8600gts extreme / 8800gs on Pentium 4

[mactronix]

Come on yourself, I reinstalled oblivion. Everything maxxxed at 1440x900 16xAF
Intown 50-70fps

outdoors 30-50fps

Maybe skulltrail board were talking about Karl Texture pack with much more detail textures. It usually makes your frames turn into half with that texture pack.

@ rado_88,
Go get your card and enjoy your gaming.
Mactronix
@ Marv
Look i play the game myself ok my characters are all very high levels and i know from first hand experiance that not only did you not get those numbers with everything and i mean everything at max. But i also know that from a clean install 37 mins isnt long enough to get your character into a position where you could have benched those figures.
I said it couldnt be done and you said it could and posted evidence that supported me.
Im not letting this get any further it cant be maxed on a 8800GS end of story. You have proved that with your link.
Further discusion on this matter isnt helpfull to the thread and i am not interested in arguing the point with you further.
Mactronix

 

[marvelous211]

Oh really? Installed oblivion within 5 minutes and wondered around in oblivion town and outside on a horse with my saved game files I had on my computer. I don't care what you play and how your system sucks. I installed oblivion at max settings and getting those results. what card do you have anyway? Don't tell me you are still on a 1650xt and trying to max out oblivion.

 

[pauldh]

Oblivion he will be quite CPU bound trying to play max details.

At 1600x1200 max details, a stock clocked 8800GS would average about 30 fps in the deepest foliage (most GPU demanding area of Oblivion) Keep in mind Marvelous 's 8800GS is far from a stock clocked


I searched Oblivion to find three map areas to bench. Biggest GPU hurting part I could find was through waist high foliage in the Great Forrest. That became my foliage Benchmark. Benching with fraps, A stock low clocked XFX 8800GS(580/1400) averaged 27 fps in that area. Keep in mind though that typically outdoors at the same settings average fps would be 30-40. In and around Towns (lots of npc's) are very CPU limited.

Foliage:


Town:

 

[marvelous211]

Is that max settings paul?

 

[pauldh]

Sorry, had to edit in the 16x12 max up top. If I ever get around to throwing the full review up somewhere, I'll be sure to make all settings clear.


Those are totally max details except no self shadows (I absolutely hate the way they look in Oblivion)

But of course, your clocks are monstrous compared to the stock 550/1400 XFX I used in the charts. Plus your CPU is above the X2 5600+ also.

Honestly with the XFX 8800GS at stock, and an X2 5600+ Oblivion is playable at 16x12 max (no self shadows) 0x/16x. There are lower dips in fps than I'd like, so I would tweak a few settings below max (shadows and shadow filtering, maybe a min grass height tweak). But make that a 650/1900 evga (or higher like yours) I doubt any tweaking is needed without fsaa.

Oh, and with regards to your gameplay..... 14x9 max with your system 0x/16x should be no problem at all. Shoot, 2xaa would be fine I'd think.

 

[marvelous211]

I don't get how Mactronix tries to tell me how I can't max out oblivion with my 8800gs when he doesn't even own one and don't have the same system.

If he's on a 9600gt it is actually worse than 8800gs in oblivion without AA. AA would be another story. Oblivion loves shader and bandwidth isn't a problem at this resolution without AA. The cad I have performs like stock 8800gt far as raw performance goes without AA. The pixel fillrate texture fillrate matches 8800gt and so does my SP far as stock 8800gt goes.

 

[pauldh]

I understand the frustration. In his defense, I often question what people claim because not everyone like us cares to be accurate. I think alot of careless (maybe not intentional) exaggeration takes place here. Instead of using fraps or checking fps, they guess it's smooth. They also may test one little part of a game, or run a benchie, and then say it's smooth in the game. Also some people want to make claims about their system likes it's their child, claims such as playing Crysis on very high 19x12 above 30fps on a single GPU. It's impossible and every review backs that up. I tend to call those out with links just to keep things in check so people aren't misled.

But a 3.5GHz C2D and a 700+/1700+/ 2000+ GS would easily play oblivion at 16x10 max no aa, never mind 14x9. I really can't guess the exact fps, I just know it would be very playable throughout the game.

My 320MB GTS hurt bad at 16x10 in Oblivion, but that was paired with a single core FX55. I had to tweak things down a bit. The CPU makes a difference even in the outdoor sections. But in my 3.0GHz Q6600 setup, without fsaa a single 8800GT can handle 16x10 Oblivion no problem at all. Shoot, even 4xaa/16xaf isn't terrible. although SLI does alot better at those settings.


BTW, I'll be doing SLI and single 8800GS vs GT benchies sometime. The cheap fry's evga SC came so now it's time to see what $200 worth of 8800GS's can do. Oddly, the fry's one has the red/orange/yellow sticker like on the box (vs the green stripes that look like my kids could cut out of electrical tape and stick on) I'll take a pic sometime of the two together.

 

[pauldh]

Not sure what game areas of Oblivion Driverheaven tests. Typical for them is a combination of long gameplay from many areas averaged together.

Look at a 680/1600 clocked 8800GS at 19x12 max details 0x/16x:
http://www.driverheaven.net/reviews.php?reviewid=525&pageid=14

52 average is higher than I'd expect, but keep in mind that is many areas averaged together, hopefully one of the areas are deep lush foliage. Too bad they don't separate the areas like FiringSquad does.

I get really into benching/comparing so i don't do real long fraps runs. Typical is 1 minute give or take (45 sec to 65 sec for my 3 oblivion runs). I do 3 areas of Oblivion, and 3 runs of each test to assure accuracy (I use the median run in charts). I then did 3 resolutions and did 0x/1x, 0x/16x, 2x/16x, and 4x/16x. So for just that XFX 8800GS, I did 108 good runs. I then did this with 3 cpu's. (A64 4000+, X2 4200+, X2 5600+) SO that is 324 Oblivion runs with that one card! I am cutting this number down (knocking out 0x/1x and maybe 2x/16x as it's killing me doing so much) for the other cards and only using the X2 5600+. I just wanted to start with an in depth look at cpu perfomance with the first card. Still will be 54-81 Oblivion tests per card * 10 cards. Then there are many games to bench, some I skimp and go timedemos or I'd never finish. You can see why it's taking me a long time though to do this 3 cpu review followed by a 10 gpu review.

 

[marvelous211]

324 oblvion runs. Oh god you're a benching slut.

I went to area with the most trees and grass everywhere in chorral area. It hovers around 40-45fps with fraps. The card doesn't even flinch with oblivion at max settings.

I'm going to install karl texture pack. That might make it choke some.

 

[pauldh]

Admittedly yes. Sad thing is I enjoy the torture and few people if any(besides me) would appreciate such an in depth set of results.

And I am struggling to organize all the data and post it. The forums and image shack will not cut it so I may throw it onto a self made quicky website. I may have to just do a large data table like expreview does. Otherwise I'd end up with fudzilla's cramped charts, or way too many charts. I need two weeks off work to cram through this thing or it will be a summer long project.

Like I said, that was just for the first card I wanted a very in depth look. That testing is all done now. Now I am onto the next card and knocking down to the most useful compromise of tests. It also was a new build, so it took a while to install/patch all games and setup the benchmarks.

 

[pauldh]

Oh, yeah mods will effect it. Test it before and after will ya?

My benchie system is pure Oblivion, no mods at all. My gaming system has many mods and does knock performance down for sure.

 

[marvelous211]

You can reinstall it if you want but i just finished a review with a skulltrail board and a 9800GTX and that wasnt doing it so i know your rig cant.
Mactronix

How was it not doing it? Perhaps there was something wrong with your system or testing with mass amounts of AA and high resolutions. Where is this skulltrail board?

Paul's driverheaven link says it's getting over 50fps at maxxxed settings on a 8800gs that is clocked lower than mine at @ 1920x1200.

http://www.driverheaven.net/reviews.php?reviewid=525&pageid=14

 

[marvelous211]

Admittedly yes. Sad thing is I enjoy the torture and few people if any(besides me) would appreciate such an in depth set of results.

And I am struggling to organize all the data and post it. The forums and image shack will not cut it so I may throw it onto a self made quicky website. I may have to just do a large data table like expreview does. Otherwise I'd end up with fudzilla's cramped charts, or way too many charts. I need two weeks off work to cram through this thing or it will be a summer long project.

Like I said, that was just for the first card I wanted a very in depth look. That testing is all done now. Now I am onto the next card and knocking down to the most useful compromise of tests. It also was a new build, so it took a while to install/patch all games and setup the benchmarks.

Maybe we should open up shop and make a website of our own? Play with the big boys and get paid free hardware.

 

[angry_ducky]

I love how this thread became first a review of the 8800GS, and then became a detailed rundown of pauldh's Oblivion benchmarking methods.

While your dedication to accuracy is impressive, I think that in the end, the average user doesn't really care about FPS; they're more concerned with whether or not the gameplay feels smooth.

I also think that the whole bottlenecking issue is mostly psychological. The consumer has an idea of what they want to purchase, and then they are scared into believing that their current system will bottleneck the graphics card, and they won't get the most for their money, and then they'll turn around and buy a whole new CPU, motherboard, etc.

I also think that some people just upgrade or overclock because they can, because it makes them feel better to have the latest and greatest, when the only thing that they'll ever do is surf the internet and send email.

For example, when I overclocked my CPU, I never even bothered doing any benchmarks (only Prime95 stability testing); just knowing that my CPU was faster than before felt like enough of an accomplishment.

Just my daily rant...

 

[pauldh]

And what's wrong with carrying on the conversation OT when the OP's has been answered and orders a card? I'd gladly go back on topic if it needs further attention. The 8800GS needed to be defended as it was unjustly called out as a non-gaming card.

You are right on the average user. Yet again, the average user doesn't frequent enthusiast forums either nor understand fps is directly related to smooth gameplay.

Admittedly, I am a stickler for details and have the rare sickness of enjoying testing/comparing hardware. But I am opposite you. If I overclock, it's because I need to or want more performance. I then enjoy measuring the gain (if any) I received from the OC. It's a need to know.

 

[pauldh]

Maybe we should open up shop and make a website of our own? Play with the big boys and get paid free hardware.

Talking my dream. if I didn't need to work now, that's what part of my free time would go toward. Just trying to stay young; eventually the love will fade and way before I could ever afford not to work.

 

[randomizer]

All pci-e 2.0 cards are backward compatible.

That is true for all non-VIA based motherboards to my knowledge (it may just be the Asrocks actually). Don't try sticking an 8800GS into an asrock VIA board, it just won't work.

EDIT: Oh, and I laugh at the person calling out the 8800GS as a non-gaming card when the OP was using a 6800. Dude you ain't got a clue. I'd just like to say that I can run 1680x1050 at 16xAA on oblivion with my 9600GT, but 16xQ is completely out of the question. I still don't play with AA though, higher framerate is more important to me. If the 8800GS can do what the 9600GT can do, then 1680x1050 in crysis on high (just ingame "high") is completely playable. I average ~34FPS at those settings, but I haven't got to the ice yet with this card.

 

[marvelous211]

Talking my dream. if I didn't need to work now, that's what part of my free time would go toward. Just trying to stay young; eventually the love will fade and way before I could ever afford not to work.

Let's stop taking and actually put our words into action. This is what I love anyways and bringing info to the mass.

 

[royalcrown]

I understand the frustration. In his defense, I often question what people claim because not everyone like us cares to be accurate. I think alot of careless (maybe not intentional) exaggeration takes place here. Instead of using fraps or checking fps, they guess it's smooth. They also may test one little part of a game, or run a benchie, and then say it's smooth in the game. Also some people want to make claims about their system likes it's their child, claims such as playing Crysis on very high 19x12 above 30fps on a single GPU. It's impossible and every review backs that up. I tend to call those out with links just to keep things in check so people aren't misled.

But a 3.5GHz C2D and a 700+/1700+/ 2000+ GS would easily play oblivion at 16x10 max no aa, never mind 14x9. I really can't guess the exact fps, I just know it would be very playable throughout the game.

My 320MB GTS hurt bad at 16x10 in Oblivion, but that was paired with a single core FX55. I had to tweak things down a bit. The CPU makes a difference even in the outdoor sections. But in my 3.0GHz Q6600 setup, without fsaa a single 8800GT can handle 16x10 Oblivion no problem at all. Shoot, even 4xaa/16xaf isn't terrible. although SLI does alot better at those settings.


BTW, I'll be doing SLI and single 8800GS vs GT benchies sometime. The cheap fry's evga SC came so now it's time to see what $200 worth of 8800GS's can do. Oddly, the fry's one has the red/orange/yellow sticker like on the box (vs the green stripes that look like my kids could cut out of electrical tape and stick on) I'll take a pic sometime of the two together.

Hey now - I wasn't gonna pirate crysis and play it just to bench it...lol.


...and as far as smoothness, that is a subjective term that cannot be quantified, so all the fraps in the world won't help...smoothness may be 24, 28, 30, or 10,000 frames per sec..so

I know I was in one of those 2 rants ...lol

And Crysis can be smooth in dx10 at 1024 x 768 on very high...hehe dunno about 1440 x 900 tho...

 

[angry_ducky]

And what's wrong with carrying on the conversation OT when the OP's has been answered and orders a card? I'd gladly go back on topic if it needs further attention. The 8800GS needed to be defended as it was unjustly called out as a non-gaming card.

You are right on the average user. Yet again, the average user doesn't frequent enthusiast forums either nor understand fps is directly related to smooth gameplay.

Admittedly, I am a stickler for details and have the rare sickness of enjoying testing/comparing hardware. But I am opposite you. If I overclock, it's because I need to or want more performance. I then enjoy measuring the gain (if any) I received from the OC. It's a need to know.

I was actually trying to get more performance out of my poor little Athlon 64, as it feels a little [strike]very[/strike] sluggish using Adobe Premiere (the only game I play right now is Portal), but I couldn't get it over 2.54 GHz, and that was with the RAM at DDR266 and 3x HTT.

At that speed, it didn't feel any faster than stock, so I didn't even bother benchmarking, and just set it back at stock. I'm hoping to get a new system soon, or at least get a new video card and RAM (I want a Quadro which does hardware MPEG encoding and 2GB of nice DDR400, as my memory is probably limiting my overclock).

 

[pauldh]

Hey now - I wasn't gonna pirate crysis and play it just to bench it...lol.


...and as far as smoothness, that is a subjective term that cannot be quantified, so all the fraps in the world won't help...smoothness may be 24, 28, 30, or 10,000 frames per sec..so

I know I was in one of those 2 rants ...lol

And Crysis can be smooth in dx10 at 1024 x 768 on very high...hehe dunno about 1440 x 900 tho...

LOL, It wasn't you and the crysis demo talks I was referring to, but rather a 19x12 very high crysis claim I remember. It's nothing new, there were claims a 7800GT ran Oblivion maxed also ages ago. Oddly my 7800GT like the ones in reviews was terrible in Oblivion. But hey if the shoe fits, and you want to wear it, .... :lol:

 

[rado_88]

Ok so today I'm getting the card. One more thing. Is it power hungry? I got a 410W Chieftec power supply. Will it need more than that? Sorry I'm asking so much questions, but I'm kinda 2 years behind with pc's

 

[rado_88]

I bought the card. It's working well, got almost 9000 in 3dmark 2005. Gainward recommends a 450w power supply for it, but it performs on a 410w too.

 

[pauldh]

Enjoy the card. If you have stability issues, it could be the psu. The important spec is the output of the 12v rail(s).