Performance Tests

We ran benchmarks on the 7800 GT with the same system that we used for the 7800 GTX tests:

MSI K8N Neo4 Platinum/SLI motherboard

AMD Athlon 64 FX-55 Processor
1 GB OCZ 2:2:2:6 DDR400 RAM
Seagate 7200.7 120 GB Hard Drive
OCZ 600 W PowerStream Power Supply

We'll also be testing the card's performance in the three games that we've been using: Battlefield 2, Doom 3, and Halflife 2. We'll be testing all three games at 1600x1200 with and without AA. We chose this as our resolution because it reflects what is most used by people who get a 7800 GT, given that most cost-effective LCD monitors still don't support resolutions much higher than this.

Since this is our first 7800 GT article, and we don't have any other versions to compare the XFX 7800 GT with yet, we'll be looking at how the GT compares to the 7800 GTX reference card. Because we were able to overclock our XFX 7800 GT a little bit higher than the reference 7800 GTX, we can see how significant it is to have one more vertex and 4 more pixel pipelines.

Battlefield 2 Performance

Battlefield 2 Performance

Doom 3 Performance

Doom 3 Performance

Half-Life 2 Performance

Half Life 2 Performance

As the numbers show, the reference 7800 GT framerates are quite a bit lower than the 7800 GTX, particularly with Battlefield 2. This is because Bf2 is a fairly new game and it places a higher stress on the graphics card at 1600x1200 than the other two games. In Bf2, there's about a 20% decrease in fps between the reference 7800 GT and GTX both with and without AA enabled. Doom 3 and Halflife 2 saw a 14% and 18% decrease in fps respectively with AA enabled, but there wasn't nearly as much of a drop in those two games with no AA.

When we tested the XFX GeForce 7800 GT Overclocked, we found the numbers to be much closer to those of the 7800 GTX. On the 7800 GT, BF2 still has about a 10% decrease in fps from the GTX, which might be noticeable during game play. But the other differences were small enough not to make much difference.

As you can see when we overclocked our 7800 GT to 462MHz and 1.2GHz, we managed to match and even slightly surpass (in Doom 3) the framerates of the GTX. This is significant, but we aren't quite sure if we'll be able to see the same results on other 7800 GTs. These numbers really show us how there isn't a very big difference at all in G70 performance with one less vertex and four less pixel pipelines at 1600x1200 and lower; all the more reason to save $100 and go with the GT if a very large expensive LCD isn't in your computer room or within your budget.

Overclocking Power, Heat & Noise
POST A COMMENT

33 Comments

View All Comments

  • Leper Messiah - Thursday, September 1, 2005 - link

    I am SO buying one of these to upgrade my 9800np. The thing is starting to die in warhammer 40K and BF2, and my resolution sweet spot is 1600x1200, which means I can max everything with this card and run fine.


    I cool my 9800np with a zalman VF-7000 Al-Cu, or whatever it is...you know what I mean. I wonder if that will fit on there, would be nice to see a 480/1300 or something with my BGA ramsinks...
    Reply
  • blckgrffn - Friday, September 2, 2005 - link

    Apparently the 7800GT cards are too cheap to warrant Nvidia requiring a temperature sensing fan like the 7800GTX cards. How about that.

    *owner of a BFG 7800GT from Dell for $345 w/25 pack spindle of CD-R's :)
    Reply
  • NullSubroutine - Wednesday, August 31, 2005 - link

    Do we have to request 1280x1024 benchmarks still? I have to believe that alot of us out there have 19" lcd's that run native(max) at this res, so we dont really care what the speeds are at 16x12. Reply
  • n7 - Thursday, September 1, 2005 - link

    NullSubroutine: There is a reason why AT doesn't bother testing at 1280x1024.

    With a 7800 series card, nearly all games are CPU-limited, meaning there will be basically no difference in performance, since the CPU is holding everything back.

    Something many people fail to compregend is that the 7800s aren't really necessary if you don't have a 1600x1200 monitor or higher.
    Reply
  • Rocket321 - Wednesday, August 31, 2005 - link

    I think reviewing 2-3 cards per article would be best. That way we don't wait weeks for a huge roundup, and we also get to see the aspects of multiple cards at the same time. I don't mind the individual reviews either, but the first one is always less useful and the last one tends to be the most useful due to having more comparisons.

    --Rocket321
    Reply
  • Houdani - Wednesday, August 31, 2005 - link

    I'm city folk, and don't have an ounce of cowpoke in me, but that won't stop me from expressing an opinion of what I think life is like out on the range.

    When herding cows, the preferred method is to gather them all up and move them as a group. In other words, ROUNDUP! Sure a stray bovine might wander off, in which case you may ride out and bring back that individual, but generally speaking it's best to round all the cows up en masse and move them as a collective group.

    Likewise, when herding video cards, the preferred method is...[wink]
    Reply
  • UltraWide - Wednesday, August 31, 2005 - link

    How come the 7800GT is so noisy compared to the 7800GT-X.

    I did notice my BFG 7800GT is too dang loud. Maybe I should return it and get a 7800GTX???
    Reply
  • Marlowe - Thursday, September 1, 2005 - link

    A friend recently bought a Gigabyte 7800 GT and that's very noisy as well! Looks like the same cooler. Mounting a Zalman vf700cu would fix it. Anybody got cards from other brands? Post in! Reply
  • photoguy99 - Wednesday, August 31, 2005 - link

    Is there any conclusion on if this is true?

    Didn't see if the benchmarks used the driver that only reviewers have.

    According to the "reliable" source below, best case is you will lose 3% performance in high quality modes.

    http://www.theinquirer.net/?article=25820">http://www.theinquirer.net/?article=25820
    Reply
  • Live - Monday, September 5, 2005 - link

    I think Derek and/or Josh really should write something about this. Since it really invalidates the HQ benches and puts questions on the review process here at AnandTech. Why wasn’t this spotted when the original reviews came out? If ATI has something like this in there upcoming products I would sure like to know when I read the review here at AnandTech.

    Hexus has a write up and an apology.

    http://www.hexus.net/content/reviews/review.php?dX...">http://www.hexus.net/content/reviews/review.php?dX...
    http://forums.hexus.net/showthread.php?t=53394">http://forums.hexus.net/showthread.php?t=53394
    Reply

Log in

Don't have an account? Sign up now