Finally. We're finally getting somewhere interesting in the graphics industry. Although they're sure to return, the days of reviewing $600 graphics card after $600 graphics card are on hiatus, and instead we're reviewing a new class of mainstream cards with earth-shattering performance.

NVIDIA's GeForce 8800 GT kicked off the trend, in one fell swoop making almost all of NVIDIA's product line obsolete thanks to the high performance and low price tag (we'll talk about that last part shortly). But what we saw there wasn't a fluke, it was a preemptive strike against AMD, who have been hard at work on an affordable GPU of their own.

This new product, like the 8800 GT, would be aimed squarely at the $150 - $250 market segment, something both AMD and NVIDIA did a horrible job at with mainstream releases earlier this year (2600 and 8600 both sucked guys).

Introducing the RV670

AMD's two new graphics cards launching today are both based off a new GPU, referred to internally as the RV670. The basic architecture of the hardware is largely unchanged from R600; there has been some additional functionality added, and a great deal of internal bandwidth removed, but other than that this is very much an R600 based part.

The biggest news of this part is that it is fabbed on a 55nm TSMC process. This is a half-node process based on 65nm technology, giving AMD an advantage in die size (cost) and potentially clock speed and/or power.

Historically, AMD's RV series has been a cost cut version of their R series designed for lower end volume parts, and that's where RV670 started. Right of the bat, half the external and internal memory bandwidth of R600 was cut out. External bandwidth dropped from 512-bit to 256-bit, but AMD stuck with 8 memory channels (each dropped from 64bit to 32bit).

Internally, the ring bus dropped from 1024-bit to 512-bit. This cut in bandwidth contributed to a significant drop in transistor count from R600's ~720M. RV670 is made up of 666M transistors, and this includes the addition of UVD hardware, some power saving features, the necessary additions for DX 10.1 and the normal performance tuning we would expect from another iteration of the architecture.

Processing power remains unchanged from the R600; the RV670 features 320 stream processors, 16 texture units and 16 redner back-ends. Clock speeds have gone up slightly and memory speeds have increased tremendously to make up for the narrower memory bus.

The RV670 GPU is also fully PCI Express 2.0 compliant like NVIDIA's G92, the heart and soul of the GeForce 8800 GT.

New Features you Say? UVD and DirectX 10.1
Comments Locked

117 Comments

View All Comments

  • bbqchickenrobot - Wednesday, May 7, 2008 - link


    But - now new Catalyst drivers have been released - so an updated benchmark needs to be completed as the drivers provide better support for the hardware and thus, better performance.

    Also, you used a non-AMD MoBo and Chipset... if you went with XFire + AMD 790 chipset + Phenom X3/X4 processor (Spider platform) you would have seen a better performance as well. There are other benchmarks that are/were done with these components (spider) and the results weren't nearly as mediocre. Just a little tip...
  • Adamseye - Tuesday, February 12, 2008 - link

    I cant see how every review I have read differs from your charts, the 2900 xt can't be faster then the 3850.I mean I spent a month researching cards and the winner was the 3850 overclocking it to 3870 speeds. To think that AMD spent all that time to make a new 2900xt and name it the 3850-70, is just foolsih. from the benchmarks you provided only an idiot would buy the new gen cards for 60-100 buxks more when the 2900xt is on par. Could you please explain to me how this happened? I feel like ordering a 3850 was a waste of money because the old 2900 is better anyway.
  • aznboi123 - Saturday, February 2, 2008 - link

    Welll dang that bothers me...666...>,<
  • spaa33 - Monday, December 3, 2007 - link

    It looked to me that the biggest complaint on the HD Video Decode article was that the 2600/2900 options did not provide an off switch for the Noise Reduction. Did you notice if this option appeared to be present in the newer drivers of this card (3850)?

    Regards,
    Dan
  • emilyek - Tuesday, November 27, 2007 - link

    So AMDTI is still getting stomped by year old hardware?

    That's what I read.
  • jpierce55 - Saturday, November 24, 2007 - link

    This is really a good review, some others are very Nvidia biased. I would like to see you do an update with the new drivers in the near future if possible.
  • gochichi - Friday, November 23, 2007 - link

    Anand,

    First Nvidia with its 8800GT... I clearly recall seing those at about $200, now they're $300 or more. At least these may come bundled with a game... they also "hold the crown".

    Now the HD 3870 has gone up to $269.99 (at newegg) and availability is every bit as bad as the 8800GT.

    This review assumes that AMD/ATI was going to deliver in volume, at a fixed price and they haven't delivered either. It would be really nice if you could slap their wrists... as individual consumers we are being tossed about and we don't have the "pull" to do anything other than "take it".

    Shouldn't AMD be accountable to deliver on their promises?
  • SmoulikNezbeda - Thursday, November 22, 2007 - link

    Dear Anand,

    I would like to ask you what exactly results in individual games represents. Are those average FPS, or something like (min + max + ave)/3 FPS. On one czech website there were similar results to what was presented here, but they were showing (min + max + ave)/3 FPS, which is a complete nonsense as this would be advantageous for cards which have more volatile results. In case when they were comparing average fps the radeon had the same results as GT card. Also I would like to ask you whether you have used the same demo for both cards or you were playing a game and therefore testing a game in different situations?

    Thanks in advance

    Petr
  • SmoulikNezbeda - Thursday, November 22, 2007 - link

    Dear Anand,

    I would like to ask you what exactly results in individual games represents. Are those average FPS, or something like (min + max + ave)/3 FPS. On one czech website there were similar results to what was presented here, but they were showing (min + max + ave)/3 FPS, which is a complete nonsense as this would be advantageous for cards which have more volatile results. In case when they were comparing average fps the radeon had the same results as GT card. Also I would like to ask you whether you have used the same demo for both cards or you were playing a game and therefore testing a game in different situations?

    Thanks in advance

    Petr
  • Sectoid - Sunday, November 18, 2007 - link

    If I'm not mistaken the 8800GT is DX10 only right? Is DX10.1 so insignificant as to not count to the favor of the 3800's over the GT's? Don't get me wrong, I'm not trying to defend AMD; I just want to know if it's a good idea to sell my 8800GTS 320mb still at a good price now(I live in Brazil and they're still pricey here) and buy a 3870 or a 8800GT with 512mb. I recently bought a 22" monitor and the GTS is somewhat disappointing at 1600x1050. Nah, it's just that crappy game world in conflict. It runs similar to crysis demo at max! I have to play at medium and the textures are really crappy for a high-end pc 8-month old :(
    Who knows, maybe I'm already CPU or memory bound with a core 2 duo 6400@24xxMhz and dual ocz platinum 2 1gb 800mhz(2gb total)...
    Thanks in advance for any more input on the qualities of DX10.1 :)

Log in

Don't have an account? Sign up now