Back to Article

  • ritalinkid18 - Monday, February 25, 2008 - link

    I think this review underestimated the price difference. The key here is cost. It said that cost wasn't a factor when comparing the two in question, but then used as a factor when comparing the SSC to the 8800 gt. The 8800gt's pricetag makes cost a big factor.
    Not to mention the SLI vs Crossfire, which is undoubtably the 9600gt's win. ( Case and point:"> ) This may be about cost and a cost saving upgrade down the road should be included.

    The fact that the 9600gt is the clear winner here in completely understated.
  • 7Enigma - Monday, February 25, 2008 - link

    Derek and Anand I had a quick question. I, like several others, would have preferred you to have a high(er) end card like the 3870 or 8800GTS (or even just the 512meg version of the GT). I understand the purpose of the review is to show mid-range cards, but its the SLI configuration that annoys me.

    In a recent Daily Tech article the author eludes to 2 of these 9600's in SLI being a "better" value than a single more expensive GTS (since the 9600's would technically be less expensive). The problem with this statement as we all know is that it is software dependent on whether the extra card will actually be implemented well. Otherwise you have a single 9600 that would get blown away by a GTS. Unfortunately less informed readers of DT that then come over to Anandtech for the review would be shown the 9600 in SLI as the leader.

    I understand you don't want to muddy the graphs/charts with the last 100 video cards you have reviewed, so here is my question:
  • 7Enigma - Monday, February 25, 2008 - link

    While its great to use the latest and greatest CPU/RAM/Mobo (to make sure the system is GPU constrained), is it possible to stick with a static testbed for at least a couple months (say 3-6)?

    Reason being, right now if you had not switched from a QX9650 to a QX9775 between your last review (3870X2), and not switched from Vista 32-bit to 64-bit, we could possibly check back and compare to previous reviews to do our own comparisons. I understand driver revisions are going to still be a factor but as long as we don't see large differences in performance between the drivers it could be a good comparison.
  • 7Enigma - Monday, February 25, 2008 - link

    To keep things even more consistent between reviews just keep a HD image of the drive and reformat back to the same static testbed whenever doing a video card review (more expensive but easier would be just a HD swap). We understand you cannot have a dedicated system that doesn't do anything but review video cards, but if you could quickly just swap out a HD or reformat to the same state as a previous review that would end these constant questions about, "Why didn't you test so and so?".

    You could then say, "Look it up yourself!"

    Here's the quote from the DT article in question:

    "As far as performance goes, NVIDIA claims a couple 9600 GT's in SLI mode will perform slightly better than a single 8800 GTX. With the cost of an 8800 GTX still above the $400 price point, the 9600 GT SLI setup is a better investment. Just to mention it, NVIDIA states the 9600 GT puts out twice the frame rates as the 8600GTS when comparing SLI performance."
  • 7Enigma - Monday, February 25, 2008 - link

    test Reply
  • 7Enigma - Monday, February 25, 2008 - link

    Test Reply
  • jamynn - Monday, February 25, 2008 - link

    Why are the results for Crysis, Oblivion and Quake Wars 880GT 256 so much lower in this article than they are in their original review in this article?">

    Definetly makes the 880GT256 look alot worse than it originally tested. Looks Suspicous.
  • MadBoris - Saturday, February 23, 2008 - link

    Ok, I was waiting for this review to popup on the home page but it never did on my visits why is that?
    It seems now I have to scour the site, in this case the video section, to look for it on some whim. The home page has really become disappointing, sometimes it shows an interesting blog but doesn't include this review, the home page is doing a disservice to visitors. I hope you guys can fix that or start sending all new content to the home page.

    Secondly, I would have really liked a comparison to the 8800GT 512 in this review, the question to me is the 8800GT 512, for only $20 - $30 more, worth it over this card.
  • niva - Friday, February 22, 2008 - link

    I still have my Tyan Tachion 9700 Pro in my 6 year old system and runs just fine and I use every day for office/email/internet and gaming! Of course I don't run any of the stupid games which require new gen cards. I'll keep it until it dies I think. Ultimate card evar though at the time it came it was high end as heck!

    Seriously though, we should all be happy that we actually HAVE a mid-price range market now. I am, thanks Nvidia!
  • Groovester - Thursday, February 21, 2008 - link

    It seems strange that no mention was made of the 8800GS card. Perhaps it wasn't available when AT decided to test the 9600, but it would be nice had they tested the two. After all, they are NVidia's two newest mid-range video cards. According to the 8800GS specs, if they are going to retail for about the same price, it looks like the 8800GS would perform better. And I doubt that I would be using the new color "saturater" of the 9600. Reply
  • pmonti80 - Friday, February 22, 2008 - link

    That's probably becuase it's a transicional product, in one or two months you won't be able to buy one. Reply
  • poohbear - Thursday, February 21, 2008 - link

    nice review, but i expected to see more cards compared with the 9600gt from a site like anandtech, especially the 8800gt 512mb version which everybody's been buying. are you guys on a budget or something? Reply
  • anachreon - Thursday, February 21, 2008 - link

    Somehow this review feels a little sloppier than the past AnandTech video card reviews I have come to trust. The cards represented in various tests are inconsistent, and the lack of a 512mb 8800 GT, as well as AA and AF, in the tests is baffling. Reply
  • DerekWilson - Friday, February 22, 2008 - link

    I don't understand what you mean about the cards represented being inconsistent. We tested the exact same six cards in every tests and the same 2 multiGPU configurations as well ... the only graph that lacks anything is the WiC 16x12 graph because we could not get the 3850 to complete the benchmark at that resolution.

    The 512 MB 8800 GT isn't really in competition with these cards in terms of price. Since AMD dropped the price so dramatically, it's more of a direct comparison, and if we had known before hand we would have included something else from the next price point up (like the 512MB 8800 GT).

    We can't test everything for every review, and we've got to make trade offs. Sometimes we make the wrong call, and not including the 512MB 8800 GT was one of those time. We'll certainly include it in follow up testing.

    Derek Wilson
  • pmonti80 - Friday, February 22, 2008 - link

    Dereck I think what he means is that at 1st sight the results are little bit strange. I had to check several reviews to see that the results are the same (how could i ever doubt you? ;)).
    An example of strange results at 1st sight is the 256MB 8800GT. Also the fact that filters give an advantage to the 9600 GT and the test without filters give an advantage to the 3870.
  • GTaudiophile - Thursday, February 21, 2008 - link

    So if I have a eVGA GeForce 7900GT, which would be the better upgrade? A 8800 GT with 512MB RAM or a 9600GT? Can you get a 9600GT with 512MB RAM? Reply
  • xsilver - Thursday, February 21, 2008 - link

    7900gt to 9600gt wouldnt be a colossal upgrade. It would probably be better to get the 8800gt or 8800gts otherwise stick it out with what you've got until the next 9xxx part rolls around. Reply
  • LoneWolf15 - Thursday, February 21, 2008 - link

    No AA/AF makes these benchmarks nearly useless. Also, while Oblivion is a great game, it is now a dated game, and no longer a good standard to measure cards by.

    I'm sorry, but I couldn't base a buying decision off of this review.
  • semo - Thursday, February 21, 2008 - link

    wouldn't it be better to put in 2 higher clocked dual core processors. aren't 4 cores more than enough for games today? Reply
  • peldor - Thursday, February 21, 2008 - link

    What's up with no AA tests at all and in some cases no AF? Seems like half a review without those. Reply
  • Spacecomber - Thursday, February 21, 2008 - link

    Thanks for the review and helping us to keep up with the latest in video card releases. I'm glad to hear that you'll be going back to include the 8800GT 512MB and the 3850 512MB. It's a crowded field with all of these cards that have come out in the last few months, but I think it is important to try and keep an eye on all of them when making these comparisons.

    It may not be practical at this time, since the 9600GT isn't available for sale, yet, but it would be nice to see a bang for your buck kind of chart for these new cards. I know that anandtech has done something like this in the past. The arbitrary part will be deciding what benchmarks (games and resolutions) to use for this. Still, I would find this interesting to see.

    Any ideas as to what resolutions (i.e., size monitors) that people looking for the $150-$250 card likely might be using?
  • Verdant - Thursday, February 21, 2008 - link

    use a midrange lunch right about now, ...mmm Subway Reply
  • fic2 - Thursday, February 21, 2008 - link

    "major complaint we had of previous midrange lunches."

    Do you get fries with those lunches?
  • Sunrise089 - Thursday, February 21, 2008 - link

    You guys really needed to test a 8800GT 512 or 8800GTS 512. I understand this is a midrange part, but when it tops the benchmark charts in some games, it's absolutely essential to know what card is actually faster. Looking at the Enemy Territory test for example, one could conclude that this could be the single fastest GPU available.

    PS - For those talking about the "best midrange GPU ever" - forget about it. This card certainly trumps the last-gen midrange, so it's much better than the 8600s, and on par with the 7600s. The 6600s though actually topped the last-gen high-end. 6600gt over 9800pro is IMHO the example of the ultimate midrange card.
  • andrew007 - Thursday, February 21, 2008 - link

    It is ABSOLUTELY essential to add 8800GT numbers. Without those numbers I just have no idea how well (or not well) this card performs. It doesn't make sense to not include a card that sells at a similar price and is probably the most popular right now and is currently used as a yardstick. Reply
  • nubie - Thursday, February 21, 2008 - link

    Natch, the 7900GS is quite simply the ultimate mid-range (although I do have a couple 6600GT, and they are really good). I suppose the argument could be made that they are really entry-level High-End, but the prices are mid-range.

    For $95 on ebay and dropping, the 7900GS isn't hard to recommend. With Ramsinks and an aftermarket cooler it can easily hit 650mhz stable on the core with a slight voltmod (some voltmodded Zalman'ed ones are $100)
  • Sunrise089 - Thursday, February 21, 2008 - link

    7900GS isn't a bad choice either, but it looses out on 4 counts:

    1) It cost more - as you said, it's entry level high-end by total numbers sold, even if it isn't by percentage of maximum possible GPU cost.

    2) You had to work your way further up the line to clearly beat the last gen - it's most impressive to beat the last top-end card with a card more stages removed from the top part (6800ultra-6800gt-6800-6600gt is more than 7900gtx-7900gt-7900gs)

    3) Perception wise, it has the "high end" naming scheme where the second diget of the name is 8 or 9.

    4) Most importantly it came out much later. This is the essential qualifier of a great card IMHO. The 6600gt beat the 9800pro within a month of the first 6000 series availability. The 7900gs was a later-released part where nvidia had more time to work on economics-of-scale to keep the price down.
  • ChronoReverse - Friday, February 22, 2008 - link

    Not to mention the x1950Pro beat the snot out of it. Reply
  • nubie - Thursday, February 21, 2008 - link

    1) Sorry, 8600GT was always more expensive than 7900GS. (no facts of course, I am probably not 100% true)

    2) 7950GX2-7900GTX-7950GT-7900GT-7900GS If you want to be technically accurate.

    3) Pffftthth, as I say, they were available for $130-150, the 8600GT wasn't (isn't??)

    4) Ah, but I really don't care about ATI.

    So in summation, I agree with you, but it is a very close call, facts remain that the 8600GT was only a good buy if you needed the "True Purevideo HD" Full hardware decode, and performance around that of a card $50 cheaper.
  • nubie - Thursday, February 21, 2008 - link

    Oops, I almost forgot, the 7900GS is an x1950XTX killer, for about the same price the 7900GS could be overclocked easily to 650mhz and eat one for lunch (as mine has been doing for 2 years now :D), so your 6600GT/9800 analogy is applicable to my 7900GS one. Reply
  • ChronoReverse - Friday, February 22, 2008 - link

    The 7950's can't beat the x1950's much less the 7900 series. The gap has only widen as more modern games came out. Reply
  • Aberforth - Thursday, February 21, 2008 - link

    Crysis benchmark @ medium settings? gimme a break, Obviously this review is designed to show 9600GT in good light. Reply
  • rcc - Thursday, February 21, 2008 - link

    As I read it, it's designed to show how it compares to similar cards. Sure they could provide test in in Crysis at full settings, but it's a bit like providing grand prix lap times on a Tercel.

  • Staples - Thursday, February 21, 2008 - link

    Either that or this was done very sloppy. The 8800GT 256 is mysteriously missing in all the benchmark which do not require a lot of video memory. In these tests, the 8800GT 256MB would win. I smell a rat. Reply
  • Kurotetsu - Thursday, February 21, 2008 - link

    "The benchmarks don't prop up the card I bought so it must be bogus!"

    If the 3850 was leading in that test, you wouldn't be complaining at all.

  • Griswold - Thursday, February 21, 2008 - link

    (_,_) Reply
  • pmonti80 - Thursday, February 21, 2008 - link

    It's just a review made at the kind of settings most of the cards analized will be used. How useful can be a review were the cards are around 15 fps? Reply
  • ncage - Thursday, February 21, 2008 - link

    For those who want a more lengthy writeup with more games tested and a bigger selection of video cards then check out the review at tomshardware. I wanted a comparison between the 8800GT 512MB & the 9600GT which is available at toms. I will tell you they are pretty close in benchmark numbers but the 8800GT still wins in every case which im quite happy about since i just bought a 8800GT :). We will see how good value the 9600GT is after the etailers do their price gouging. Still Nvidia is doing a VERY good job in bring competition to the market. Reply
  • knitecrow - Thursday, February 21, 2008 - link

    bah... i am not going back to tom for anything Reply
  • murphyslabrat - Thursday, February 21, 2008 - link

    I am sure this is on everyone's mind: overclock the thing already! Reply
  • dm0r - Thursday, February 21, 2008 - link

    Ca we say the best midrange nvidia card ever?
    Good review.Ill keep tunned for more testing.I also would like to see the 256MB variant.Thanks
  • kmmatney - Thursday, February 21, 2008 - link

    No, I don't think you can say best mid-range card ever. The past year has been so bad, it just makes this look like a good deal. This is what the mid-rnage should have always been like. Some better mid-range cards in the past

    Better mid-range cards that I've owned in the past

    Voodoo3 1000 - $45 card, performed better than $100 cards at the time
    Ti4200 - $120, very overclockable
    Radeon LE - $65 - bios update to perform like a $165 card
    6600GT - defacto mid-range card for a long time
    Radeon X800GTO2 - Bios update to 16 pipes, X850XT PE speeds

    There are others, these are just very good mid-range cards that I've owned, that I would say offered the same or better bang-for-buck as the 9600GT.

  • BigLan - Thursday, February 21, 2008 - link

    The 6600 was a great card for it's time, and further back than that the 4200ti was incredible - though you could argue that it wasn't technically mid-range. Reply
  • dm0r - Thursday, February 21, 2008 - link

    oh, forgot to mention the temperature tests Reply
  • knitecrow - Thursday, February 21, 2008 - link

    It is a good preview but I would like to see more in depth benchmarking, esp. with titles like Gears Of War, and Bioshock Reply
  • Spivonious - Thursday, February 21, 2008 - link

    Is the 3850 included in the benchmarks at 256MB or 512MB? Reply
  • hadifa - Thursday, February 21, 2008 - link

    Tomshardware has done some tests with the 512MB version Reply
  • ImSpartacus - Saturday, February 23, 2008 - link

    I don't know why anyone bothers with the 256mb version anymore. The 512mb is cheap as dirt and does wonders on those higher resolutions.

    I was planning on getting one, but this 9600gt is looking a little better.
  • Spivonious - Thursday, February 21, 2008 - link

    Nevermind, I read the whole last page now :) Reply

Log in

Don't have an account? Sign up now