7900 GTX sli anyone doing it?

Darkwave · 9 May 2006 at 11:34

Just ran it myself in 1280x960 with full settings, HDR enabled, 4x antialiasing and 16x anisotropic filtering (not HQAF, I turned it off since it started protruding textures for some reason.)

Frames, Time (ms), Min, Max, Avg
481, 16667, 22, 40, 28.859

FPS: 24, 24, 25, 27, 28, 28, 28, 25, 26, 26, 30, 32, 32, 31, 31, 37

Bit more demanding than usual since I've been playing with grass tweaks to make it look even prettier, but still gives easily playable performance as always, even with antialiasing and HDR enabled at the same time.

easyrider · 9 May 2006 at 13:05

I'm not getting another 7900 GTX I have decided.
Not worth it. :eek:

fornowagain · 9 May 2006 at 13:30

egt said:
I was going to suggest this too. Problem is, the results can be "doctored" too easily and we need ppl's honesty for it to work. As we've already seen, there's at least one poster who's not too good with the truth.

Here's my results at 1280x1024 16HQAF+0AA everything max settings and no INI cheats:

2006-05-09 07:03:19 - Oblivion
Frames: 357 - Time: 16741ms - Avg: 21.325 - Min: 13 - Max: 36

2006-05-09 07:03:45 - Oblivion
Frames: 338 - Time: 17513ms - Avg: 19.300 - Min: 12 - Max: 31

2006-05-09 07:04:23 - Oblivion
Frames: 325 - Time: 16996ms - Avg: 19.122 - Min: 11 - Max: 34

2006-05-09 07:04:52 - Oblivion
Frames: 479 - Time: 16829ms - Avg: 28.463 - Min: 11 - Max: 57

Take note of my last benchmark, thats doing the run while looking at the sky which gives a false reading!

First, let me check this. Everything in CCC to max 16xAF HQ on, AA off, AI standard, Mipmap-High Quality. In game every slider to the right, all shadows on. No ini tweaks. Load the save file, auto walk and start Fraps (just miss the tree) and turn off Fraps (ver 2.7.2) just as you hit the rock. Don't touch the mouse at all.

I'll quote your results to compare, but my XTX at 680/1600 1280x1024 16AFHQ 0AA+HDR

2006-05-09 12:57:53 - Oblivion
Frames: 626 - Time: 15660ms - Avg: 39.974 - Min: 31 - Max: 52

2006-05-09 12:59:00 - Oblivion
Frames: 631 - Time: 15602ms - Avg: 40.444 - Min: 32 - Max: 50

2006-05-09 13:00:04 - Oblivion
Frames: 613 - Time: 15493ms - Avg: 39.566 - Min: 32 - Max: 49

2006-05-09 13:01:14 - Oblivion
Frames: 609 - Time: 15480ms - Avg: 39.341 - Min: 31 - Max: 49

What's different? everything is set to max so I'm at a loss. And I can assure you I'm not economical with the truth. Your results have exactly half the frames when you take the time per frame.

P.S. This is not intended as flame/fan bait. I'm not going there, it'll end in tears.

fornowagain · 9 May 2006 at 15:11

EGT, Did you put HDR on?

I was also looking at the minimums as well, yours drops to 50% ish, mine drops to 75%, again I dont know why. Any ideas?

fornowagain · 9 May 2006 at 15:13

Exsomnis said:
Just ran it myself in 1280x960 with full settings, HDR enabled, 4x antialiasing and 16x anisotropic filtering (not HQAF, I turned it off since it started protruding textures for some reason.)

Frames, Time (ms), Min, Max, Avg
481, 16667, 22, 40, 28.859

FPS: 24, 24, 25, 27, 28, 28, 28, 25, 26, 26, 30, 32, 32, 31, 31, 37

Bit more demanding than usual since I've been playing with grass tweaks to make it look even prettier, but still gives easily playable performance as always, even with antialiasing and HDR enabled at the same time.

Can you try that again without the AA and +16AFHQ at 1024 so its apples to apples?

Darkwave · 9 May 2006 at 16:04

fornowagain said:
Can you try that again without the AA and +16AFHQ at 1024 so its apples to apples?

When deciding wether or not to use AA with HDR, I didn't actually notice much of an FPS drop with it enabled, and I don't remember what exact frames I got with HQAF enabled. My benchmark results were mainly to show that Oblivion does indeed run fine with really high settings on a single card.

fornowagain said:
What's different? everything is set to max so I'm at a loss. And I can assure you I'm not economical with the truth. Your results have exactly half the frames when you take the time per frame.

Disclaimer: This is not intended as fanboyism or flamebait of any kind either, and if any argument results from this post then it will be totally ignored.

Now that's out of the way... I, and a couple of others, have asked egt about his Oblivion performance before since it does indeed seem oddly slow on his system, but these "discussions" usually descend to the level of hostile flaming extremely quickly. I think that egt has taken my past comments on his Oblivion performance as flaming, when I actually hope he gets it fixed and would gladly help out if I could.

There's no doubt about it though, comparing the Fraps reading in this thread is enlightening indeed. I'll do a 0xAA/16xHQAF run when I reinstall the game, I've been messing around and swapping parts out to play with clocks and did a clean install of Windows.

KarlMcWade · 9 May 2006 at 16:45

just thought i jump in here I,m running a dothan at 2.7 with 3 gb ram and a x800xtpe water cooled 560 core 580 mem
and it runs fine at 1280 1024 maxed detail and bloom lighting
i,m not cpu limited

egt · 9 May 2006 at 16:46

fornowagain said:
EGT, Did you put HDR on?

I was also looking at the minimums as well, yours drops to 50% ish, mine drops to 75%, again I dont know why. Any ideas?

Yes, HDR on, all graphics settings max but no AA.
AI High:

2006-05-09 16:36:22 - Oblivion
Frames: 406 - Time: 16446ms - Avg: 24.687 - Min: 16 - Max: 44

AI off:

2006-05-09 16:40:04 - Oblivion
Frames: 343 - Time: 16215ms - Avg: 21.153 - Min: 13 - Max: 35

My results tie in with Gorods results as he is using an X2@2800Mhz which seems to help this game. Your results look freaky high, tbh m8.

fornowagain · 9 May 2006 at 17:06

egt said:
My results tie in with Gorods results as he is using an X2@2800Mhz which seems to help this game. Your results look freaky high, tbh m8.

Or maybe there is a common problem. I have a very similar system to Gorod and the game is very playable, which is why I've always wondered what the fuss was about and I tried his "demo". If I increase settings incrementally it all drops slowly, nothings missing, its all there, so I don't don't know why his/your frames are low. I've seen plenty of systems getting the same results.

The AI has a bigger effect than I expected. What it actually doing?

1280x1024 16AFHQ 0AA+HDR All settingd to max.

AI off
2006-05-09 17:01:19 - Oblivion
Frames: 543 - Time: 16095ms - Avg: 33.737 - Min: 23 - Max: 54

AI standard
2006-05-09 17:02:44 - Oblivion
Frames: 644 - Time: 16249ms - Avg: 39.633 - Min: 31 - Max: 55

AI Advanced
2006-05-09 17:04:19 - Oblivion
Frames: 687 - Time: 16105ms - Avg: 42.658 - Min: 35 - Max: 55

egt · 9 May 2006 at 17:16

We need more ppl to to do this test, not as a willy waving contest, but to make an honest database

Gorods x1900 results:

Not same test, but results with an 512mb x1800 and lower gfx settings:

Darkwave · 9 May 2006 at 17:28

egt said:
We need more ppl to to do this test, not as a willy waving contest, but to make an honest database

I agree, it would be nice if we could get a spectrum of how different systems perform in Oblivion, because something frightening just occured to me... Could it be that Oblivion isn't entirely compatible with all hardware? I mean it wouldn't be the first time that a game has been buggy with a certain brand of botherboard or a certain speed of RAM, who knows.

Their site isn't working for me, but if I know EliteB*******, they're probably using DFi boards like you are for testing. That'd explain the oddly low HDR+AA results with the X1800XT if there was a compatibility issue.

egt · 9 May 2006 at 17:59

Another run with all gfx card settings dumbed down (max speed):

2006-05-09 17:48:26 - Oblivion
Frames: 516 - Time: 17243ms - Avg: 29.925 - Min: 22 - Max: 43

These are the same settings but looking at the sky (cheat):

2006-05-09 17:50:40 - Oblivion
Frames: 599 - Time: 16811ms - Avg: 35.631 - Min: 23 - Max: 54

KarlMcWade · 9 May 2006 at 18:19

Frames 2653 - Time (ms) 60000 Min 27 - Max 67 - Avg 44.21
taken walking from burma walking NE down mountain pass
2.7 dothan 3 gb ram
x800xtpe agp 560 core and 580 mem
1280 x 1024 max everything, bloom lighting, no AA or fsAA
10 month old xp pro install
47 proc in backround
6 apps running msn steam bitlord avg ati tool fraps ms word

egt · 9 May 2006 at 18:48

Reputable site benchmarks here (Anandtech ). Please note these results are on a A64 FX60 with in-game settings dumbed down so not as stressfull as Gorods test:

Fornowagain, impressive stuff, you're beating x1900xt Crossfire results!

fornowagain · 9 May 2006 at 19:08

Wow, the lengths.

Just to keep it honest. Apprently.

1280x1024 16HQAF+0AA everything max settings and no INI cheats.

2006-05-09 18:56:16 - Oblivion
Frames: 795 - Time: 20233ms - Avg: 39.292 - Min: 30 - Max: 61

18.8Mb video

fornowagain · 9 May 2006 at 19:12

egt said:
Fornowagain, impressive stuff, you're beating x1900xt Crossfire results!

I'm lying then am I? It does what it does. Without running the exact same path it pointless putting up sheets from all over. All you have is your results versus mine, I'm not lying or cheating. Nor are you I assume, so something is different.

Do you get 50's in town? I do, seems about right to me.

speeduk · 9 May 2006 at 19:13

Here are my results if anyone cares.

Opteron 144 @ 2.8ghz
2gig ram
x1900xt @ 670/780

1680x1050
16xHQAF
No AA
Bloom

Slowest outdoor environment (mixture of some npc's, a building and lots of trees)

Everything max in the options.

Min 19fps avg 23 max 27

Trees @ 1/2, no self shadows or on grass. Ini tweak to reduce grass density by 20%.

Min 28 avg 34 max 41

Anyone trying to play oblivion on max settings is silly imo. A few ini tweaks and reducing the tree distance hardly affects I.Q at all and gives perfect framerates.

Edit - added HDR + 8XHQAF, *2xAA* 1680x1050 results.

Everything max in options

Min 21 avg 23 max 26

1/2 tree distance, no shadows on self/grass

Min 23 avg 28 max 33

HDR and AA are real FPS killers. In scenes buildings and trees the framerate drop is huge and in a normal scene with just some trees/grass etc the framerate is fine.

egt · 9 May 2006 at 19:28

fornowagain said:
I'm lying then am I? It does what it does. Without running the exact same path it pointless putting up sheets from all over. All you have is your results versus mine, I'm not lying or cheating. Nor are you I assume, so something is different.

Like i said, you're beating Crossfire results, maybe you should post your vid over at XS

fornowagain · 9 May 2006 at 19:40

egt said:
Like i said, you're beating Crossfire results, maybe you should post your vid over at XS

All I'd like is an explanation.

These sort of posts say all you need to know about this guy.

http://forums.overclockers.co.uk/showpost.php?p=7034860&postcount=123
http://forums.overclockers.co.uk/showpost.php?p=7018857&postcount=42

Anandtech said:
We see no benefit to SLI or CrossFire here, due to whatever limitation we're running into at these settings. What we will investigate in future articles is exactly what is causing this limitation; we would assume we're CPU limitated even though we're already running an Athlon 64 FX-60. That doesn't bode well for other processors, as there simply isn't much more we can throw at the game.

Their town graph has an XT beating an XTX? According to that graph an XTX is faster than crossfire. Its just silly looking at odd graphs. You can only compare the exact same path. I notice they use 6.4+chuck. I don't, mines just chuck all on his lonesome.

speeduk · 9 May 2006 at 19:57

Just tried the save game @ 1680x1050 16xHQAF HDR no aa

Frames, Time (ms), Min, Max, Avg
1839, 60000, 13, 40, 30.650

You do realise that the grass density in this area makes it the slowest part in the game though right?

**Heres with some ini tweaks (not going over the top still good I.Q) same settings.**

Frames, Time (ms), Min, Max, Avg
2498, 60000, 29, 49, 41.633