Sign in with
Sign up | Sign in
Your question

Shouldn't my SLI 2x GTX 260 216 be scoring higher on 3DMark Vantage?

Last response: in Graphics & Displays
Share
April 22, 2012 8:09:06 PM

I'm not too great with the details of graphics card tweaking. Please help a newb out!

Main card is EVGA GTX 260 Core 216 55nm edition.
SLI card is EVGA GTX 260 Core 216 65nm Superclocked edition. (SC is set as SLI card due to overheating issues.)
CPU is Core i7 920 with TRUE heatsink, overclocked to 3.6 GHz, doesn't run too hot ever.
Memory is 12GB G.Skill DDR3 1600 running in Triple Channel, clocking at 8-8-8-24 666MHZ (ooh, spooky!)

http://3dmark.com/3dmv/4032309 - My system with 2x SLI and slight overclocks, stock voltage, no hardware mods.
http://3dmark.com/3dmv/4028674 - My system with SLI disabled and no overclocks.

Here are links to other systems performing BETTER than mine.. some with only 1 card!

2 cards:
http://3dmark.com/3dmv/1471306
http://3dmark.com/3dmv/1421611
http://3dmark.com/3dmv/1293218

1 card:
http://3dmark.com/3dmv/3729614
http://3dmark.com/3dmv/2853982

So.. what gives?

Does having a dedicated PhysX card make THAT much of a difference?

or..

If the graphics driver on most of these is unsupported, are they possibly running some sort of unofficial super driver to get this score? If so, is it against the rules to post a link and/or to use one?

or..

Are they just hacking their score so they can look better.. which would be ridiculously lame and I don't see why someone would do that.

or..

is there something I'm missing with my setup that I can tweak to get better performance?


I don't have the money to buy all new cards, so I'm trying to make the best with what I have for now. If I can squeeze the extra 30 fps that these guys seem to be doing, I'd be more than happy for a long time. Any help would be much appreciated! If you have any questions regarding my setup, I'll answer what I can.
a c 153 U Graphics card
April 22, 2012 8:19:42 PM

I believe the reason for those scores is the Physx score is invalid (if you look at the top of the links). Before the newest release of Vantage you could chose which way to run Physx, either off the CPU or off the GPU this wasn't very fair to AMD users, and it made it hard to get everyone on the same page. Eventually people just chose a way to do it, the creators caught on and added that clause to all the scores (the invalid Physx clause).

Regardless I have SLI 260s, let me run Vantage for you and see what I get. I'll get back to you.

However I'm 100 percent your setup is just fine. Why don't you try running the test with the Phsyx option on GPU and see what happens.
m
0
l
a c 153 U Graphics card
April 22, 2012 8:38:52 PM

Just so you know I ran it and got 22438. 19013 GPU and 48799 CPU.
m
0
l
Related resources
April 22, 2012 9:00:45 PM

Thank you much for your feedback. Is there any chance you saved your report? If not then whatever, but if you did, any chance you could tell me what your FPS was on Jane Nash and New Calico? And are you saying that it was PhysX that gave those people 90FPS benchmarks on both those games? Thats mainly what I'm wondering... if those guys can get 90FPS with their SLI, then how can I? Even if it involves getting a third 260 for PhysX?
m
0
l
April 22, 2012 9:01:55 PM

Also if its not too much trouble, what brand and model 260s do you have? Core 216s, regulars?
m
0
l
a c 153 U Graphics card
April 22, 2012 9:37:24 PM

I just have the regulars dude. Probably just overclocked a bit higher then your own. It should be noted that I run my cards hotter then I should, but at this point I don't care, I'm in need of an upgrade (as I'm sure you're aware).

I don't even think you can set the option in Vantage anymore like you used to. You used to be able to run Physx off the graphics card, and not the CPU. If you still CAN do that (I couldn't find the option myself) go ahead and try to and see what happens.

A card can computate Physx while still in SLI. So this would give the Physx scores HUGE boosts.

Basically only compare it to scores that have a VALID Physx score.

No I didn't save the score but I can rerun it for you if you wish. I see now what you are saying about them getting such high FPS, there is more to it than just the Physx, yes.. But it probably from uber overclocks, maybe even water cooling. I'll check my FPS when I re-run but I know it was no where near that high.
m
0
l
April 22, 2012 10:00:01 PM

His CPU score seems low to me if he's OC'd
m
0
l
April 22, 2012 10:23:46 PM

Helltech said:
A card can computate Physx while still in SLI. So this would give the Physx scores HUGE boosts.


Wait, really!?! I figured that if I was SLI'ing, I'd not want to strain the second card, so I've had my NVIDIA Profile set to have the CPU do the PhysX. Now I do have a pretty beefy processor, quadcore hyperthreaded i7 920 @ 3.62 GHz, but I'm gonna try running Vantage with PhysX set!!

You may have just been a HUGE help!
m
0
l
a c 153 U Graphics card
April 22, 2012 10:27:41 PM

Well yeah, when you play Mafia II and use SLI the second card will be used for Physx also, if you have it set up that way, I do at least. Always have SLI set up so that in SLI the second card uses Physx. Which is automatically set (I believe).

Regardless of that your score on Vantage wont be "valid". In fact you should just forget vantage all together, there are many other amazing benchmarks that are more consitent than vantage, malmental pointed out many of them.

EDIT - Also when I gave out my first scores for Vantage I just totally realized even I was using invalid Physx.

What I get when I have PPU off in Vantage (which is what makes it valid).

3DMark Score

19903 3DMarks
--------------------------------------------------------------------------------
Graphics Score

19064
--------------------------------------------------------------------------------
CPU Score

22931
--------------------------------------------------------------------------------
Jane Nash

57.4 FPS
--------------------------------------------------------------------------------
New Calico

54.23 FPS
--------------------------------------------------------------------------------
AI Test

3208 operations/s
--------------------------------------------------------------------------------
Physics Test

29 operations/s
--------------------------------------------------------------------------------
Texture Fill

40 texels/s
--------------------------------------------------------------------------------
Color Fill

11 pixels/s
--------------------------------------------------------------------------------
Pixel Shader

63.93 FPS
--------------------------------------------------------------------------------
Stream Out

35.25 FPS
--------------------------------------------------------------------------------
GPU Particles

40.5 FPS
--------------------------------------------------------------------------------
Perlin Noise

70.33 FPS
m
0
l
April 22, 2012 10:36:19 PM

Hm. Set PhysX to GPU2 as recommended, and saw a decrease...

http://3dmark.com/3dmv/4040084

That was however, before I saw anything regarding PPU... Gonna try that now.
m
0
l
a c 153 U Graphics card
April 22, 2012 10:39:29 PM

triplefam said:
Hm. Set PhysX to GPU2 as recommended, and saw a decrease...

http://3dmark.com/3dmv/4040084

That was however, before I saw anything regarding PPU... Gonna try that now.


Your score is already valid, which would mean "Disable PPU" is already checked (presumably because you have the newest version of vantage I'm guessing).

Honestly you're scores are just about the same as mine. I'm sure those ridiculous scores you are seeing are people who have modded bios and liquid cooled/liquid nitrogen cooled their systems.
m
0
l
April 22, 2012 10:41:45 PM

Helltech said:
Your score is already valid, which would mean "Disable PPU" is already checked (presumably because you have the newest version of vantage I'm guessing).

Honestly you're scores are just about the same as mine. I'm sure those ridiculous scores you are seeing are people who have modded bios and liquid cooled/liquid nitrogen cooled their systems.


I guess I just wanna see if I can get 90 fps like those guys did. I want to rule out the idea that there's something wrong with my setup. You got pretty much the same score as me with PPU disabled, so it seems that may have been it afterall.

Gonna try this now
m
0
l
a c 153 U Graphics card
April 22, 2012 10:43:57 PM

There are always going to be insane scores posted out there from people who do crazy sub-zero overclocking, and there is no way at all anyone on aircooling would touch those scores. I wouldn't worry too much about it.

We basically have the same rig, and are getting roughly the same scores, so there is nothing at fault with your system. I have pushed mine as far as I know it will go.

EDIT - I kind of mislead you at the beginning of the thread, so sorry about that. I just assumed at first their amazing scores were because of the Physx differences that came along with Vantage at its youth, which in my defense is true and it changes scores significantly depending on how you configure it. It wasn't until I ran the test myself (haven't done it in years), that I realized you were comparing your scores to ones that are more than likely sub-zero overclocking scores.
m
0
l
April 22, 2012 10:55:37 PM

No problem whatsoever! I know now to have SLI set to GPU2, or Auto, which still suggests GPU2.

http://3dmark.com/3dmv/4040101

Turning "Disable PPU" off gave me +3000 vantage points, and gave a VERY small FPS boost, but a HUGE strain off the CPU, the Physics test being MANY TIMES faster.

Good to hear that there's nothing wrong with my cards, and that definitely makes sense to me that these guys were probably doing some intense overclocking.
m
0
l
April 22, 2012 10:57:56 PM

Thanks again everybody!
m
0
l
April 22, 2012 11:00:42 PM

Quote:
no, never set your PhysX to the CPU.
(especially in SLi)
you can leave it set to AUTO or set it to your second GPU and try that.
I have 2 SLi units and all my units run GeForce.

also if you wouldn't mind, try running without hyper-threading.
I would like to see the results (difference).


I'd be happy to, thanks for the help! Problem is.. I don't know how to disable hyperthreading :sweat: 
m
0
l
April 22, 2012 11:05:46 PM

Right on, I'll get to it in just a few!
m
0
l
April 25, 2012 1:15:15 AM

Quote:
it would be in the BIOS, there is an option top disable.
and lots of times the unit (depending on app or game) will run better.
but on average hyper-threading can be like a 15-20% boost (excellent circumstances).


Here we go!

HYPERTHREADING ON: (1/8 physical/logical processors, 4 core)
http://3dmark.com/3dmv/4040101

HYPERTHREADING OFF: (1/4 physical/logical processors, 4 core)
http://3dmark.com/3dmv/4043751
m
0
l
April 25, 2012 1:51:14 AM

Quote:
not a lot of difference in 3D Mark testing..
you got a strong unit there bro.


I don't get it, lol was that sarcasm?? :lol:  Cause I was hoping for a much greater increase :( . Perhaps I should run some PCMark benchmarks, or, are you aware of any better CPU benchmarking software?
m
0
l
April 25, 2012 2:20:18 AM

Eh well. Considering I can run BF3 on almost High, 1080p Caspian Border 64 player with explosions and everything and only dip below 60 during VERY intense moments, I'm happy for a little while I think :) 

Nonetheless I'm a statistics junkie when it comes down to it, so considering at some point I was gonna run Unigine anyway, I suppose I'll get my results and then support the database (that is, the MULTIPLE ones you posted!) by posting my results!
m
0
l
!