a 770 is the SAME as your card clock for clock and the gpu cannot oc any higher than a typical 680. only the memory is better and can clock higher. oc both the 680 and 770 and you are looking at maybe a 5% advantage overall from the better memory.Wow, the 770 is quite a bit faster than my 680 than I thought. Not enough for me to want to buy one though. I think I'll wait till Maxwell.
According to 3dCenter's Performance Index (compilation of 6-10 reviews) here's the performance breakout at 1080p 4xAA in relation to how much faster the 780 is than the following cards:
770 - 16% faster
7970GHz - 13% faster
680 - 22% faster
670 - 33% faster
7950 Boost - 38% faster
The top two could overclock to match a stock 780 but it would have to be a pretty good overclock (~1200Mhz for the 7970 and ~1350Mhz for the 770). A 680 with an uber oc would get within spitting distance. The 670 and 7950 would both need amazing chips to get in the same neighborhood as a 780 and that is assuming near linear performance increases.
Congrats on the cards by the way. Sounds like you're enjoying them.
Do you have a link for that performance index?
It may need less than 1350Mhz to match the 780. I was just extrapolating from their relative performance. If the 780 is 16% faster than the 770, you'd need to overclock the 770 at least 16% to match it. The 770's average clockspeed in the Anandtech and Hardware France reviews was 1136Mhz and I multiplied that by 1.16 and added a little to account for the non-linear performance to overclocking relationship. Might be closer to 1320-1330Mhz if the 770 scales extremely well.I doubt it would be that much for the 770. Looking at the HardOCP review of the MSI Lightning, although the 770 at 1241 was clocked lower than the 7970 at 1275, it was slightly faster in Metro Last Light, Crysis 3, slower in Tomb Raider and the same speed as the 7970 in Far Cry 3.
that is incorrect. oced Titan will hit its TDP quicker than an oced non reference 780. basically the 780 can end up just as fast in some games and only 2-3% slower overall.I hate OC vs stock comparisons when they compare 1 stock card to 10 overclocked ones; I think the more valid comparison is comparing everything overclocked - the 780 exceeds Titan performance when OC'ed. But of course the Titan can be OC'ed as well - and then you're right back to where things started. The titan maintains the same performance gap above the 780. (which admittedly is a small 10% gap) So what does overclocking change? Nothing. The Titan is still 10% faster when you overclock both cards.
I guess, over time I put less thought in it. It's nearly a guarantee that you can OC a SKU to reach near the next SKU performance (eg OC a 760 to match the 770), but you can still OC the other cards as well to maintain the exact same performance gap. And they will still be faster by the same or higher margin. I mean, you can OC a 760 to reach 770 performance if you get a good chip. But then you can turn around and OC the 770 and then, again, get the same 15% performance increase over the 760...... Would have been nice if everything on that chart had been overclocked. I just don't see much value in an OC vs stock chart (every 770 on the chart is factory OC'ed) unless they OC everything across the board.
I kinda like Linustechtips' methodology, every review he does overclocks every card to it's maximum potential in every comparison. Why? It's free performance, and nearly every SKU has 10-15% headroom from overclocking - And it doesn't muddy the waters by comparing a ton of OC'ed cards to a stock card. Doing stock vs OC'ed doesn't really paint a clear picture of the maximum performance potential of every SKU, because even the worst overclockers still get a decent +10/15% from OC'ing.
Don't get me wrong, the 770 is an awesome card for a great price. Definitely one of the best price/perf cards to be released recently. And when you think about it, you can get a 780 for 650$ or 2 770's for 800 bucks - Clearly the 770SLI will trounce the single 780 in terms of performance by a huge margin for only 100-150$ more. That being said, I suppose i'm just ambivalent towards the methodology used in some OC vs stock reviews...
that is incorrect. oced Titan will hit its TDP quicker than an oced non reference 780. basically the 780 can end up just as fast in some games and only 2-3% slower overall.
yes I am referring to stock BIOS. and also the point is that not every gap is equal either. a 670 and 680 oced will be within 3-4% of each other. an oced 660ti will still trail an oced 670 by at least 10-12%. some cards just have tons of overclocking headroom too such as the 7950 and 7850 as I am sure you know.Ah! Well it really depends to be honest. There's a modded vBIOS for the Titan which substantially increases available TDP for overclocking, and i've seen some pretty beastly overclocked benchmark scores at OCN. Certainly nothing I can match with my 780 even with my best overclock. Aside from this, I think the Titan is a special case designed for multi monitor gaming (given that it has 6GB of VRAM). I don't think the average single monitor gamer would buy it, although i'm sure there are a few.
But with the stock BIOS you're perhaps correct. The main point i'm driving at is that every SKU has overclocking headroom out of the box and comparing one stock card to 10 OC'ed cards in a review doesn't paint a proper picture. I really like a methodology with everything overclocked, if one card is being overclocked, you know? I mean, you can overclock a 760 to match a 770, a 770 to match the 780, but when you overclock everything across the board, nothing changes. The performance differentials between each SKU remains the same if you overclock everything.
yes I am referring to stock BIOS. and also the point is that not every gap is equal either. a 670 and 680 oced will be within 3-4% of each other. an oced 660ti will still trail an oced 670 by at least 10-12%. some cards just have tons of overclocking headroom too such as the 7950 and 7850 as I am sure you know.
$900 for cards when the next gen is around the corner?
You stuck it out with your 580's this long and at the end of the cycle you decide to buy 1 1/2 year old hardware at nearly the same price it's been available for nearly two years?
This is puzzling to say the least. Return those cards and buy a single 7970 for $300 or 7950 for $250 and overclock them. Put the money in a rainy day fund, kids savings account for school or vehicle fund for when they are old enough, take a vacation to the beach with your family, etc...
It may need less than 1350Mhz to match the 780. I was just extrapolating from their relative performance. If the 780 is 16% faster than the 770, you'd need to overclock the 770 at least 16% to match it. The 770's average clockspeed in the Anandtech and Hardware France reviews was 1136Mhz and I multiplied that by 1.16 and added a little to account for the non-linear performance to overclocking relationship. Might be closer to 1320-1330Mhz if the 770 scales extremely well.
Same for the 7970GHz. 1050Mhz*1.13 = 1187Mhz. The 7970 might need 1200-1225Mhz depending on scaling.
You have to consider boost. The GTX 770 performance is not at its reference clock, but rather at its boost clock(s).My point was that if the GTX 770 was outperforming or matching the 7970 at a lower clock speed (Tomb Raider being the exception), how can it require a higher clock than the 7970 to match the 780?
Yeah, but as I said in this thread, I don't like to buy first generation hardware. I got burned doing that with the GTX 480, so I usually wait until the re-tweaked version comes out..
So I won't upgrade my 770s until the tweaked Maxwell cards come out, sometime in 2015.. I have no doubt they will easily last till then..
You have to consider boost. The GTX 770 performance is not at its reference clock, but rather at its boost clock(s).
Yeah, but as I said in this thread, I don't like to buy first generation hardware. I got burned doing that with the GTX 480, so I usually wait until the re-tweaked version comes out..
So I won't upgrade my 770s until the tweaked Maxwell cards come out, sometime in 2015.. I have no doubt they will easily last till then..
You do realize you could have bought the same cards for nearly the same price over a year ago and you decided it was a good time to buy them at the same price at the end of the cycle.
No, a year ago the 680 4GB editions were close to 600, or over 600 USD particularly for the enthusiast models like the FTW, Classified, OC etcetera.
Brand new GTX 680 Classifieds are still selling for more than what I paid for mine, despite being slower.
You must be confusing them with the 670s, because the 680s were way more expensive. I just googled the launch price for the GTX 680 4gb FTW and it was a whopping 629 USD.....and the Classified was around 660 USD.
Also as I said earlier, they are still priced higher than my 770 4gb cards brand new at many retailers.
http://www.fudzilla.com/home/item/27093-evga-launches-gtx-680-ftw-edition-graphics-cards
My point was that if the GTX 770 was outperforming or matching the 7970 at a lower clock speed (Tomb Raider being the exception), how can it require a higher clock than the 7970 to match the 780?
That's what I thought as well.Two reasons, 1) the 770 starts out at a higher clock (1136 vs 1050) and 2) the gap between it and the 780 is slightly bigger than it is between the 780 and 7970Ghz (16% vs 13%).