- Status
Skurge
Diamond Member
- Aug 17, 2009
- 5,195
- 1
- 71
The 7990? How could that be? Then a 7970 would be around GTX580 in performance (even lower when you assume near 100% CF-scaling).
Or Kepler?
@Boxleitnerb
I think apoppin, ment it as sarcasm, going by the innocently whistling smily.
Doesnt make sense that a 7990 is only 40% faster than a 580... a 6990 already is something like that.
I think apoppin, ment it as sarcasm, going by the innocently whistling smily.
Doesnt make sense that a 7990 is only 40% faster than a 580... a 6990 already is something like that.
dust
Golden Member
- Oct 13, 2008
- 1,328
- 2
- 71
@Arkadrel:
Why does it not make sense? HD5000 was not that large of a step versus G200 either. Maybe power goes even down a little. GTX580+40-45% (without cherrypicking) seems reasonable.
@Zed03:
The problem is AFR. Every dualchip card since the reintroduction of SLI in 2004 has used that. And they are not useless, you just have to know how to deal with it
Why does it not make sense? HD5000 was not that large of a step versus G200 either. Maybe power goes even down a little. GTX580+40-45% (without cherrypicking) seems reasonable.
@Zed03:
The problem is AFR. Every dualchip card since the reintroduction of SLI in 2004 has used that. And they are not useless, you just have to know how to deal with it
@Arkadrel:
Why does it not make sense? HD5000 was not that large of a step versus G200 either. Maybe power goes even down a little. GTX580+40-45% (without cherrypicking) seems reasonable.
@Zed03:
The problem is AFR. Every dualchip card since the reintroduction of SLI in 2004 has used that. And they are not useless, you just have to know how to deal with it
How to deal with it: Make sure the single card in your dual card setup hits vsync.
Then whats the point of the dual card?
OT answer:
What? What single card?
Microstutters appear most noticably when you are GPU bound. It is true that when using vsync often you shift the relative load towards your CPU, thus microstuttering lessens or is not noticed anymore. Similar thing if you leave vsync off. More fps, the timeframe deltas become so small that they are not significantly different from the average timeframe -> (almost) no microstutters.
Only when using settings that truly tax your cards, microstutters become a real problem. Usually that is below 40-50fps, depending on the game. And even then you can do something against it by using a fps limiter which has been proven to lessen or eliminate the stutters altogether. Before you say: What's the point of capping my fps - you only do it when you actually have stutters, i.e. low fps (due to GPU bottleneck). For instance if I have 40fps avg and 35fps min with really high settings, with SLI/CF that would feel like 25 maybe, really choppy. Cap the framerate at 35 and voila, smooth gameplay.
If you use such settings that you hit 60fps because you need 60fps, you don't have (that much) of a problem anyway.
Edit:
Why isn't the spoiler function working properly? Like actually hiding the text behind a button...
RussianSensation
Elite Member
- Sep 5, 2003
- 19,458
- 765
- 126
Ok, then proceed to buy an HD6970 then. No one is forcing you to buy a $700 graphics card. I am not going to buy a $500 graphics card until it falls in price, but I certaintly understand why it's going to be priced this high. Also, if the card is not 30% but 50-60% faster than HD6970, would that change your mind? Or would you need it to be at least 100% faster?
The world is a wonderful place full of options. We have people who take public transit to those who own a parking garage full of supercars.
Last edited:
I still thinking its going to be more than 50% faster than a 6970, which means 35% or so faster than a 580.
to me that doesnt seem so bad. Howmany people own a 580 and complain about it being slow? add 35% ontop of that, there are probably no games you wont be able to play 1920x resolution with more or less everything on.
to me that doesnt seem so bad. Howmany people own a 580 and complain about it being slow? add 35% ontop of that, there are probably no games you wont be able to play 1920x resolution with more or less everything on.
RussianSensation
Elite Member
- Sep 5, 2003
- 19,458
- 765
- 126
That's one way to look at it. I think a lot of people are looking at it from a context of spending $500-550 on a GPU generation that ultimately competes with Kepler. Since NV and AMD are committed to an 18-24 month generation cycle, you are faced with a decision to wait 6 months (or longer) to possibly get a much faster card in Kepler. Also, since most of the demanding games have been out (which means most people have been playing them using older hardware), chances are that if you already played those games just fine on your older hardware then you might as well wait 5-6 months until something more demanding arrives and see what Kepler brings.
If for example, HD7970 is 35% faster on average over GTX580, but GTX680 is 75-80% faster, then suddenly that $550 purchase will seem rather poor. I have no idea what the performance will be of either card, nor their pricing, just putting the 35% figure into perspective of how it might not be that great. Of course in the context of a $450-500 GTX580, getting a card that's 35% faster for almost the same price would be great.
But as you know, enthusiasts don't like waiting. They'll upgrade to the next best thing, even if it means upgrading every 6 months. Can't discount the possibility that HD7970 might be more than 35% faster than GTX580..... :awe:
Last edited:
Question is if nvidia's kepler is really gonna end up that fast.
It ll probably end up 15% faster than the 7970, and be 6months or more behinde in launch dates.
Thats if Nvidia only aims for a 55% faster card than their 580.
Thats just as likely to happend.
It ll probably end up 15% faster than the 7970, and be 6months or more behinde in launch dates.
Thats if Nvidia only aims for a 55% faster card than their 580.
Thats just as likely to happend.
nVidia cards always under perform at launch. Look at Fermi (worst performance per watt in the business), GTX 560 (which required ti variant to be competitive), etc..
We seem to assume that both sides release closely as a fact of nature. What if we're entering a world where they split the generational release times so as to constantly leapfrog each other? Out of phase, in other words. Waiting for the next great card becomes a LOT more complex.
RussianSensation
Elite Member
- Sep 5, 2003
- 19,458
- 765
- 126
If major generations are slated for an 18-24 months schedule, then the only leapfrogging expected is for refreshes (such as GTX480--> 580 and HD5870 --> HD6970). Do you suggest we will see HD7970 with 50% more performance over 6970, and then GTX680 beating the 7970 by 15%, and then in 6 months an AMD card that beats 680 by another 15%? Last time AMD beat Fermi by 6 months but they launched HD5000 series in the Fall of 2009, capatalizing on the huge advantage of the holiday season. This time, HD7000 series is launching in historically slow Q1 2012 (Dec 22nd is the launch date but volumes probably won't pick up until Q1). You are overstating the significance of launching early outside of peak sales season and understating NV's ability to respond with a competitive offering. They did just fine with Fermi. The market for high-end GPUs is only 14%. NV can lower prices on their current GTX500 series and still remain competitive. They stand to lose high-end performance in the short-term.
Last edited:
No, the real question is how much Tahiti info out there is BS fake crap, how much is opportunistic marketing, and how much is realistic real world performance.
The|Hunter
Member
- Dec 5, 2011
- 145
- 1
- 81
I never assumed that AMD launched early. I DO assume that ANY party launches as soon as possible. Why is it assumed to be 'only' a six month lead? If more, then the 18-24 month generational split could be achievable. We have 1 past example. This time might be different. Don't assume that only Nvidia learnt anything from the previous scenario.
I have a solution for this, buy a HD 7970 now and then sell it when GTX 680 comes out :awe:
If the hd 7900 are really coming out on the 22nd of this month -paper launch as it may be- it would still be smart to send some pieces to reviewers for early benchmarks, otherwise their potential costumers won't have enough time to inspect the waters before they take out their wallet ... what with christmas being a few days away and everything....
Last edited:
I'm guessing the sites who have a good relationship with AMD (read: They actually honor NDAs) already have the cards and are finishing up the reviews for when the NDA lifts on Thursday.
toyota
Lifer
- Apr 15, 2001
- 12,957
- 1
- 0
still showing 32 ROPs. maybe they have some really odd architecture for the GCN cards.
- Status
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter