Low 3DMark06 Score EVGA 9800gx2?

xblackdemonx

Distinguished
Jan 24, 2007
71
0
18,630
Hi, I've bought a EVGA 9800GX2 recently and it seems like i have a low score.
My Rig is not that crappy though, i've seen that k|ngp|n managed to get 24k marks with little Overclock.
I've got a max of 13300 marks on stock settings. Does anyone know what could be causing this ? is it my CPU that Bottlenecks?

k|ngp|n score:http://www.xtremesystems.org/forums/showthread.php?t=180875

here are my specs:

XFX 680i SLi mobo
Single EVGA Geforce 9800GX2 stock cooled and un-modded
Intel Q6600 Core2Quad @ 2.4 GHZ
4x1GB Crucial Ballistix 1066MHZ 5-5-5-15
CORSAIR HX620 620w PSU

Thanks :hello:
 

billiardicus

Distinguished
Mar 2, 2008
186
0
18,680
I dunno, but if you OC that Q6600 you score will go way up

I have Q6600 and a 8800GTS. Stock I got 11400. With Q6600 at 3.5ghz, and the 8800GTS at 800/1950/1100, I got 15800.
 

xblackdemonx

Distinguished
Jan 24, 2007
71
0
18,630
LOL well first of all i don't play wow...anymore (ex-addict)
and even if that guy has a CPU @ 5.7 GHZ why would he get like 11k more points, i would at least expect a 15-16k pts ... no ?
 

Evilonigiri

Splendid
Jun 8, 2007
4,381
0
22,780
The vidcard is being held back by the cpu, so obviously OCing cpu will bring out the potential of the vidcard. I must say, even at 3.6GHz on a quad will bottleneck the 9800GX2 in 3dmark06. That's probably why there's such a huge increase in points at 5.7GHz.
 

xblackdemonx

Distinguished
Jan 24, 2007
71
0
18,630
hum... thats crazy that this card needs so much CPU power ...
ya i guess you are right, i OCed from 2.4 to 2.8 and went from 13300 to 14910 pts.

i guess ill try to OC it at least to 3 ghz :)

Thanks
 

Perp

Distinguished
Feb 27, 2008
144
0
18,680
Also there is a large score component that is CPU only. In that area alone his scores will crush yours.

I scored 17700 with Q6600@3.6 and a single 3870x2.

 

xblackdemonx

Distinguished
Jan 24, 2007
71
0
18,630
Ouch, how did you get the Q6600 up to 3.6 GHZ ?
can you please tell me how you've done your setup ?

Ram timings, Vcore, etc .. :p

Thanks
 

billiardicus

Distinguished
Mar 2, 2008
186
0
18,680
I can get my Q6600 to run at 3.6, but the vcore has to be ~1.5, and I've got to max the FSB and MCH voltages (+.3 for each on my P35-DSL3). I left the ram running at stock requirements. I don't run this setting because my temps were ~70C.

It's pretty hard to do 3.6ghz on air, but there's a few guys that have done it. I think most are using a lapped TRUE and cpu.
 

Dunkel

Distinguished
Jan 19, 2008
454
0
18,780
@billiardicus

1. Did you get to 3.6 with 400x9?

2. Did you run Prime at that clock?

3. You added +0.3 to MCH and +0.3 to the FSB Overvolt?

4. Do you have a fan on the NB?

Sorry for all the questions. Thanks.
 

xblackdemonx

Distinguished
Jan 24, 2007
71
0
18,630


Nice, I managed to get to 3GHZ,

Vcore 1.31v
Ram 1068mhz @ 6-6-6-15

and i got 15921 in 3dmark06 :whistle:
:bounce:

Thanks a lot for the tips
 

SpinachEater

Distinguished
Oct 10, 2007
1,769
0
19,810
No man...No! You can't compare your scores to a professional OCer in the first place. Second, he is using uber high end DDR3 that costs over $500 and is OCd with new timings, a different and better mobo, an OCd GPU (738 check it), and an OCd CPU almost twice as fast as yours AND that is a 45nm. A little overclocked? Yeah, no, a lot overclocked IMO. You are trying to race your soccer mom mini van with a Ferrari.

If you want to compare 3Dmark06 scores look at Gx2 reviews where they aren't using a liquid nitrogen cooled setup. I know the first scores were on air but still you can't ignore the rest of the system because as you can see it drastically affects 3Dmark scores [:thegreatgrapeape:5]
 

xblackdemonx

Distinguished
Jan 24, 2007
71
0
18,630
Ya you are right, the DDR3 doesn't matter though since it only counts the video card and CPU.

But ya, that CPU OC is crazy ...

i thought that if he got 24k just on air OC that i would get at least near 16 k without overclocking anything ... i guess i was wrong :(

but anyways... thanks everyone for your help and your tips ;)
 

Perp

Distinguished
Feb 27, 2008
144
0
18,680


400*9 for 3.6ghz

1:1 ram ratio at 800mhz

Stock voltages for everything but the CPU.

CPU voltage @ 1.36

I have a VID of 1.2375 so no doubt there are people who need more voltage to hit a stable 3.6ghz.
 

Perp

Distinguished
Feb 27, 2008
144
0
18,680
@Dunkel

There are lots of people zooming past 3.6ghz with lapped CPUs and the best air cooling available. I'm shooting for more latter, but I won't have time to pull my PC apart to do the work for another week or two.
 

dagger

Splendid
Mar 23, 2008
5,624
0
25,780
You are using motherboard with 680i chipset. It will not go up to 3.6ghz no matter what you do due to FSB bottleneck. You need 780i or P35 for 3.6ghz, x38/48 for 3.8 and beyond. :p
 

billiardicus

Distinguished
Mar 2, 2008
186
0
18,680


1. Yeah 9x400
2. I did run prime, but for only ~20mins to see what the temps would be. Temps were ~70C, so backed off to 3.5ghz. Since then I've improved my cooling (I've got a Xigmatek S1283, I replaced the fan w/ a 110 CFM, and added more case fans), but I haven't really messed with 3.6 anymore because IMO I'm addding so much voltage and heat for so little gain. There's no point unless you're after a quick 3dMark06.
3. Yes
4. No. But the CPU fan sits right above it. I imagine there's pretty good airflow around my NB.

For everyday use, I run my CPU at 3.3ghz. I got to 3.5 stable (ran FFT for 8 hours), but I can run 3.3@1.35vcore with no MCH or FSB overvolt. 3.5ghz requires 1.45vcore and +.2 FSB and MCH overvolt. I don't think the small increase in speed (.2ghz) is worth it.

Depending on your chip, mobo, etc you should be able to get ~3.0ghz pretty easily on air. IMO, it's not worth to try for a whole lot more than that.

- You start needing "exotic cooling"
- You burn a lot more power and may need a high end PSU
- You get very little performance increase

 

Dunkel

Distinguished
Jan 19, 2008
454
0
18,780
Thanks for the info. I'm running my E2160 stable @3.2 so I'm pretty happy. But like your 3DMark comment, I'm trying to post and run at >3.3GHz and I just can't do it. I can't even post. I added +.2 to MCH a few days ago and I think my board went out. RMA to gigabyte, we'll see what they say. :eek:
 

SpinachEater

Distinguished
Oct 10, 2007
1,769
0
19,810


Ummm...no. Where do you think 3DMark is running from? It doesn't benchmark your RAM persay, but the RAM clocks and timings will make measurable differences in 3DMark scores.
 

pb4udie

Distinguished
Apr 16, 2008
5
0
18,510


70C @ 3.6? lol is that w/o the cooler on the cpu? No seriously try to rearange the wires in your chassis to get better air flow. I'm working in a PC shop/repair service. A month ago I was assembling a rig consisting of ASUS P53, q6600, XFX 8800 GTS 512 MB XXX in Thermaltake Soprano. The special thing about the rig was that a coworker of mine talked in the customer to buy Thermaltake Big Typhoon. He was buying a new cooler at the time so we wanted to test it and c how well it performs. The bottleneck was the memory. Corsair TWIN2X2048-6400C4DHX was yet on it's way to the store so we had to improvise with single Kingmax 2GB 5-5-5-15 @800 Mhz. So to get to the point... Around 12.5k 3D Mark06 all default 35C. After several hours of child play the final resoult was 16301 in 3D Mark06 with default gpu and memory cloaks cpu q6600 @ 3.4 Ghz 377,77 x 9, and I only stopped there becouse all of my coworkers were begging me =) the pont is that after 20 minutes of prime 95 the temperature went up by 2C, that is from 35 to 37C. The q6600 was G0, BigTyp was the original 1 not the VX (IMO original is better than VX), no lapping (plane old 99% silver compund thermal paste) =) Now, as allways the same results can't be guaranteed becouse 2 exactly same pieces of hardware do not exist. But as a tip for noobs i would sugest sticking to Asus mobo Big Typ and a good thermal paste.
 

SpinachEater

Distinguished
Oct 10, 2007
1,769
0
19,810



In some cases yes, in some cases no...it depends on what hardware component is the bottleneck. And BTW, all of your hardware components contribute to the FPS count, not just the GPU. Tighter timings at the least will stabilize your FPS and smoothen gameplay. You don't even have to try this yourself, I would suggest doing so however just so you know, but look through the forums, when people tighten up their RAM timings they up their 3DMark scores.

So this brings me back to my main point....OCd RAM will improve your 3DMark scores. You have to take in consideration that Klngpln was using OCd DDR3 and you have stock DDR2. It makes a difference.