Sign in with
Sign up | Sign in
Your question

What would you most like to see?

Last response: in Graphics & Displays
Share
March 6, 2008 12:27:37 AM

nVidia GeForce 9800GX2 & 9800GTX benchmark scores from 3DMark06

or

nVidia GeForce 9800GX2 & 9800GTX Frames per Second screenshots & stats from Actual Games!


If these websites are capable of sitting there for 10 minutes running a benchmark with these unreleased cards, then WTF are they not testing games with them!?!?!?

For instance... CRYSIS.

Or maybe all these benchmarks are made up and nobody can really predict how these cards will perform with games like Crysis because they dont have the cards yet!

So.... if the benchmarks arent true, neither are the specs. And if they arent true... then the 9800GTX & GX2 can still be winners.

Fingers crossed.

But back to my point....... if you had a 9800GTX what would you check first? Its 3DMark06 score or would you stick Crysis on @ Ultra High and find out some worthwhile stats?

More about : question

March 6, 2008 12:38:36 AM

Oblivion,Tomb Raider Legend (next gen on) and Company Of Heroes, my favourite games :D 
March 6, 2008 12:41:52 AM

Oblivion has to be the best game to test performance, put in a 9600gt into a friends system last week and it still wouldnt allow me to run Oblivion with all settings maxed out flawlessly (still some chugging!)
Related resources
a b U Graphics card
March 6, 2008 12:57:45 AM

Yeah, I hear ya. If I had a 9800GTX I would skip 3dmarks for sure and test it vs my single and SLI 8800GT in crysis, then Oblivion, COD4, etc. Better yet, find an 8800GTX to test it against in the same games.

I'd rather see real life games comparing multiple cards at the same settings for sure. But if no other cards are included for comparison, then 3dmarks with system specs may give us more info.

What would it mean to see a crysis screenie showing 40 fps if we have no idea what some other card renders in the same situation? Example- Any one of these sceenies would mean little by itself, but all four together give some info. Top to bottom X1950XT, 8800GTS 320MB, 8800GT 512MB and SLI 8800GT 512MB.




a b U Graphics card
March 6, 2008 3:26:00 AM

Honestly, the only benchmark we need is crysis, thats it.
March 6, 2008 4:22:23 AM

Annisman said:
Honestly, the only benchmark we need is crysis, thats it.



Not quite. That is actually the worst since it doesn't do any GPU justice. You can't tell how a GPU does overall based on Crysis marks.

To dev1se, NDAs prevent the release of official benchmarks before the official release of a card. I would imagine that sites get right down to testing as soon as a sample is in hand but they cannot officially post results that violate dates in the NDA.
a b U Graphics card
March 6, 2008 4:32:55 AM

SpinachEater said:
Not quite. That is actually the worst since it doesn't do any GPU justice. You can't tell how a GPU does overall based on Crysis marks.

To dev1se, NDAs prevent the release of official benchmarks before the official release of a card. I would imagine that sites get right down to testing as soon as a sample is in hand but they cannot officially post results that violate dates in the NDA.


Oh, don't B.S. around the fact man, if a GPU can run Crysis well, then it is very safe to assume it can handle any other game out there alright. That's the point im getting at.
March 6, 2008 6:11:42 AM

What now I want to see are the official specs of the 9800 GTX, and NVIDIA has to confirm it.

And also the official benchmarks of the 9800 GX2 and the GTX as well.
a c 130 U Graphics card
a b Î Nvidia
March 6, 2008 10:52:32 AM


@ dev1se
You know its all well and good setting every thing to ultra high and seeing how close you can get to running a game propperly but most people will be moving the sliders to the left quite a bit.
You do need a decent range of games and other cards to get a worthwhile result from any set of benchmarking tests as pauldh is saying.
Oblivion is a good test of a PC and so is Crysis but either one on its own cant be considered conclusive. Both these games have scalebility (unlike Assassins Creed) so you can play them to the limits of your machine.

Personally I would do just what you are saying but the second thing would be to run 3dm and see what the increase from my last card was.
Mactronix
March 6, 2008 12:17:44 PM

Annisman said:
Oh, don't B.S. around the fact man, if a GPU can run Crysis well, then it is very safe to assume it can handle any other game out there alright. That's the point im getting at.


I see what you are saying since everyone is waiting for a card that can tackle Crysis. To expand on my point, each card only has like 5-10 FPS difference from the next in Crysis benchmarks depending on resolution and settings and SLI and xfire efficiencies get stomped into the mud.

Considering a margin of error in measuring FPS and the fact that GPU architecture favors certain game engines over others...I wouldn't base performance on just one game where you are only going to see a 10 FPS difference. Maybe you play Crysis and only Crysis so you are only interested in a card that knocks it out of the park. That is fine, but there are people who do other things with their cards.
a b U Graphics card
March 6, 2008 2:14:28 PM

SpinachEater said:
I see what you are saying since everyone is waiting for a card that can tackle Crysis. To expand on my point, each card only has like 5-10 FPS difference from the next in Crysis benchmarks depending on resolution and settings and SLI and xfire efficiencies get stomped into the mud.

Considering a margin of error in measuring FPS and the fact that GPU architecture favors certain game engines over others...I wouldn't base performance on just one game where you are only going to see a 10 FPS difference. Maybe you play Crysis and only Crysis so you are only interested in a card that knocks it out of the park. That is fine, but there are people who do other things with their cards.

SLI/Crossfire efficiencies get stomped into the mud? 10 fps doesn't matter? I don't agree. Problem is Even two of anything currently gets hammered by crysis when you crank the res and settings. But It's still the way to go and scales very well in the GPU demanding levels. Shoot, you can see near 100% scaling in crysis, but it will be 10 fps vs 20 fps. See 19x12 here: http://www.firingsquad.com/hardware/geforce_9600_gt_sli...

But I'm quite happy that SLI makes the settings I want to play playable, and scaling like below isn't bad. 13 fps may not be much, but it's over 80% scaling and this level crushes one card yet remains playable with two cards.

compressed to fit on imageshack, but still a large file so I'll just post the link:
http://img159.imageshack.us/img159/289/crysis8800gt1vs2...
March 6, 2008 2:28:04 PM

Sounds to me like hardware is falling behind in performance. Use to be able to buy top of the line GPU/CPU and not have to worry about any game to come for 2 years. Either that or game developers are just way ahead of the hardware there programing for.
March 6, 2008 3:02:59 PM

mpavao81 said:
Sounds to me like hardware is falling behind in performance. Use to be able to buy top of the line GPU/CPU and not have to worry about any game to come for 2 years. Either that or game developers are just way ahead of the hardware there programing for.


Where were you when doom3 and oblivion came out?
March 6, 2008 3:09:34 PM

pauldh said:
SLI/Crossfire efficiencies get stomped into the mud? 10 fps doesn't matter? I don't agree. Problem is Even two of anything currently gets hammered by crysis when you crank the res and settings. But It's still the way to go and scales very well in the GPU demanding levels. Shoot, you can see near 100% scaling in crysis, but it will be 10 fps vs 20 fps. See 19x12 here: http://www.firingsquad.com/hardware/geforce_9600_gt_sli...

But I'm quite happy that SLI makes the settings I want to play playable, and scaling like below isn't bad. 13 fps may not be much, but it's over 80% scaling and this level crushes one card yet remains playable with two cards.

compressed to fit on imageshack, but still a large file so I'll just post the link:
http://img159.imageshack.us/img159/289/crysis8800gt1vs2...



Yeah, stomps in the mud is an exaggeration, sorry. I am shamed :pfff:  I agree with you, SLI does bring very good improvements with Crysis compared to a single GPU. However, looking at the difference between the 9600GT and 8800GT in SLI (5.5 FPS in the 1920x1200)...personally I would want to see other scores to help me decide which card is right for me. If you compare the single 8800GT (15.6 FPS) and the 9600GT (12.9 FPS)...the numbers are just too abysmal to even make a conclusion.

I wish these benchmarks would stop trying to make one setup look the best. Like two GTs are better than one GTX. SLI with the GTX would have been nice for comparision.
a b U Graphics card
March 6, 2008 3:13:56 PM

Yeah, that makes sense.

I would totally love to see someone test the complete GF86 and 88 series as well as GF96 and upcoming 98 series in SLI and single card. Large undertaking, but boy would it be interesting if they tested enough games and settings.

Kinda like Legion did, but also include SLI of each.
http://www.legionhardware.com/document.php?id=710&p=0
March 6, 2008 3:23:11 PM

Yeah I love to see benchies on a complete set like that so you can see how and where each one stands out from the crowd. It sucks when the crowd is cherry picked.
March 6, 2008 3:40:51 PM

skittle said:
Where were you when doom3 and oblivion came out?


busy working 60+ hours a week, didnt have much time for computers lol
a b U Graphics card
March 6, 2008 3:57:36 PM

A new computer WITH a 9800GTX inside show up at my door for free?

Overall, gaming benchmarks.
March 6, 2008 4:06:48 PM

pauldh said:
Yeah, that makes sense.

I would totally love to see someone test the complete GF86 and 88 series as well as GF96 and upcoming 98 series in SLI and single card. Large undertaking, but boy would it be interesting if they tested enough games and settings.

Kinda like Legion did, but also include SLI of each.
http://www.legionhardware.com/document.php?id=710&p=0

He he, we should do it! ;) 
CPU scores can obviously be written off as they're in all likelihood going to be different and at varying states of tune, but how many different systems are there on here?!?
A rather large undertaking as you said though! :lol: 
March 7, 2008 1:40:02 AM

skittle said:
Where were you when doom3 and oblivion came out?


I think he means from those games onwards, that this has been occuring, that game graphics/etc is getting further ahead of hardware.
March 7, 2008 2:10:34 AM

Annisman said:
Honestly, the only benchmark we need is crysis, thats it.


I don't even play Crysis and don't plan to. I bought Farcry used, but ended up giving it away. Some of us prefer CRPG's to FPS, and we'd like to see how a card does under those rather different conditions.

Here are some FRAPS screenshots of Oblivion and The Witcher. I'll admit, I don't have a 20" or higher LCD yet, so it's only 1024 x 768 on a CRT, so I might be CPU limited, but it's still decent at Ultrahigh quality:

Oblivion with ultrahigh quality.

Scenic view at night near Bravil:



Anvil Docks at night:



The Witcher

Cut Scene goodbyes



Fighting Barghasts at the Inn:



After the fight:


What will you have?



Anandtech benched The Witcher at 47fps at 1680 x 1050 in the first cutscene, so I'll see how it goes when I get a better monitor. I'd say the highest I get in The Witcher now is 50-60 and the lowest is in the mid 20's. With all settings maxed out except Very Sharp Detail instead of 8x AF.

mactronix said:
... the second thing would be to run 3dm and see what the increase from my last card was.
Mactronix


My last card was a 7600gs on an Nvidia 405 chipset MSI board. I switched to another budget MSI board and moved up to the 3870x2. I got very low 3DMark06 scores on my last card (the CPU and RAM were the same). Now, I get over 3 times the score, but not as good as those with overclocked Wolfies or Q6600's. Can't wait to get a 790 board and a Phenom 9750 in May (but the monitor comes first).



3DMark Score 9524 3DMarks
SM 2.0 Score 3906
SM 3.0 Score 5827
CPU Scroe 1708


1. Return to Procyon 30.72 fps
2. Firefly Forest 34.38

CPU1-Red Valley 0.54 fps
CPU2-Red Valley 0.87 fps

HDR Tests

1. Canyon Flight (SM 3.0) 73.45 fps
2. Deep Freeze (SM 3.0) 43.1 fps

dev1se said:


Or maybe all these benchmarks are made up and nobody can really predict how these cards will perform with games like Crysis because they dont have the cards yet!


I don't think the benchmarks are made up. I just think the settings are tweaked so the benchies do very well. Plus, they usually have an Intel EE quad core, so the CPU influences very high 3DMark06 scores, even at stock. What the 9600gt's in SLI seem to benefit from is that Nvidia fudge where the PCIe bus is boosted without it being obvious. That's how I believe the 9600gt in SLI beat a 3870x2 in a couple of FPS recently.

Take all benchmarks with a shaker of salt. Results may vary based on your CPU, memory, overclocking or stock, sometimes motherboard and chipset. Many things effect both benchies and "real world game tests" like H does with the wrong drivers for an ATI card and the right drivers for Nvidia.

I want to see the 9800gx2 vs. the 3870x2, then I want to see a triple SLI 9800gx2 and a 9800gtx vs. two 3870x2's. Then I want to see it done all over again when the 4870x2 arrives in June or thereabouts. Once Nvidia's next GPU arrives, then rinse and repeat. I want to see it done at every reputable tech site, but results will still vary when people get their new cards home.
!