Test Descriptions and Settings

Age of Conan

For Age of Conan, we opened up the graphics settings and clicked the high button. For added fun we went into the advanced page and upgraded to SM 3.0, though nothing else on this page was changed. We left AA off, as the graphics on this one were a bit tough already, and getting a good experience might require dropping down to medium settings. It's a shame that AoC doesn't allow us to test at 800x600. On an MMO, screen space can be pretty important, so we do understand the decision. Our tests consist of a swim towards Tortage Island and a quick run to the jungle. The test is a straight line run and very repeatable. We used FRAPS to record the average frame rate.

Enemy Territory: Quake Wars

We recorded a custom net timedemo for this one. All of our tests are done using the highest possible settings with 4xAA. The Doom 3 engine is getting a little outdated and just about everything can run it well. One of the more interesting issues is that we need to quit out of the game every time we change resolution or our scores get really crazy.

Race Driver: GRID

This game plays surprisingly well at lower resolutions on these cards. We used the ultra high settings, but disabled AA here. Some of the shaders this game uses really benefit from AA though, and the 4670 is actually able to handle AA pretty well on this one at 1280x1024. The test for GRID is the first straight (right into a wall) of a track. We start at the back of the pack and start FRAPS running as soon as we take off. FRAPS is stopped when we hit the wall.

The Elder Scrolls IV: Oblivion

This test has been with us in some way or another for a very long time. The game is DX9, and we use Ultra High default settings for this one. For our AA/AF testing, we force those options through the driver control panel. The test is a straight line run with FRAPS. Even though there's a little age on this game, it is still a great title to get people hooked on gaming. It's open and fun and it doesn't look bad at lower resolutions with the settings cranked up. And these little cards can deliver as well.

Crysis

Yes, the obligatory Crysis test. This one is done using Medium settings across the board and no AA. We were running in DX10 mode and 64-bit as well. Our test is the built-in GPU benchmark run 3 times, and our score is the average of the second two runs. The game does lose some of its luster under Medium settings, but it is still very playable. Depending on the card you have, you would want to spend some time seeing what settings you could get away with pushing up to High.

The Witcher

For The Witcher we crank everything way up. We would have done more AA tests with this one, but there is this annoying new trend of limiting the maximum resolution AA can be enabled on based on framebuffer size. Aside from being annoying in our testing, architecture has a lot to do with AA performance and framebuffer is sort of a naive metric to use. In any case, this is another FRAPS test, but we benchmark one of the early in-engine cut scenes

Test Setup
CPU Intel Core 2 Extreme QX9770 @ 3.20GHz
Motherboard EVGA nForce 790i SLI
Intel DX48BT2
Video Cards

ATI Radeon HD 4870
ATI Radeon HD 4850
ATI Radeon HD 4670 (512MB GDDR3)
ATI Radeon HD 3870
ATI Radeon HD 3850
NVIDIA GeForce 9600 GSO
NVIDIA GeForce 9500 GT

Video Drivers Catalyst Press Driver 8.9 Beta (Radeon HD 4670)
Catalyst 8.7 (Radeon HD 4850, 3850)
Catalyst 8.8 (Radeon HD 4870, 3870)
ForceWare 175.19 (9600 GSO)
ForceWare 175.16 (9500 GT)
Hard Drive Seagate 7200.9 120GB 8MB 7200RPM
RAM 4 x 1GB Corsair DDR3-1333 7-7-7-20
Operating System Windows Vista Ultimate 64-bit SP1
PSU PC Power & Cooling Turbo Cool 1200W
Enter the 8800 GS... err... I Mean The 9600 GSO Starting at the Low End: Radeon HD 4670 vs. 3650
Comments Locked

90 Comments

View All Comments

  • arturnowp - Wednesday, September 10, 2008 - link

    I really don't like the conclusion. You can always say spend same more. You have a video card for 79$, just add 20$ and get something faster. But how much faster? 20$ it's 25% more. Does 9600GT provide 25% more performance? What's powor consumption of 9600GT. Not to mention this card is simply much bigger. We're at 100$ but why not spend around 125-130$... I'm sure most buyers want add extra money just to have something quicker if it doesn't provide "next level" of performance. Also companies like Dell or even Apple with chose smaller cards for their's computers.
  • neomoco - Wednesday, September 10, 2008 - link

    we all know the problem comments about biased articles on anand
    i haven`t made one yet but the final words on this article are hilarious ...
    my opinion is the final words should have started with something like this :

    wooowww impressive card ... amazing price/performance ... highly recommended at its price ... it decimates everything nvidia offer ... same performance if not > as 9600gso at lower price ...

    whenever they said something good about this card(rarely)they imediatly put brackets and add something negative ex:

    "Unfortunately, that's a more difficult question to answer than it was with the higher end parts." -lol
    "The hardware does outperform the competition at the same price point (though that isn't saying much)" -hmm

    and much more ... i may not know too much but my opinion is this amazing card should have recieved a much better review.

    let me give you an example of a nvidia review article title ... i wont say wich one it was

    "NVIDIA GeForce xxxxx : The Only Card That Matters"
    and an article introduction
    "It's really not often that we have the pleasure to review a product so impressively positioned. The xxxx is a terrific part, and it is hitting the street at a terrific price.Whatever the reason for the xxxxx, we are glad of its existence. This truly is the part to beat in terms of value. "

    i`ve never seen something even close about amd products and they had great products so to me your articles seem a little biased but we already got used to it . maby im imagining things

    peace
  • RagingDragon - Wednesday, September 10, 2008 - link

    In this article, the reviewer pans the Nvidia 9500GT and 9600GSO even more severely than the AMD 4670. Also his reviews of the AMD 4850 and 4870 were extremely positive. So I don't think it's fair to say he's biased against AMD or in favour of Nvidia. However, he obviously has a hate on for all current < $100 cards... Nvidia's 9500GT is particularly galling - it's just a re-rehashed 7600GT! And the 9600GSO seems pointless, I just checked prices at a local online store and found EVGA 9600GSO cards costing more than their 9600GT cards.

    But I think the review is too harsh on the AMD 4670, which resoundingly beat everything else in it's price range, and it is a big step in the right direction. These cards don't do what I want (1920x1200 at high details settings), but that doesn't mean they're junk, just that I'm outside the target market. While they offer little value to me, they should appeals to others with different needs/wants.
  • pattycake0147 - Wednesday, September 10, 2008 - link

    Did you read the 4870x2 review? It definitely had and bias against the 4870x2.
  • Loknar - Wednesday, September 10, 2008 - link

    Anandtech is not Pro-Nvidia, if that's what you want to imply.

    I remember the days of the Radeon, when the likes of TomsHardware was still drooling over Geforce2, and Anand chose to painfully explain the issue of image quality - which other reporters were too lazy to attempt. Same goes for the difficult and technicalities detailing the superiority of the Athlon XP over the Pentium 4; Anand took the rough route when other sites found it easier to say "Pentium is awesome, dude".

    You should consider the 'bias' in some articles is in fact "enthousiasm" about the new product/technology - which makes for a more fun-to-read article than blog-like constant bickering.
  • toyota - Wednesday, September 10, 2008 - link

    this is a GREAT card for oem comps. its small and the user can stick with the stock power supply and get a massive increase in fps over integrated graphics. plus these cards will probably be just $50 in a few weeks.
  • drfelip - Wednesday, September 10, 2008 - link

    I performs better than a 3850 and uses less power. When I need to upgrade my 3450 I think I'm going for a 4670. As you can see I don't need much 3D power, though.
  • needystevie - Wednesday, September 10, 2008 - link

    Does this card support hybrid tech?
  • scruffypup - Wednesday, September 10, 2008 - link

    AR, a 3870 can be had for $90-$100
  • toyota - Wednesday, September 10, 2008 - link

    well the 4670 is only $80 MSRP and will likely be much cheaper in a few days and also likely have rebates or sales. plus the 4670 fits the needs of most oem comp users. its tiny, runs cool, and doesnt need external power.

Log in

Don't have an account? Sign up now