When it comes to choosing a new graphics card I find myself asking the same questions every time and I'm curious to see how other people approach these decisions; especially in light of the current cards that everyone's looking at: Radeon 6950, GTX570, GTX580. Bit-Tech's members vary wildly in their purchasing habits - I've seen plenty of you here who are happy with a GTX 260 (Hi GoodBytes!) while others are rocking Dual or Triple GTX580s (I'm looking at you Pete J!) so I'd like to see some responses from both sides of the divide on this. Bear with me, this could be a long one! Inb4 wall of text crit etc. - Many of these questions are relevant to other components such as CPUs, but I find myself upgrading graphics cards more often. ==== 1. Mid-range or High end? 2. If I go with the mid-range card will I see enough of an upgrade from my current card to warrant the cost of upgrading? 3. If I go with the high end card will I be wasting money on diminishing returns because of the progressively smaller differences in performance further up the price scale? ==== The first question is the one I'm struggling with the most right now, but it's tied into the second and third questions so I'll deal with those together. In the past I've always settled on mid-range or upper-mid-range graphics cards because of the price/performance ratio. Voodoo 3 -> GeForce 2 -> Geforce 4 TI 4600 -> 6600GT -> 7600GT -> 8800GTS 512 G92 -> GTX 275 (Note: I'm not an nVidia fanboy, I swear! Those just happened to be the choices that worked for me at the time of purchase and not all of them were smart!) Most of those cards were mid-range for the time of their release apart from the TI 4600 and arguably the GTS512 and the simple fact of the matter is that I was completely satisfied with very few of those purchases. When looking at reviews and online stores I always find myself hovering around the mid-range cards under the assumption that those are the best value, but I'm starting to wonder if that's really the case when factoring in more than just the price/performance ratio at the time of purchase. - A huge part of the whole problem is purely psychological - Knowing that I could have better performance or visuals than I'm currently seeing with my new purchase. When you buy a mid-range card you generally do so knowingly and are aware that you're not seeing the full performance that you could have had by spending more money. Of course, there's a strong element of diminishing returns there and I think most will agree there is a point at which throwing money at your system won't really net a performance or visual benefit to match the amount of money spent - And I'm not even talking about the Tri-SLI systems here. - When I had a 6600GT I was fairly satisfied with it, but always felt as though I was missing out because the performance difference between a 6600 and a 6800 was fairly significant in that series. A few months on and I had gotten over the novelty of having any new card at all and found myself thinking I should have gone for a 6800GT(X) so as to get those last 10-20 fps that would take me over 30/60fps at my desired visual settings. I felt the same with my 7600GT, thinking I'd have been happier with a 7800GTX. When I got my GTX 275 I found the same again, that it was just short of the right level of performance for the visuals I wanted from contemporary games and that I probably should have spent that bit more to get a 285 or 295. - On top of that feeling of dissatisfaction I find that I upgrade sooner after each mid-range card, whereas even though a high-end card usually costs disproportionately more than its performance benefits can justify, it would likely last longer in terms of satisfaction. I'm starting to think that I would feel more like I had spent my money wisely on an investment than on something disposable, even if it's obsoleted by the next series or refresh 6 months later. Am I seeing another side to the decision now that I hadn't before, or am I just playing right into the hands of nVidia and AMD by thinking high-end cards are worth their disproportionately higher prices because of the psychological benefits of product satisfaction? For those of you who buy mid-range cards, do you just buy what you can reasonably afford or justify, stick it in the machine, find your comfortable settings and enjoy your purchase until it no longer handles new games comfortably - Or do you find yourselves slightly dissatisfied the way I do? ==== 4. Is it better to get a mid-to-high end card and overclock it to the speeds of the high-end card, or is it better to just get the high-end card and overclock it higher still? ==== "Get the cheaper card and overclock it to the speed of the more expensive one." "lolol people who buy the more expensive one are idiots because i overlocked my 8800GT to the speed of a stock 8800GTS 512" And so on.. Often when a new series is released we see a lot of comments like the above on the reviews of the highest-end model from people who ritualistically buy midrange cards only and I can't help but think that in some cases it's PJS (Purchase Justification Syndrome) of a strange kind; where people find themselves regretting their midrange purchases and seek to make themselves feel better by thinking that with a bit of overclocking luck their new midrange card is really a high-end in disguise. I don't often see it said in response, but do people not realise that the high-end cards will generally achieve significantly higher clock values because of the wonders of speed-binning? In some cases the high-end cards max out only a short distance ahead of their mid-range siblings when overclocking, but I don't think that's typically the case. I'm fairly sure a reasonably-overclocked GTX 570 will never match a GTX580 overclocked by a similarly reasonable amount, for instance - Especially when considering factors such as the differences in total RAM and memory bandwidth. Of course you can get very lucky doing the above too and there are some weird exceptions to all of this, such as the current situation with 6950s needing only a BIOS reflash and clock-cycle change to be converted into their 6970 siblings that cost upwards of €50 more at retail. This is an anomalous situation and I personally expect it to be fixed in the next revision of those cards, once AMD and the board partners see how few 6970s are being bought by those-in-the-know. It feels quite dismissive to say that people who buy mid-range cards and insist on commenting on how much smarter they are than their peers who buy high-end cards are just jealous or venting PJS, but I'm starting to think that might account for a lot of it. ==== 5. Is the stock cooling good enough, or do I need to add €50-90 to the price of the card for aftermarket cooling, making this the wrong choice of card? ==== I'm an avid watercooler and I absolutely detest the vacuum-cleaner noise of most graphics card HSF units, so I habitually stick a waterblock on every card I buy. Unfortunately, that can add a hefty amount extra to the cost of the card, from €20 for a block adapter plate up to €90 for a new full-cover block. I love quiet systems, so I'm inclined to believe that it's a justified extra cost even without the increased overclocking headroom afforded by watercooling - But when looking at the price of watercooling parts relative to the price of the hardware I'm cooling, sometimes it looks a bit silly. Graphics card - €350, Full-cover block - €80, Single-slot PCIE bracket - €10, Silence - Priceless. (You know where to send the money, MasterCard.) This is also related to question 4 because it's arguably more justifiable to spend a significant sum on decent aftermarket cooling such as a full-cover block when you're pairing it with a card powerful-enough to 'deserve' it and get the most out of the extra cooling. I sometimes think it's a bit boy-racerish and silly when I see people putting €90 water blocks on cheap graphics cards when silence isn't their goal. Does the cost or availability of aftermarket cooling affecting your actual choice of card, or do you just choose your card as a priority and hope for decent aftermarket cooling later if it isn't available at time of purchase? ==== 6."Is now the right time to buy or do AMD/nVidia have something up their sleeves for next week/month/quarter?" 7."Is it better to pay more now as an early adopter or wait, pay less, but buy something that's obsolete sooner?" ==== These questions are always the most annoying to read an answer to, because they generally aren't answered until it's already too late. Bit-Tech is a great help in this area, moreso than many other hardware review sites, I feel. The closing paragraphs of the GTX580 review offer advice to hold off for a few weeks to see what AMD serves up in response, because the guys at Bit have an inkling that something is on the way. Without such advice the constant tick-tock of graphics card series releases and refreshes would make it very difficult as a consumer to get that kind of 'insider scoop' and to be able to predict if now is the right time to buy or not. Another factor that makes questions 6&7 awkward is that it can be tough to know when the prices of a new card are going to drop to a reasonable point and waiting too long can place you in the awkward position of buying a new card right in the middle of its shelf-life. That situation leads to the kind of dissatisfaction I mentioned in the first questions because your new purchase becomes obsolete sooner. ======== Someone is bound to say that I'm overthinking all of this, but I honestly feel that these are thoughts a lot of us have when we prepare to drop hundreds of our hard-earned euro/pounds/dollars on shiny new hardware; but I don't think the psychological side of the decision is often dealt with in discussions. It's too easy to just look at benchmarks and price figures and think "This offers the best bang for my buck - I'll get this." I don't think the decision is really as simple as we often make it out to be. What do you think?