Update: AMD has confirmed to us that there were some issues with the BIOS on our sample board. Rather than 2 disabled SIMD units, our review sample 4830 had 3 disabled SIMD units. AMD has assured us that no retail boards will be affected and this is only a problem that affected reference boards built as review samples. We are working on resolving the issue with our review sample and will complete updated tests as soon as we can. This will affect our performance results, but until we run the tests we can't be sure how much more performance we will get out of retail 4830 hardware.


Update 2: We have updated all the performance graphs in the article with data re-run on a card that actually has all the SIMD's available. There was a difference, but it hasn't changed the overall conclusion of the article. For more information, see our update article with details on the problem, the situation, and performance differences.

Since the launch of the RV7xx GPU, AMD has been steadily filling out a top to bottom Radeon HD 4000 series lineup. The first markets addressed were gamer centric with the 4850 and 4870. Next in line was the hardcore enthusiast class with the dual-GPU 4870 X2. Since then we've seen the 4670, the 4550 and the 4350 filling out the bottom end value and mainstream segments. But there was a bit of a performance and price gap between the 4670 and the 4850. This gap has now been filled.

Today we see the introduction of the Radeon HD 4830 which is to be priced at or below $130. This part is poised to split the difference between the 4670 and 4850, and filling in this market segment should finish rounding out AMD's line up of RV7xx based cards for now. At least we hope.

In the past both ATI and NVIDIA have flooded the market with way too many different models that overlap in price and performance in ways that just confuse their customers. While AMD has been releasing cards at a fairly steady pace, all these parts have been well positioned and have served to disseminate their new architecture. We have been really happy to see how quickly AMD has gotten their new GPU out into the world.

By now, we've covered the architecture and other versions of the hardware quite a bit. The really interesting bit about this launch is the price and the prices of competitive hardware.

All About Price and Rebates
Comments Locked

56 Comments

View All Comments

  • nirolf - Thursday, October 23, 2008 - link

    That low core/mem looks promising. It would be interesting to see if you can get close to 4850 with some tweaking.
  • Mathos - Thursday, October 23, 2008 - link

    And I think I found my next video card upgrade. Have been either waiting for the 4850 to drop a bit more, or something to come out between the 4670 and 4850 to come out, and this one hits exactly where I figured it would. Nipping at the heels of the 9800gtx in a few benchies there, and at the Res that matters for me 1680x1050. This looks like something I can pair with my 3870 toxic edition till I can afford a full on 4850 or 4870. Then if I wanted to I can get rid of the 3870 and run Xfire with the 4830 and 4850.
  • DXRick - Thursday, October 23, 2008 - link

    The charts for power consumption are totally different (show a lot more consumption, especially at idle) than the charts done for the 4670 article: http://www.anandtech.com/video/showdoc.aspx?i=3405...">http://www.anandtech.com/video/showdoc.aspx?i=3405...
  • JarredWalton - Thursday, October 23, 2008 - link

    From the other article: "At idle our entire testbed (Intel G45 + Core 2 Quad Q9450) used only 67W with the Radeon HD 4670." Note that this article uses QX9770 and 790i, among other differences.
  • DXRick - Friday, October 24, 2008 - link

    I had no idea that different chipsets (or mobos) and cpus could result in such a dramatic difference in power consumption. I sure want my computer to consume as little as possible when I am not using it.

    Are there any other articles here about this?

    Thanks!
  • strikeback03 - Monday, October 27, 2008 - link

    http://www.anandtech.com/casecoolingpsus/showdoc.a...">http://www.anandtech.com/casecoolingpsus/showdoc.a...
  • Jovec - Thursday, October 23, 2008 - link

    More importantly, all your previous reviews show roughly a 40+ idle and load watt difference between the 4850 and 4870, yet this review has it down to 3-4 watts at both. Was there a problem with 4870 power consumption that has now been fixed?
  • Jovec - Saturday, October 25, 2008 - link

    And now these 4850 and 4870 numbers show a wider margin again. With a different testbed I'd expect different numbers, but the relative difference on the same testbed should be the same. These numbers are more in line with AT's other 4850/70 power numbers. The original article's numbers need to be explained.
  • JarredWalton - Thursday, October 23, 2008 - link

    I believe that was one of the issues with previous Catalyst drivers: for some reason the power saving stuff wasn't working on 4870. It's good to see that finally addressed.
  • Jovec - Thursday, October 23, 2008 - link

    Seems likely for idle numbers, but I'd be curious what power saving can be done under load.

Log in

Don't have an account? Sign up now