Which is why it's bad. We don't want or need the most notable performance improvements, or losses, obscuring or "softening out." The averages are useless.Originally Posted by Kaotika
Which is why it's bad. We don't want or need the most notable performance improvements, or losses, obscuring or "softening out." The averages are useless.
Last edited by icemanchilled; 04-17-2012 at 10:18 AM.
The 7970 can be had for £345ish at some places... Seems about right for the feature set.
Huh? So in your opinion, if you play wide variety of games, it's better to take individual results which make the card of your choice look good, and then think "yay, this is this much faster" and ignore the results where the card isn't that much faster than the card you're comparing it to? Or how else, if not averaging, did you plan to get overall look of the cards you're comparing?
Averaging 16 games tells a lot more than taking a pick of 5 games, unless those 5 games are the only ones you play.
7970's starting to undercut 680's in price allready here. As low as $470.00
http://www.newegg.com/Product/Produc...CE&PageSize=20
Too little too late and a price cut doesn't magically make it any faster than it is, now does it.
Use your head for once, Slower card = cheaper. Glad AMD sales guys have realized that.
They had no pressure to lower the prices before the supply started to catch up with demand, while the cards have been steadily available lately, they have been moving stock all the time.
Regarding availability which you mentioned earlier ("nowhere to be seen"), I don't know about where you live, but if you're referring to US markets it could be the same as with GTX680 it at least was for quite some time - "the old continent" aka EU had plenty of supply (you could always find stock somewhere), while US was more starved.
While the GTX680 is somewhat faster in gaming, the compute performance took a nosedive, and it's in many cases losing to 580 too on that department, not to mention 7970 which is in a league of it's own in most cases for the time being.
http://muropaketti.com/artikkelit/na...md-vs-nvidia,2
The article is in finnish, but the graphs are universal as long as you remember "suurempi tulos parempi" means "bigger result is better" and "pienempi tulos parempi" means "smaller result is better"
Last edited by Kaotika; 04-17-2012 at 12:52 PM.
Wow. The 680 gets stomped in most of those tests.
That's exactly what we already do. Here's an example to show how crappy and dumb it is use averages; What if, out of those 16 games, you play 5 in particular and in those five the 7970 performs better, whilst the 680 wins the other 11? How does your averages account for that? It can't.
You talk like it's some kind of mind bending conundrum. TPU is the only site that does this, ALL the others manage fine without it.
Averages tell you nothing. How many people here are playing 16 different games at the same time CB? How many people actually split their spare time across 16 different games? I doubt anyone does. Most people probably play 2 or 3 at the most, so those averages apply to no-one.
Again, when we buy a new card, do we benchmark a whole bunch of applications and then proceed to work what the average FPS is across all those games? Of course not, because it's pointless and tell you nothing.
Yep, and applicability to gaming (even via DirectCompute) = nigh on zero
Swings and roundabouts really, when the GTX 580 was stomping around showing off it's GPGPU prowess at launch, it never really affected much as there was a lack of quality GPGPU apps to use it with. Likewise with the 7970, superb GPGPU power, but still a lack of quality GPGPU apps and surely even less to pick from because it doesn't use CUDA.
I would be more interested in how it runs my favourite game at the moment which is Battlefield 3 @ 1920 Full details with 4X AA.
7970 barely manages 50fps, while the GTX 680 gets over 65fps which is significantly more comfortable.
Posting Permissions
Reply With Quote
