Is there any software avalible to push tahiti beyond overdrive limits?
Is there any software avalible to push tahiti beyond overdrive limits?
I assume Afterburner would work just fine at launch, if anything it'll just need to be updated.Quote:
Originally Posted by majkel84
For AMD's marketing purposes why would they shoot themselves in the foot with a GTX580 3GB? :)Quote:
Originally Posted by paulbagz
Barely beating a 580? hmm..... Lets hope thats not true, if so im sure many people will be holding off for GTX 780/770
Huh? The results so far look like a carbon copy of 5870 launch..If anything even better if the slides are even half truthfull /realisticQuote:
Originally Posted by nascasho
Looks like 7990 is going to be a beast..
and with a beastly price... ouch.
The 78XX are just die shrunk caymans, that's why it'll be similar to 6970 ;)Quote:
Originally Posted by 6thsurvivor
Meh, the 7970 results aren't really impressive. Maybe if they moved to XDR2 results could have been more interesting.
If true awesome results for the 7970
When compared to the 6970(cayman) vs gtx 580, those graphs seem off considering the 6970 is using 11.11B drivers. These results seem especially off for the Civ 5 results and Batman Arkham city results. It seems like there is a huge amount of fudge factor and like typical marketing slides, are pretty much worthless. Better to wait for real results before making judgement.
Clearly lol, even if not reference or not, better to at least compare them on some semi-relevant plane rather then one with double the VRAM.Quote:
Originally Posted by minpayne
-PB
Yes, at same time, just a precision, it seems there's only 1 driver usable now for the 7970 ( you can download it if you access on NDA.AMD.COM ). And with the big change on the code of the gpu ( and so how the driver work with the hardware) , i suspect some drivers will be needed ...Quote:
Originally Posted by tajoh111
I personally wait on Anandtech reviews... not cause i like how they bench the card ( not the case ), but cause with the excellent preview of gcn made by them in June, i hope they will go in deep of the architecture this time.
Well, that's really not the case when you see things with green tinted rose glasses. :lol:Quote:
Originally Posted by mAJORD
I don't think that's true, 78xx will use Pitcairn chip, and that baby is a member of GCN (Graphic Core Next) family, along with Tahiti, Cape Verde, & New Zealand. :)Quote:
Originally Posted by iddqd
But then all the people who claim more than 1gb or whatever arbitrary number they choose "is enough" would be complaining ;)Quote:
Originally Posted by paulbagz
Will we see this again
http://img706.imageshack.us/img706/4198/67104522.jpg
:D
>__________________________________________<Quote:
Originally Posted by STEvil
Can't please everyone kinda situation eh? :P
-PB
It'd definitely stir things up considering we haven't seen that happen since the x1950xtx... I don't think we will this time though, considering a big reason for those differences in performance numbers between the 580 and 7970 in that slide is that 1.5gb isn't enough for high aa at 2560 and damn well not enough for 3+ screens with AA, and the 580 isn't going to be what AMD are up against.Quote:
Originally Posted by cold2010
That, I'm waiting in anticipation for this show-down between both camps new cards. GPU cycles (not refreshes, but the initial releases) are my favorite time of the hardware year.
Probably not, each process switch makes each mm2 more important. So if AMD continues to use less die space and both architectures are reasonably close on performance per transistor then each mm2 AMD don't use gives nvidia a bigger advantage if they continue to use the 'All in' strategy they've used for some time. On 22nm AMD will either have to match nvidia's die size or have a superior architecture because giving up 100mm2 is going to hurt. Sure AMD can claw some of it back with their dual GPU cards, but when it comes down to it, people rarely compare the 5970 to the 480.Quote:
Originally Posted by cold2010
True, but the 580GTX in 3GB configuration isn't really a vanilla card, a few card makers have released them but then some people will release 1.5GB versions of the 7950 too, like they released 1GB versions of the 6950. So the omission is fair and valid.Quote:
Originally Posted by paulbagz
However I don't think it doesn't really matters, nVidia have made more of a effort this generation to insure they do not have the same issues they had with 40nm, they'll most likely go large again, and I think the gap in release between the two might be shorter this time.
I'm sure (or rather I hope) that when professional reviewers put out their numbers we'll see some benchmarks that showcase the difference the extra memory makes.
Lets not forget that the game engine can be as much at fault for poor performance at high resolution as can be the lack of vram.
They ought to have just kept on using the 5870 shroud, that thing was so much nicer than the HD 6000s / 7000s shrouds.
eyefinity 3 screens and such then a 3gb card makes more sense and need to provide the performance there.Quote:
Originally Posted by DilTech
1920x1080 then todays cards works well enough for.
what typically happens is, do the card and the configuration you use it for make a difference with the new card or not?
is that difference noticeable for you or not?
it not, then the new card isnt worth it.
benchmarks aside, but if the results dosnt gets noticed in your machine, why buy new then?
Was read a bit the beyond3D forum, and one have see something interesting... 2x 6 Pin = 7950 ?Quote:
Originally Posted by minpayne
About the red pcb .. well dont take too much care about it, thoses are AMD cards, surely HIS, MSI, etc will have their own black pcb for the retail cards. ( simply AMD/ATI made by AMD/ATI are always red pcb )
Weird that people are disappointed its pretty good performance IMO, its always been like this.
4870=<GTX 280 Link
5870=<GTX 480 Link
6970=<GTX 580 Link
7970=<GTX 780/790 "Which ever one has 1500 cores"
AMD has not done a big daddy GPU for quite some time, they rather go with the dual GPU route which is a super easy with smaller cores its more about mm2 than freq.
EDIT: Things like GTX285, 4890, 6930, etc do not matter since they are freaks and mutants :P
4870X2 had it all black I remember, but again it's of very little relevance!