Why is the 9800 GX2 not on the list? I tried it earlier at a friends house and Oblivion got about 80 FPS!

Ok, would two 7950GTs run it on max/near max with good performance?

Beause NTK (the author) hasn't been here in a while to make updates.

It should. Probably up to a 1920x1200 resolution. Any resolution lower than that should be alright as well, but if the resolution is too low you lose the benefit of SLI.

I need help asap, someone! Will the graphics card ati radeon x1550 run the Goty Oblivion, as it isnt listed in the recommends?

Run? Sure. Just don't expect too much power out of that card. You're looking at low to medium settings with it up to maybe a 1024x768 resolution depending on your fps tolerance. It's a rather weak card.

If it is the PCI version of the card, it may struggle a little more than the PCI-E versions. But people have enjoyed the game on worse.

Great list. Should be sticky.

You must RP a necromancer in Oblivion.... Because you sure brought this topic back from the dead.

Hmm. Good idea for my next character.

Actually it was mentioned in another thread today.

Okay, as is normally my policy, I don't add cards until they're already on sale; last evening, NewEgg apparently started stocking GTX 200 cards. I'm not positive on the benchmarks, since unlike as was the case for most of 2006 and 2007, Oblivion has fallen a bit out of favor for use in benchmarking due to the difficulty of getting consistent results due to the utter lack of included benchmarking tools or timedemos. (same on you gamesas!) However, from what I can tell, if you use a resolution LESS than 1600x1200 or so, the GTX 280 is the new champion card. However, I noticed that in Crysis, it suffers an embarassing defeat across the board to the older GeForce 9800GX2, and even has a credible threat from the Radeon 3870 X2. The GTX 260 is notably crippled by comparison, and hence it actually appears to regularly fall BELOW the two dual-GPU cards.

Likewise, the Radeon 4850 apparently saw itself on sale for all of a few hours before being yanked; apparently someone didn't feel like waiting until the actual release date before selling them. Although the information is VERY sketchy here, given that the 4850 appears to be just an underclock of the 4870, (while the GTX 260 is outright missing 12-20% of the hardware in the 280) it would put it perhaps comparable to the 4870X2, as well as the GTX 260. Expect some places to switch around as supplies become better and better reviews become available.

Well, in the past, at least, because it went through iterations so fast (the "Will my computer run Oblivion?" topic went through 1-2 threads, or 200-400 posts, per DAY) it was seen as not much of a reason to sticky it, since it went too fast. Though perhaps a consildated thread on hardware might be in order...


Naw, this thread cannot die, so hence necromancy is not of real point.

At higher resolutions, the GTX280 should do better than the 9800GX2, thanks to its 1GB of VRAM instead of 512.

Anyway, http://www.anandtech.com/video/showdoc.aspx?i=3334&p=15 shows the GTX280 beating out the GX2 at 2560x1600 with 4xAA+16xAF, but losing to it at lower resolutions (and by quite a bit at 1600x1200 or less).

That's not always the case. The use of VRAM as far as scaling the resolution really starts to drop off. Plus, the way SLi/Crossfire works, part of the VRAM load from higher resolutions (i.e, the different layers) are combined BEFORE the frame portions are sent to the other card. So in that case, the higher raw theoretical power shows up better; this has been shown in games like Crysis, where at high resolutions the 9800GX2 utterly spanks the GTX 280.


Hmm... That's an interesting result. I suppose I'd have to go with that; largely what I'd been relying on is testing in OTHER games, which showed much of an opposite effect: as you raised the resolution, the GTX 280 started to fall behind.

I love the new category you've added.

Egad Batman!

Honestly though, with the 4850 having only been out a few hours before being taken down, are you sure you can put it into the "Egad!" category? If you're basing it off of previews, you could just as easily put the 4870 and X2 in with it.

Thanks. Since there were a number of cards that were substantially above the 9800GTX in performance, I felt it time to add another category, since the 3870X2 and 9800GX2 never really did fit in with the 3850 and 8800GTS 320; each category is meant to be CLOSE to about twice as potent as the one before it.


Well, I actually have a pretty solid idea of where all cards would fit well before they release... But in a tradition that has QUITE infuriated people over time, I've held off placing any card there until the very day that it appears in stores, even if by accident. I.e, it only appears on the list if there's a chance someone actually legitimately has their hands on it. Plus, the GTX 280 deserves having a few days for it to hold the crown, no?

[edit]Wait, I forgot that the 9800GX2 still beat the GTX 280 anyway. Ah well, so much for THAT!

I am disappointed in these new cards.

Only 30 FPS with QTP 3 and max settings (include 4xAA and 16xAF) on a GTX 280 at 1920x1200?

That svcks.

Guess I can forget about MAEVWD and QTP3 and dozens of more visual add-ons, let alone LODing every mesh.

You disgust me nVidia. So much for something "not loading Oblivion down." Dozens of combos could do it for the EGAD category.

I'd take the 30 fps with a grain of salt. Oblivion has no timedemo and there is no standard benchmark for it. Hence, benchmark values from different sites are not comparable, and it is impossible to tell how the values relate to real playability. A few sites use good benchmarking methods that show performance in the most difficult areas, your average fps overall would probably be about double. Other sites should ludicrously high fps.

The only way to get comparable numbers would be for us to devise a few standard tests that we would run on our machines under controlled conditions. Then we could post our numbers in a thread like this and make a true ranking, and we could also add comments if the numbers alone would be misleading. I.e. SLI/CF systems can score insanely high average fps but they can also have very uneven performance and at times can play less smooth than single-card setup (not to mention micro-stuttering and driver hell).

Luchaire has explained a nice way of benchmarking Oblivion in http://www.gamesas.com/bgsforums/index.php?act=findpost&pid=12380990, and I described an application of Luchaire's method at the end of http://www.gamesas.com/bgsforums/index.php?act=findpost&pid=12411472. Without such benchmarks we don't have any comparable values at all. Besides, having reliable reference values would make it easy for people to see how their setup fares compared to similar systems, with gross underperformance pointing to hardware b0rkenness or suboptimal setups (or simply a CPU that has been running at 1/3rd its normal clock speed for a couple of years, as was posted in this forum recently).

Just out of curiosity, where abouts would the 8800M GTS stand? I got a notebook for my birthday and it has this card. It is a great improvement from what I have been using (7600GT on my desktop,) but the driver support stinks.

Well, mobile cards are never as fast as the desktop counterparts and typically fall a step behind in the ranking of the card. If I had to guess, I'd say it probably has the performance close to a 9600GT or 8800GS based on the specifications of the card.

Better than I thought it would be. Thank you Tigs.

Just as a note, I've added the 9800GTX+ to the list; some benchmarks perhaps might suggest that it'd go over the hump into the "egad" category, but I remain skeptical, since it's core is a modification of the G92.

Also, apparently, http://www.tomshardware.com/forum/252026-33-radeon-4870-512mb-newegg. As such, I took it as an excuse to throw it up on the list as well. It appears that even with the new HD 4000 cards, Oblivion actually switches to favoring nVidia; the 4870 normally schools the GTX 260 in almost everything else for 60% the cost, and falls disturbingly close to the GTX 280 in spite of costing some 46% as much.

I'd also be a bit careful in trusting the benchmark results as an absolute figure; remember that no two people's testing in Oblivion will yield the same results given the lack of any scripted timedemos for the game, (note to gamesas: FIX THIS for Fallout 3) so their usefulness is limited to as a relative tool; you can compare it to other cards benchmarked at the same time to see how they compare, but if they got X number of FPS, chances are slim you'd get the same.

Remember, as I noted in the description:

• I'm uncertain if it's possible - I recognize that someone will be able to probably bog the cards down if they tried.
  
• I said "smooth performance" - 30 fps is smooth for Oblivion; this isn't Counter-strike, folks; you aren't going to instantly die should your framerate drop to 59 for one second. I was careful to choose my words, and not say, "60fps."

Also, given that different mods will change the performance strain at DIFFERENT AREAS, it should be possible to combine some and some added on ones would NOT actually increase the load noticably.

Yeah, I think you'd be able to get smooth performance with a GTX 280, even with QTP3 and MAEVWD, and a slew of other goodies, at resolutions of AT LEAST up to 1600x1200, x4 AA x16 AF, if not higher. Of course, this presumes the CPU is willing; it's not the video card's fault if you place the bottleneck on a different part.

http://www.anandtech.com/video/showdoc.aspx?i=3341
Look at ATi's new cards. They are beasts at there price points. Wow, even has comparable performance to the much more expensive Nvidia 200 series.

I am running a Radeon 1300XT pro in my Dell 8300, I beleive it has a 250 watt power supply, with 4gigs of system ram, and 512megs in the 1300, with the dual core PIV 3.0ghz processor..

Based on the http://www.gamesas.com/bgsforums/index.php?showtopic=797401, list.

Is there a better card than the one I am using now that will work with the power supply I currently have?

The X1300XT is essentially a rebadged X1600pro...if you were able to power that card, then you should be able to power the 7600GS just fine. The 7600GS is a very low power card and is a decent amount better than your card.

Don't forget http://www.pcghx.com/aid,648976/Test/Benchmark/Radeon_HD_4800_PCGH_Benchmark_Review/&page=10 4800 series review from pcgh.