1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Graphics Graphics Card Bang-for-Buck Comparison Charts

Discussion in 'Hardware' started by boiled_elephant, 27 Sep 2011.

  1. boiled_elephant

    boiled_elephant Merom Celeron 4 lyfe

    Joined:
    14 Jul 2004
    Posts:
    6,569
    Likes Received:
    805
    EDIT - note that the prices this analysis is based on are now out of date. If you want the chart to be accurate again, download the spreadsheet and input some prices you find to see how cards compare. I was going to keep doing this myself on a monthly or quarterly basis, but I don't think anybody visits this anymore.

    Update:

    For absolute FPS-to-Price graphs and work with average FPS values, see Zinfandel's awesome work on that here and here. Here's the bang-for-buck analysis:

    [​IMG]

    This is based on minimum FPS, averaged across Bad Company 2, ARMA II, Black Ops and Dirt 2 (data taken from bit-tech's reviews). Prices are UK prices only and are taken from scan, aria, dabs, overclockers, ebuyer, amazon UK and ebay (buy-it-now only, failing all other stores).
    (The other resolutions aren't important - the revelation here was that, for all resolutions, these cards stay in the above ratios with regard to bang-for-buck. Whatever resolution you're on between 1680 and 2560, this chart applies. If you want to check for yourself, the other graphs are on sheet 2 in the Excel file below.)

    So I can finally answer the dreaded HD 6870 vs. GTX 460 1GB and GTX 560Ti vs. GTX 570 questions. Here's the lowdown:


    Top Bang-For-Buck Cards

    £85 - Nvidia GeForce GTX 550Ti 1GB
    £100 - AMD Radeon HD 6790 1GB
    £135 - Nvidia GeForce GTX 560 1GB
    £164 - Nvidia GeForce GTX 560Ti 1GB
    £147 - Nvidia GeForce GTX 470 1.3GB
    £250 - Nvidia GeForce GTX 570 1.3GB
    £348 - Nvidia GeForce GTX 580 1.5GB


    How you use this: identify your maximum price and find the card in the above that's closest to it. That's the best bang-for-buck card in your price range. If you're tempted to go higher or lower, try to go to the next higher or lower card in this selection, as the ones in between are generally worse value-for-money purchases.

    So for the 560Ti/570 choice, for example, as they're both in this list it just comes down to which meets your needs, and/or which you can afford. If you can afford both but only need the performance of the 560Ti, get that; if you can afford both but know you need the performance of the 570, get that. If you can afford both but care more about cutting costs than about raw performance, get the 560Ti (cheaper AND higher absolute bang-for-buck.) If you can only afford the 560Ti but need the performance of the 570 - tough ;)

    Note: If you find a card on a special offer that takes more than £10 off its price in the table below, you can safely assume that it trumps a slightly more expensive card.

    Example: the AMD Radeon HD 5850 1GB at £162 is bad value for money compared to the Nvidia GeForce GTX 560Ti 1GB at £164, but if you found an offer that brought the 5850 down to £150 it would become the better bang-for-buck choice.

    [​IMG]

    These were the lowest prices on 28-9-11, but of course they change constantly. If you want to stick your own prices into the tables and see how they compare, you can download the Excel file here.


    edit - It's worth mentioning that this is wholly unbiased and that I'm not an Nvidia fanboy. I'm not even an Nvidia fan - my last three cards have been the Radeon 9550, the X1900XT and the HD 4870. Traditionally, ATI always held the bang-for-buck crown; I was as surprised as anyone to find that Nvidia are actually beating them for value in most price brackets at the moment. Make of it what you will - it'll probably change again in a month.



    Original post:

    I thought some people might find this interesting or useful. A few days ago in the FS forum there was this:

    It got me thinking. After some thinking I got industrious and decided to chart the bang-for-buck value of the current graphics cards, modifying a chart from bit-tech's review of the 6950 1GB.

    I treat minimum FPS as the metric of gaming 'bang'.

    I take Bad Company 2 to be most representative of current gaming demands (RTS are all CPU-limited anyway).

    I pooled prices from Scan, Aria, Ebuyer and Overclockers.

    Some newer additions are missing because the article dates from Feb. 2011, but nothing magnificent has happened since then anyway.

    [​IMG]

    So the 560 Ti is still the higher-end monster. And for lower prices, surprisingly enough (to me), the 460 is still the mini-monster. The 6870 is quite close behind, though, and enjoys better performance than the 460 in some scenarios.

    Of course, it all changes when you throw in one of Overclocker's frequent special offers. Knocking £20 off a 460 or a 6870 changes its value coefficient significantly.

    Open to suggestions of improvements to the method, or of subsequent cards that might weigh in well that I should've included. Quite enjoyed putting this together.
     
    Last edited: 18 Aug 2012
  2. favst89

    favst89 What's a Dremel?

    Joined:
    23 Jul 2010
    Posts:
    390
    Likes Received:
    13
    I like this idea, well done. I would also like to see the effective bang for buck different screen resolutions.
    I appreciate it would take more time and you've picked the middle ground for most people.
    However, as an example in battlefield again the HD5970 beats a 560ti by about 23 fps (minimum) and even beats a 580, just. At 1920x1080. When you move to 2560x1600 it suddenly falls just below a 560ti and well below the 580.
     
  3. fdbh96

    fdbh96 What's a Dremel?

    Joined:
    29 May 2011
    Posts:
    1,894
    Likes Received:
    33
    Battlefield is quite CPU restricted aswell compared to other fps games
     
  4. mucgoo

    mucgoo Minimodder

    Joined:
    9 Dec 2010
    Posts:
    1,602
    Likes Received:
    41
    Were do you get 460 768mb for £85???

    If your actually serious about pursuing a proper chart system for this then the best system would be average the minimum fps across the test games for 1920 and 2560 resolutions with each res having a separate result value.

    Would it be possible to make somekind of wiki style spreadsheet so the burden of keeping this sort of thing up to date is eased?
     
  5. Nazata

    Nazata What's a Dremel?

    Joined:
    2 Sep 2011
    Posts:
    160
    Likes Received:
    6
    +1 boiled_elephant, good guide an price comparison and quality, especially for those like myself who are new to building gaming rigs :)
     
  6. longweight

    longweight Possibly Longbeard.

    Joined:
    7 May 2011
    Posts:
    10,517
    Likes Received:
    217
    Nice work, surely average FPS would be a better guide?
     
  7. Fingers66

    Fingers66 Kiwi in London

    Joined:
    30 Apr 2010
    Posts:
    8,688
    Likes Received:
    919
    Nice. Techreport.com do a similar thing by scoring multiple games across different resolutions then summarise the score like this: http://techreport.com/articles.x/20629/12. However, I haven't seen anyone do one with UK prices and availability before.

    I would imagine that it would take a while though...:D.

    +rep for the effort put in on this mate.
     
  8. Sloth

    Sloth #yolo #swag

    Joined:
    29 Nov 2006
    Posts:
    5,634
    Likes Received:
    208
    Another +rep. There are, of course, so many different games and settings and resolutions and ways of judging FPS for each but as a good way to set the starting comparison to see where further research should be focused this is excellent.

    For anyone who can get a card for a little bit less or more cost they can grab a calculator and figure it out for themselves.
     
  9. SirFur

    SirFur PC Gamer and LAzy B0nes

    Joined:
    8 Apr 2009
    Posts:
    1,189
    Likes Received:
    14
    As a suggestion...maybe to use a game that pushes all the GPU to their limits, something like Just Cause 2 maybe? This would enable the better cards to shine more/not...

    However, the prices above are comparing stock performances...most cards will have a small-ish price added for a decent amount of overclock, and not all cards have been reviewed by the same reviewer to provide an accurate comparison....but thats a lot of work.

    Nice stuff though!!
     
  10. boiled_elephant

    boiled_elephant Merom Celeron 4 lyfe

    Joined:
    14 Jul 2004
    Posts:
    6,569
    Likes Received:
    805
    Thanks for the feedback guys, appreciate it.
    It's at overclockers. I only discovered them recently, via the hardware bargains thread in the FS forum.
    These are excellent suggestions. This was a sketch effort to see how informative a value coefficient would be; next up will be an average across games, as you say. And yes, when I turn this into a proper system it will be Excel-based, rather than just MS-Paint and a calculator :)

    I considered it, but decided that minimum FPS is more crucial due to it being more determinate of your gaming experience. A higher average FPS doesn't do much, as it tends to occur mostly in the least important sections of gameplay where nothing's happening, whereas the minimum FPS is hit during the heavy moments when you most need as much FPS as your system can produce.

    A middleground would be to average between the, uh, average and the minimum, and I did consider doing that, but truthfully I don't think average FPS wins back many points against a poor minimum FPS. Any gamer would sacrifice some average FPS for a higher minimum FPS, and a large gap between the two is often taken as a reason to avoid a card - it speaks of instabilities, unreliable performance and driver issues. This is why bit-tech's graphs rank by minimum.

    Time to break out Excel.

    edit - if there are any UK stores other than the 4 I mentioned that people think I should include when checking prices, lemme know.

    edit - I'm thinking of making a cutoff point at 25 min. fps, too, because while a card below that might technically be great value for money, most of us probably won't be considering it.
     
    Last edited: 27 Sep 2011
  11. SirFur

    SirFur PC Gamer and LAzy B0nes

    Joined:
    8 Apr 2009
    Posts:
    1,189
    Likes Received:
    14
    How about Dabs and Amazon?
     
  12. GeorgeK

    GeorgeK Swinging the banhammer Super Moderator

    Joined:
    18 Sep 2010
    Posts:
    8,649
    Likes Received:
    478
    Excellent work +rep

    I've found CCL to be good recently wrt prices - might be worth a compare...
     
  13. Zinfandel

    Zinfandel Modder

    Joined:
    2 Aug 2010
    Posts:
    3,233
    Likes Received:
    176
    Ok, I'm bored and unemployed for another two weeks and thus...

    This is all the cards from the last two generations and all FPS marks at all resolutions. First off, I've used the average frame rate and not the minimum. I realise this goes against Bit Tech wisdom but as far as I'm concerned, I'd rather have an average of 50 with the occasional drop of a few FPS below another card with an average of 39. There are cases in the BT charts where a card that has >10 Avg. FPS ranks lower because its minimum is 1FPS less, which as far as I'm concerned is mental.

    It's all very simple. Just all the FPS at all resolutions added together and divided by the number of samples (Last gen cards were not tested on Arma II). That obviously skews the results as new gen cards have 12 sets of data, 3 of which are considerably lower than average because it's Arma and thus there are two sets provided. One with Arma included and one without.

    It's important to say though, as far as I'm concerned the data set is no where near as complete as I'd like it to be. A far broader selection of games would be better and at different settings but we have what we have. Obviously I can't guarantee how accurate this all is, I'm not getting paid for this haha.

    So, this is the score for FPS Vs Cost excluding Arma.

    [​IMG]

    This is the Average FPS excluding Arma.

    [​IMG]

    This is the score for FPS Vs Cost including Arma

    [​IMG]

    And the Average FPS including Arma

    [​IMG]

    Pick your price, find the best performance. This exclude Arma. As I say, it gives an unfair reduction in score to cards tested with it.

    [​IMG]

    There you go.

    If anyone cares I'll do some stuff at specific resolutions.

    Also, if anyone wants the spreadsheet where you can just change the prices and it'll all update for you drop me a message and I'll send it to you.

    Zinf.

    PS. Sorry BE if this seems like I'm stepping on your shoes, I was just bored.

    EDIT: Also as BE suggested, offers/changes in price effect things massively. The 6870 comes out lower on the Coef but for £20, I'm still going to stick with recommending it even if it's not the £125 it was a week or so ago.

    EDIT 2: Also, the 768 looked too much of an anomoly but it's been checked again and again. It just works out well for the price so if you've only got £85... That's the card to get, if you trust 3 games as a data set at least.

    EDIT 3: Sorry about the tags on a couple of them. Meh.
     
    Last edited: 28 Sep 2011
    adidan, Fingers66, The_Beast and 2 others like this.
  14. Parge

    Parge the worst Super Moderator

    Joined:
    16 Jul 2010
    Posts:
    12,994
    Likes Received:
    609
    This thread is an awesome resource! + rep!
     
  15. shah

    shah Minimodder

    Joined:
    29 Jan 2010
    Posts:
    753
    Likes Received:
    21
    Great work both of you :D
     
  16. trig

    trig god's little mistake

    Joined:
    10 Aug 2006
    Posts:
    2,848
    Likes Received:
    43
    nice job fellas...this had to take some time...nothing but appreciation...gives a guy like me in the states some basis for recommending products in builds to folks across the pond...

    i find it amazing that the 5850 is still up there...lol...
     
  17. stefan442

    stefan442 Minimodder

    Joined:
    14 Mar 2011
    Posts:
    367
    Likes Received:
    14
    yep. fantastic piece of information. +rep left right and centre sir.

    Wish you had done it before i bought my gfx card :p Would have meant less research on my part... :)
     
  18. Spanky

    Spanky Minimodder

    Joined:
    14 Apr 2010
    Posts:
    767
    Likes Received:
    93
    Well, i bought a GTX 480 a short while ago from Aria for £160 (lets no forget the 2 yr warranty). I tore the niosey cooler off and popped the Zalman V3000f on. I actually upgraded from a 5850 and as im running 2560x1440 it has dramatically increased my minimum FPS.

    Many people scoffed when i suggested the 480 but its turned out to be a cracking card, i ramped the clocks to a very comfy 800/1600 and kept the RAM at stock because this is where i noticed the large temp increase's. Still, it sits at 37 idle and hits 73 under load, and when i say load im talking about a couple hours playing black op's , i have no real need to thrash the **** out of it in furmark.

    Charts are great. Bang for buck .... I aint so sure.
     
  19. trig

    trig god's little mistake

    Joined:
    10 Aug 2006
    Posts:
    2,848
    Likes Received:
    43
    that's it...cook that baby...come on
     
  20. Zinfandel

    Zinfandel Modder

    Joined:
    2 Aug 2010
    Posts:
    3,233
    Likes Received:
    176
    Frame rates by resolution as requested. Excluding Arma II. 6990 not included in 1680 x 1050 as no result posted.

    [​IMG]

    [​IMG]

    [​IMG]
     
    David likes this.

Share This Page