Tom's Hardware Charts: 2009 Mainstream Graphics Update

Two Settings For Graphics Cards

To enable comparisons for the High and Low Settings tests, we tested a batch of graphics cards for both categories. We chose the Radeon HD 3850 and HD 4670, along with the GeForce 8600 GTS and 8800 GTS 320 models for this head-to-head. While the Radeon HD 4670 works well with DirectX 10 and our higher-end settings, the GeForce 8800 GTS proves something of a problem child. With a game's graphics quality slider set to high, this card’s 320 MB of RAM is improperly optimized, very slow, and unusable at 8x AA. When Low Settings are in effect, this card is much faster, if not under-utilized in light of significant CPU limitations.

Our small test table shows total results in frames per second (FPS) and should shed some light on differences and distinctions in performance. The GeForce 8600 GTS works well at lower graphics quality settings, but when asked to deliver higher graphics quality in a DirectX 10 environment, it just barely limps along. This also applies to the Radeon HD 3850, which is designed more for use with DirectX 9 and somewhat lower graphics quality settings. The Radeon HD 4670 functions well in both categories and is noticeably stronger and more capable as graphics quality increases. In fact, it’s surpassed only in the Low Settings tests by the GeForce 8800 GTS 320 MB.

Overall Performance Results
Low Settings
High Settings

GeForce 8800 GTS (320 MB)

3,247.8 FPS

533.1 FPS

Radeon HD 4670 (512 MB)

2,458.6 FPS

608.5 FPS

Radeon HD 3850 (512 MB)

2,281.1 FPS

454.7 FPS

GeForce 8600 GTS (512 MB)

1,737.8 FPS

384.9 FPS

In the real world, you’d probably boost the graphics quality slider somewhat for the GeForce 8800 GTS 320 MB. For our charts, we were forced to use a setting with medium graphics quality, which didn’t really pay off because it makes the basis for comparison with other chipsets questionable. We also include 3DMark06 tests in our suite, which are independent of graphics chipset speed because we always use the default graphics quality at a 1280x1024 resolution. This too provides a direct basis for comparison across all tests.

Create a new thread in the US Reviews comments forum about this subject
This thread is closed for comments
    Your comment
  • no stalker cs, wth. thats a new graphically demanding title.
  • hmmm ... weird choice of options ...

    I would consider low as medium detail settings with no AA/AF up to 1680x1050 and 1900x1200 at low again without AA/AF ... any card that cannot meet this at playable fps is HTPC material at best.
  • also can you please PLEASE finaly implement multiple criteria selects ?

    I just wanted to have a look at the 9800GT in all the benchmarks at 1900x1200 no AA/AF. however I can either select the cards or only the benchmark for all cards. any fine tuning is not possible.
  • Shouldn't the mainstream segment go a little past the 8800GTS and the HD4670 at this point?
  • Exodite3Shouldn't the mainstream segment go a little past the 8800GTS and the HD4670 at this point?

    The 4670/8800 are still powerful cars and will meet basic gamer needs. Hell, fallout 3 at high is playable for me on my pentium D, so what more do I need? (HD4670 underclocked by the way.)
  • anamaniacThe 4670/8800 are still powerful cars and will meet basic gamer needs. Hell, fallout 3 at high is playable for me on my pentium D, so what more do I need? (HD4670 underclocked by the way.)

    I also think the 4670, 9600GT, and 3870 are proper mainstream cards. The Old 8800GTS 320MB for me is a little bit questionable though.
  • Wow, great article; affirming and eye-opening. It affirms what I've thought for a long time, that surely many games are quite playable on cards like the HD4670. For players interested in the mechanics of the gameplay and/or the story line, this card is entirely suitable. I would like to have seen the HD4650 on the charts also, do you have benches for it?
    Eye-opening too, in that I can see why those who absolutely must have the eye candy, and might not care about other aspects of the game, want to spend $500, $600, or even more on graphics cards (and a PSU to support them!). While I hope they earned that money themselves, I can see much more clearly why they want to spend it.
  • it would be nice to know the reference system the cards were tested on. Unless I missed it someplace I didn't see it listed.
  • I'm currently running a Core2Duo 6600 (2.4Ghz) with an old 8800 GTS 640,

    up to this point I have yet to find a game it cant handle at my resolution of 1680x1050, which I would consider to be the mainstream resolution for gaming.

    So using the 8800 series, 4670 series is more than adequate for mainstream as far as I'm concerned.

    Moving into the 1900x1200 and above resolutions, the 9800 / 4850 series would probably be better suited.
  • Might want to update the cost of the 4890. Newegg has one on sale now for $180 with a list price of $200.
    The 260 seems a little high too.
  • I don't really the meaning of this article.
    Where is the chart?

    This look like a study report who refine all term before given the result.
    But with no result at the end.
  • to add a little to the "you don't need monster GPU's"
    i played Fallout 3 on a X1300 Pro not that long ago. Far cry 2 was the only game it could not play anymore. that's why i moved to an HD 4670 which lets me play all games, mostly at highest setting, sometimes even with AA. i have to admit that i play on 1280*1024 but still immpresive for a budget card from January 2005. it even sported acceptable framerates.

    The HD4670 is even better, a recomendation i would have given to everyone. not anymore though as the prices of the 4850 are so low at the moment and it IS a better card.
  • Last Remnant? Seriously?
  • btw, which mod boots fallout 3 graphics to almost photo realistic? Leaving it up to me to search for it amongst a sea of mods...
  • I think, in general - the chart was problematic before, and now its even a bigger mess than ever before!

    The BASE model GPUs are fine. Any gamer who knows ANYTHING about the cards will know that there is a bit more performance with OC. But most people buy 8600s are not as interested in OC as someone who buys a GTX 285... so maybe 1 or 2 OC cards for reference.. but over all, its not

    By separating the bottom end cards out from the top, you guys (THG) make it more difficult for low-end game card owners to visualize the difference between a $40 card and a $400 one.

    We DON'T need 4 versions of the GTX 285! What are you guys doing, Advertising in the Charts NOW?! Between these 4 GTX 285 cards, the that the slowest is 90.2fps and the fastest is 91.0fps!

    The chart is harder to find what you want, the text is tiny with mesess off the various brands and their model numbers.

    The chart has 15 Brand specific cards that show nominal differences from the reference cards at stock speed.


    Here is an exmaple of junk:
    Sapphire HD4850 1G
    (HD 4850 1024 MB)

    Its no faster than the stock card. Make the chart better by saying:
    ATI HD4850 1024 MB (i)

    Make the Card type BOLD, tad bigger font. Memory size the font you use not. the (i) can be a graphic or word for "Product details". That's it.

    The colors should be:
    RED = ATI
    Green = Nvidia
    Dark Red = ATI Cross Fire
    Dark Green = Nvidia SLI

    No blue, no name brands.

    Include pretty much ALL current cards - so we can use them for reference... which is THE POINT of the chart.

    For some older cards, include base 8x & 7x series and HD 2 series.
    Because some people still use older cards, just include a handful of popular cards as long as they're PCIe like a X1900, X1600 and a 6600GT.
    And to really help out people, as horrible a they are:
    IGPs like an Intel GMA and and ATI & Nvidia (1 of each from current boards)

    Older or lower cards:
    GTX 2* (all of them)
    GTS 250
    GTS 150 (Find OEM card so a 150 owner knows how they scale)
    9800 GX2
    9800 GTX
    9800 GT
    9600 GT
    9500 GT
    9400 GS
    8800 Ultra
    8800 GTX & SLI
    8800 GT 512
    8800 GT & SLI
    8800 GTS 640
    8600 GT & GTS
    8400 gs
    8200 IGP
    7900 GT
    7600 GT
    6600 GT
    6150 IGP (if the 8200 IGP is any faster)
    5200 (There is a PCIe version - a very popular useless card)

    48x0 cards - ALL (CF on 50/70/90)
    4670 + CD
    3870 + CF
    3850 + CF (And the X2)
    3200 IGP (On board)
    2600 XT
    2600 Pro
    2400 Pro
    1900 XT (A single 1900 is fine)
    1550 (Current Low profile)

    Intel Onboard.
  • Oops, I forgot to change title to Nvidia for that group... I was on a roll.

    Currently THG has 67 total cards, 50 in the "high" end area. None of them in SLI/CF mode (in their own chart with direct comparisons with their single card variants is good)

    In my list above, 28 cards (none in SLI), 20 ATI and and a single Intel = 49 cards. It would be CLEANER than what we have now and have a bigger variance than have 2~4 cards that are the same here and there.

    Find a GTS 150... someone should have one. It should be on par with the 9500GT... and when a owner of a 150 comes to the site, they can see where they are on the list.

    Don't need both versions of the original 8800gts.

    I know the 6150, 8200 and 1550 are very low end, but they are used in lots of computers and people do ask "Why do my games suck on my $800 computer"? This will show why.

    Please clean this up.
  • I personally am not amused to see brand names on a chart that should give us objective information and not put 'some' brands on it with their sooped up cards. Any one that would buy those knows how they perform.
    It's sad to see that it isn't what it used to be and too bad not someone like belardo, who does seem to understand how it should be done, works here...
    Big miss here although the games are better chosen I think. More differences in engines and waiting for the next crysis lookalike it is what engines are used now for gaming.
    Though I must say that stalker should be in it for the dx 10.1 benches too. I don't agree with it being unfair to nvidia that does not have dx10.1, hell there slacking on the job! Too bad they don't have it but let's not complain then when their benchresult suck in comparison to amd/ati's offering...
  • What I did notice is how well that rusty trusty old 8800 ultra is doing! Man it beats the sh*t out of most mainstream cards that are a full YEAR younger... Should say something about nvidia and working on getting better results outside of the gtx 400 + regions...
  • Why isn't the radeon 4770 crossfire setup on any of the charts?
  • [nobody’s going to equip an overclocked $1,000 PC with a $50 graphics card to play 3D games.]

    Some things to add:
    -When My Athlon X2 died, I upgraded to an i7 system, but conserved the old Gf8800GT, because i don't want to waste money until DX11 cards are released.
    still I would like to see a high end CPU with a low end card, to know the difference to a GPU upgrade.
    -Many dudes ask me advice on what system to buy. If you publish a high end CPU with cheap card, I can point my arguments agaist such purchase, by showing your benchmarcks. If such rig does not make sense, still is useful to check it.
    -Frequently somebody don't need a gaming PC, so buys a powerfull CPU, but then needs to add a video card, and maybe use the rig for some gamming, so the questions is: how much money it takes an i7 to do basic gamming? what is the difference to a better card?