me is why nothing i do causes scores to go higher.. i mean look the cpu will run 2.7ghz look at that Ram 265 2-2-2-5 ,, no overclocking or changing of the nvidia settings is allowing me to score higher.. nothing....

Are the NEO2 boards known to have flakey AGP slots/bus ? i havent seen anything but as of now i havent searched real hard.

I'm just simply lost cuz i have no clue wth is going on with my system and why my old clawhammer was getting way better benching than this system does.. i mean realistically this system should be just tropping that other one...

any advice ?

 

[blckgrffn]

Well, you need that Audigy back for the 03' benches for sure, your sblive is killing you by crunching all the "hardware" stuff on the CPU. I would start there. This could be contributing to your lower scores in games as well, especially if you have EAX enabled.

Moving on to '05, are you getting roughly the same FPS as you were before? The only thing that clawhammer had going for it was the L2 cache, but 01 and 05 might actually be pretty sensitive to that. Basically, the cache may mean more than the bandwidth.

The scores should be about the same, given that your 6800GT and 6800U are clocked the same. Actually, clocked identically, the U will be slower as it has more relaxed memory timings so that it could have a reliably higher memory frequency.

Hope some of that helps

Nat

 

[PoopyPants]

yeah that L2 helped i know that i can feel it.

but what confounds me is the lack of any gain in scores especially overclocked as high as i am..
im getting very close scores from mild overclock to max overclock on cpu and GPU ..

now im not looking for massive increases but with 3dmark 05 that 6800 GT and clawhammer ran 6300+ and i cant even break 5700 with the venice and Ultra. that baffles me...
now the Ultra has faster ram ,, not looser timings,, the timings are actually looser on a GT

its just wierd to me that the scores arent increasing... and that '01 score just confuses me...

 

[blckgrffn]

Actually, the timings are the other way around from what you said. THUGSROOK has a thread here that details flashing your U with GT bios to get a performance increase and control over your fan speed...

 

[yourdeardaniel]

sb live
DFI mobo better
1mb cache vs 512kb

 

[blckgrffn]

The DFI is actually a bit of a bear to configure out of the box to get optimal performance as it has so many tweaks. That said, the best score I ever achieved on my DFI with a NC 3400+ @ 2.6 with 1 gig mushkin blue 2-3-2-10 @ 217 mhz and my 6800U at 445/1170 was only about 25,000 in '01, so your tweaks must have been pretty good

Nat

 

[wanderer27]

Okay, I guess I've been out of the Loop on this one for awhile.

You're all saying the SB Live! Card is acting as a drag on his System ?

I always thought they were supposed to offload the sound processing to the SB board .

Is there a thread/article here somewhere where I can catch up on this ?

 

[Topweasel]

they still require CPU cycles and the Audigy uses less then the Live.

 

[sundev]

As others have said, try switching the sound cards between the two computers.. though I really don't see that increasing the 3DMark05 score by 600 points.

A little off topic, but how did you get your A64 3200+ Claw to 2500MHz?? I have the exact same CPU (and motherboard), and I can't get much past 2300MHz.

 

[Brunnis]

Not wanting to be a smart ass here, but the soundcard has absolutely zero impact on 3DMark scores.

 

[blckgrffn]

Try 03, yep, you would be wrong

 

[Brunnis]

Originally posted by: blckgrffn
Try 03, yep, you would be wrong

Hehe... Nope. Ony the game tests (the first four tests) make up the score.

 

[KoolDrew]

Originally posted by: yourdeardaniel
sb live
DFI mobo better
1mb cache vs 512kb

 

[Leper Messiah]

nice OC on the Clawhammer, even if it is a CG stepping. Took 1.85vcore to get mine to 2400.

IDK think a 5% CPU ulization is going to matter that much to 3DM05, considering its GPU capped, not CPU capped like 01, and 03 (possibly) I would be looking at my drivers, seeing if they are the same adn which ones are better.

 

[AiponGkooja]

Wish we could delete our own posts...I hate my internet, for the fifth time

 

[AiponGkooja]

Originally posted by: Leper Messiah
nice OC on the Clawhammer, even if it is a CG stepping. Took 1.85vcore to get mine to 2400.

IDK think a 5% CPU ulization is going to matter that much to 3DM05, considering its GPU capped, not CPU capped like 01, and 03 (possibly) I would be looking at my drivers, seeing if they are the same adn which ones are better.

I'll have to find it again, but for the first two tests in 3DM05, CPU DOES matter quite a bit. For the third one it does not. Was on a thread somewhere where the person first oc'd the CPU and then the vid card (with CPU back to normal) to check.

Found it: http://forums.vr-zone.com.sg/showthread.php?t=27074&page=2&pp=15

Edit: Also it's showing a 7800gtx's cpu limitedness...but I thought maybe it'd be relevent...but I could be wrong...

 

[Markfw]

why don't you swap the audio cards, then test, then leave audio switched, and swap video cards.

 

[touchmyichi]

...What was the point of that upgrade? All that stuff is hardly better than your old rig

 

[episodic]

Originally posted by: touchmyichi
...What was the point of that upgrade? All that stuff is hardly better than your old rig

QFT

 

[PoopyPants]

the point was dual channel ram, i traded the video card and some ram for a 2005FPW so i got back a better card... the mobo i had no choice on its still one of the best 939 AGP boards. The ram,, well that ram just kicks the poop out of everything,,, or it did until it died shortly after getting it.

the sblive has absolutely nothing to do with the benchmarks since there are absolutely no sound samples at all in any of those tests,, why would you boys think that the sblive hurts performane in 3dmark 01 03 and 05 ? there are only a couple of tests in that entire suite that uses the sound card and i aint using the ENTIRE test suite. all you need are the Game tests to produce a score.

i found one of my old clawhammer/6800 GT score files 6100 something i wish i could find the 6300 that was my all time best.

and Venice = SSE3 , lower volts, lower heat, more performance, than a clawhammer, anyone knows that and to believe otherwise is sheer rubbish. oh and need i say once again dual channel ram vs single channel ? ESPECIALLY 260 2-2-2-5 vs 2.5-4-3-10 big different kiddies.


like i said and ill say it agian.. going from a mild overclock on the system to pushing the envelope with this hardware shows very little difference in scores. 500 points maybe if im lucky worst case is 1000 pts.

something is wrong... its like verticle sync of the entire system is on and it cant score past a certain point.


anyone here with similar hardware or who has had similar knows darn well that this setup running the settings i showed you should be gankin way better scores ESPECIALLY in 2001. AND 2005 let alone 2003..

i keep forgetting to install Aquamark . .. i think ill do that and see what happens.. if its the same poopy results then i know something isnt right.