Imouto
Golden Member
- Jul 6, 2011
- 1,241
- 2
- 81
Yes, 680 is faster out of the box. The only reason is because it has 15% higher clocks.But overall performance comes from a combination of things. Assuming an infinite amount of memory bandwidth and PCIE bandwidth, then performance comes down to how much work can be done per clock and how many clock cycles can be crammed into a given amount of time. The 680 is faster. Not hugely faster, but faster at the reference speeds.
Like I said earlier, I don't know what there is to argue about. A GPU comes out later than another and a near price point and is faster... I think we would expect that more often than not.
Yes, 680 is faster out of the box. The only reason is because it has 15% higher clocks.
AMD can release 7975 @1.1GHz and it will be faster card out of the box.
I would suggest you learn to read?
Congrats! You'll be one of the few people who can do Tri-SLI 480s @ 580 speeds vs. GTX680s in SLI at 1600P on a 5.0ghz LGA2011 i7. Drool.
Sure 925MHz against 1.05-1.1GHz. 680 is what 10% faster with 15% higher clocks?
Clocks correlation is nothing compared to voltage.Because TDP and clocks don't correlate.
Yes, the hell got frozen.There is a reason AMD released the stock clocks where they are at.
I think what you mean to point out is that even though nV is clocked higher, it is cooler, quieter, and uses less power.
Timing.
Although there was a relatively short time period between the releases of the chips, Verde and Pitcairn's bring-up, and to some extent qualification, have a reasonable level of leveraging going on so they are a little shortended in terms of initial engineering wafers back to product shipping. Actually setting the product "boundries" for Tahiti happened a while ago, on initial engineering material and few wafers out from the fab; Pitcairn and Verde on the other hand had their product boundries set when Tahiti production starts were already occuring and there is a very quick evolution in terms of understanding things with the new process / chips.
I guess the question you want to ask is whether, now that we know things have evolved, are we going back to re-look at Tahiti....
Yes, the hell got frozen.
NV has more power efficient card for gaming. NV did an excellent job.
Here is an answer from Dave Baumann why Tahiti had its clocks so low.
http://forum.beyond3d.com/showthread.php?p=1626557#post1626557
I am not sure how much power 1.1GHz 7970 would need. It depends if the voltage goes up or not.Selling a pre-OC's 7970 (perhaps binned) is not a great solution... it's a stopgap at best. See my post right above yours. Yes it will take the 7970 back on par with the 680, but it will also draw more power in the process, at least for the 3GB VRAM version and probably also the 1.5GB VRAM version. And it will have a smaller feature set, too. AMD has got to lower prices or else lose market share.
Selling a pre-OC's 7970 (perhaps binned) is not a great solution... it's a stopgap at best. See my post right above yours. Yes it will take the 7970 back on par with the 680, but it will also draw more power in the process, at least for the 3GB VRAM version and probably also the 1.5GB VRAM version. And it will have a smaller feature set, too. AMD has got to lower prices or else lose market share.
I am not sure how much power 1.1GHz 7970 would need. It depends if the voltage goes up or not.
I am not sure how much power 1.1GHz 7970 would need. It depends if the voltage goes up or not.
What smaller feature set? Are you talking about just psyhx? And since when power consumption is so important for high-end cards? 7970 is not bad when it comes to power consumption it draws much less power than GTX580, so its power consumption is not unreasonably high like GTX480 was. Even amd's dual gpu card drew less power. Even though power consumption gap was huge between Fermi and Cypress, Fermi was still considered the better card, which it undoubtedly is. GTX680 is just particularly good at that metric. But I agree that at current prices GTX680 is the better buy.
I'll try and look at my Kill-o-Watt tonight. I can say that I don't notice much difference between 925MHz and 1100MHz as far as fan noise and temp go. I'm sure it'll use a little more power, but things are great until I crank the voltage... then the fan becomes annoying quickly. I have not stopped at 1200MHz because that's as fast as I can go, I am at 1200Mhz because the card becomes just plain too loud.
Yeah with binning it is possible to suppress wattage. For instance, look at the GTX 460 FTW edition they got AT to use in their 68x0 review. http://www.anandtech.com/show/3988/the-use-of-evgas-geforce-gtx-460-ftw-in-last-nights-review
But binning costs time/money/effort and even if successful, they are at best on par with the 680 performance. Still not beating it. Still with higher wattage. Still with a smaller feature set, including lack of Adaptive Vsync. And what happens to the price of the non-binned SKUs?
Plus NVDA has a stronger brand name, and many people are either too fearful/lazy/ignorant or whatever to overclock and they will be happier letting GPU Boost take care of the nitty gritty details.
This all adds up to bad news for AMD. I do not see how AMD can hold its market share by merely matching 680's performance and price, but not its feature set or power draw. A $500 7970-1.5GB version isn't good enough. Even $450 is probably not good enough. $425 or less might be good enough, imho. The 3GB variant can probably be priced higher for those nuts who actually want that much VRAM. But most people will read the reviews and conclude that the 680's 2GB VRAM is enough for them.
A lot of small advantages add up to a significant one. I'm talking about ease of GPU Boosting for people who don't know how to OC or are scared to void their warranty and want the software to hold their hand. Lower power draw. Physx/CUDA. Adaptive Vsync (this is important to me and I bet a lot of others).
I do not believe Fermi was considered better than Cypress by the vast majority of people, though it's possible that a smaller majority or a minority of people did. Many others, though, joked about its power/thermal/watts and lousy price/perf relative to the competition.
7970@1.1GHz is faster than 680.
http://www.hardware.fr/articles/857-23/performances-gpu-boost-overclocking.html
Maybe I didn't remember that correctly I considered it as a better card than Cypress maybe that skewed my memory. I still had Cypress because I bought it full 5 months before Fermi even came out and upgrading was pointless back then because at launch GTX480 was just 15% faster than Cypress and that performance advantage came at a huge penalty to power consumption. The difference was much more relevant then it is between GTX680 and tahiti now because it made SLI GTX480 very loud, too loud for most people. For single card systems it was basically a non-issue.do not believe Fermi was considered better than Cypress by the vast majority of people, though it's possible that a smaller majority or a minority of people did. Many others, though, joked about its power/thermal/watts and lousy price/perf relative to the competition. And back then AMD had single-GPU Eyefinity all to itself.
blastingcap you are being overly dramatic. Tahiti is still a small chip. The whole card costs maybe 40$ more to produce so their margins are still reasonable given that's their high-end card. Back in 2008 nvidia competed with RV700 with 576mm2/460mm2 with way more complex PCB 448 vs 256 and somehow still made money. That was 323mm2 vs 576/460mm2 plus more complex PCB to boot. Now the difference is way smaller. GTX680 is more efficient per mm2 than tahiti and it would be an epic failure of similar magnitude to GF FX if it wasn't. Take into account that gtx680 was stripped of almost any compute features. It has only 1/24 DP performance and static scheduler.
That's the point. AMD can release a faster out of the box card in a matter of weeks/days.Based on the reviews I've seen you will need a bit more than 1.1GHz to match a similarly pushed 680's performance. More like 1.15 or 1.175GHz. If you're just aiming to match stock 680 performance (such as it is, since GPU Boost is still on), then you probably don't even need to hit 1.1GHz, something lower will do.
7970@1.1GHz is faster than 680.
http://www.hardware.fr/articles/857-23/performances-gpu-boost-overclocking.html
Yea, you keep flipping. Beating and not beating.Two times in a row now you should just read the post I made right above yours.