Lol, doubt they kept a 980X hanging around for benchmarking purposes
Wish I had the money to get a load of stuff like that, I waste enough free time to warrant doing all this myself
Fraps will benchmark games fully, considering the fps every refresh (graphable), and the amount of time it took to render each frame, so you can check for spikes (also graphable). This allows you to track average, minimum and maximum fps in each game.
i see it now, fraps does minmum aswell. nice.
hmm...
I'd like to see minimum fps charts (ideally replacing av. fps charts as it will give a better perspective on what these cards are actually capable of doing), Metro 2033 benches would be nice (might involve getting ATi to actually work on the drivers for it though), maybe 3D results when the DDD kit is released, or using the IZ3D monitor and also try and get a stable OC at the beginning of the review and do tests with OC and at stock or OC at the end of your review process and go over the tests again.
I know a lot about benchmarks and the current suite is slightly tailored towards favouring nVidia.
BC2 favours ATI because nVidia's driver team hasn't had a chance to work on it yet.
FC2 is so far in favour of nVidia it has to be considered proprietary. It's old and nobody really plays it, so why keep benching it?
Dirt 2 is fair because even though it's an ATI game, nVidia wins at it. Expect nVidia to get even better as time goes by. As a dx11 game, this is a good win for nVidia and a problem for ATI - it should be included because of that.
Crysis - fair, it's a killer of all gpu's.
HAWX is just old and bad and doesn't deserve to be used as a benchmark for dx11 cards.
I would wager that most of those games made it into nVidia's "reviewer's guide". All except BF2 and maybe Crysis, those were probably replaced with Metro and Metro with PhysX.
I doubt that very much, the game's been out for quite a while now.BC2 favours ATI because nVidia's driver team hasn't had a chance to work on it yet.
I agree with FC2, though it's more biased pro-SLI than pro-nvidia per se.
Yeah Metro 2033 is the obvious one, but when I found out nvidia were only faster because they hacked 'very high' details back down to standard 'high', I lol'ed - that probably isn't nvidia-biased either, truth be told.
The only time minimum fps is valid in terms of gaming performance is when it's shown as a timegraph like Hardocp does it. As a single number on a chart, it can be abused.
Example -
A 5870 runs BC2 at 50fps average, 80fps high, but one time in the benchmark it falters and scores 10fps. The true 'low' is closer to 30fps over most of the benchmark.
A 480 runs BC2 at 40fps average, 70fps high, but one time in the benchmark it falters and scores 20fps. The true 'low' is closer to 30fps over most of the benchmark.
Does the 480 really deserve to be twice as good on minimum framerate? Of course it doesnt, and that is why only a timegraph is useful when minimum framerates are included.
why not avg the minimums?
Well the overall average is just the average minimums subtracted from the average highs.
Time graphs aren't difficult to do, all it needs is fraps and a spreadsheet.
I dont think there is anything wrong with just averaging the whole range of numbers, you guys need t stop moaning and let them get on with it.
Some of the games are a bit odd in the review ine up i do agree with that. All the games that are int he review line up should always be the top played games of the time. BFBC2 is a definite win, where is MW2? I dont liek it but a lot of people do.
Company of heroes should never have been taken out, its old but its still popular and with full settings is pretty demanding.
Jimbo: It's not difficult, remove anomalies like that, and state the minimum figure that is often observed. Full frame rate graphs are nice of course. Out of curiosity, who reported the figures you state?
Biscuit: Using minimums raw might add some strange readings (which would no doubt be explained) but just using average alone misses out on a lot of the detail. It ignores a large number of issues.
Call of Duty, I'll be honest, doesn't belong in a benchmark of high-end dual GPU setups. The frame rate limiter is set pretty high in it, and everything here would exceed said limit, the game just isn't that demanding.
Company of Heroes on the other hand, whilst far older, is still worthy of consideration, as RTS games often suffer lower frame rates.
Hardocp does time graphed benchmarks. They are pretty cool so long as there aren't too many cards on the graph, at which point things can start to get messy. Nobody reported the figures, I just used them as an example of how a single minimum drop can give a false impression.
Best way to do it - raw data in a script - choose which cards to display and not display on the graph, render it on demand. A good anti-hotlinking tool too
Concerning the Price of the GTX 470 you can purchase one from ARIA.co.uk, arias on version for £278.84 including delivery which is 30 to 40 GBP cheaper than the HD 5870 - and more importantly is in stock
5870 is still in stock, but you're right, they're about £15 more expensive than 470s excluding outliers. But that seems to be a worthwhile price to pay for the advantages.Originally Posted by Smacks
Concerning the Price of the GTX 470 you can purchase one from ARIA.co.uk, arias on version for £278.84 including delivery which is 30 to 40 GBP cheaper than the HD 5870 - and more importantly is in stock
Yeah, Aria have some decent prices on the 400 cards, and better stock levels than most, though the only cards below the standard pricing points (i.e. £300 for the 470 and £450 for the 480) are OEM cards, so it's not an amazing deal. That, plus the prices still haven't droped to appropriate levels. The GTX470 is an HD5850 rival, not an HD5870 rival, and thus it should really be at the £250 mark, for good brands, not just OEMs.
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
LinkBack URL
About LinkBacks
Reply With Quote
