GTX 780 Ti reviews

[blastingcap]

I hope you are serious on this one, because more choice is always better then less choice.
That is what happens in CPU market. You can get your CPU without heat-sink, you can buy it with crappy cooler included, or you can get one with manufacturer banded watercooling kit! You can get CPU with HT or without it, with gpu included or without it. Everything a the price. It is your choice to make.

Oh man, if I could buy CPUs without stock coolers I would, but a lot of times I did not have the choice, like when I bought a bunch of Ivy Bridge at launch. I now have a nice collection of 1155/1156 stock Intel coolers just sitting around.

But yeah, if they did this with GPUs that would be good if you got a substantial price break on the base card as a result. I'm sure all the watercooling people would love that option.

 

[ams23]

Regardless of the "how", it is how the R9 290 comes and how the majority of users will use the stock card.

Well we really don't know how the "majority" of users will use the stock card. If many pro reviewers find "Uber" mode to be annoyingly loud (referring specifically to the ref. design), then why would the "majority" of users not find that to be the case?

 

[chimaxi83]

Well we really don't know how the "majority" of users will use the stock card. If many pro reviewers find "Uber" mode to be annoyingly loud (referring specifically to the ref. design), then why would the "majority" of users not find that to be the case?

Because those reviewers use an open test bench with measuring device 1 foot from the card?

 

[ams23]

But is it more than just noise - is the problem that the 290 generates too much heat for any cooler to manage?

What I found somewhat remarkable is that GTX 780 Ti OC (http://tpucdn.com/reviews/NVIDIA/GeForce_GTX_780_Ti/images/perf_oc.gif) is 40% faster than R9 290X OC in "Quiet" mode (http://tpucdn.com/reviews/AMD/R9_290X/images/perf_oc.gif)! That is a huge difference, all at stock voltages and fan speeds too where the GTX 780 Ti is quieter and less power hungry than R9 290X. Even compared to "Uber"[Loud] mode with OC, the performance difference is still 22%. GK110 appears to have pretty high performance headroom all things considered.

 

[ams23]

Because those reviewers use an open test bench with measuring device 1 foot from the card?

Ryan Smith @ Anandtech said that the noise is even louder when closing up the case (probably due to sound pressure waves inside the case and due to more distortions with the case closed).

 

[Erenhardt]

I bet majority of 290 users will use their 290X like this one here:

780Ti amazing OC thanks to good cooler.
Overclocker edition 780 GHz B1 stepping max OC out of 2 tested samples was 1123Mhz which was stock boosting clock.
290X with its terrible leafblower clocked to 1220.
290 clocked up to 1220 mhz aswell. That was cooled with mk-26 aftermarket heatsink.
I have a feeling that good aftermarket cooler will benefit 290X the most out of these.

 

[Erenhardt]

What I found somewhat remarkable is that GTX 780 Ti OC (http://tpucdn.com/reviews/NVIDIA/GeForce_GTX_780_Ti/images/perf_oc.gif) is 40% faster than R9 290X OC in "Quiet" mode (http://tpucdn.com/reviews/AMD/R9_290X/images/perf_oc.gif)! That is a huge difference, all at stock voltages and fan speeds too where the GTX 780 Ti is quieter and less power hungry than R9 290X. Even compared to "Uber"[Loud] mode with OC, the performance difference is still 22%. GK110 appears to have pretty high performance headroom all things considered.

You are not comparing two different reviews are you? Have you noticed that stock 290X is faster in one review then in the other?

 

[ams23]

You are not comparing two different reviews are you? Have you noticed that stock 290X is faster in one review then in the other?

Umm, look again, there is no difference in R9 290X scores between the two data sets (this is the exact same reviewer). Note that I am NOT using the actual OC'd score in "Quiet" mode because performance when OC'd actually went down compared to no overclock, I am using the higher score in "Quiet" mode.

 

[Erenhardt]

My bad, I colors misguided me.
Anyway, one card is crippled by its cooler, one is not.

 

[Nomanor]

Looks like aftermarket 290 in CF is the way to go.

$400 x 2 vs. $700

 

[ams23]

Anyway, one card is crippled by its cooler, one is not.

No doubt about that. AMD definitely should have spent more time on the reference model design.

 

[blackened23]

My bad, I colors misguided me.
Anyway, one card is crippled by its cooler, one is not.

And this is why no one should defend the design decision at AMD to put that stupid thing on there. At stock voltage the 780ti and 290(x) are about even in terms of power draw (NOT overclocked or overvolted) - the Titan and 780ti may draw 5-10 watts less, or sometimes they're even in gaming loads. Yet the 780ti has way better acoustics. With this being the case it's pretty obvious that AMD could have and should have done better with their cooler, especially with the card not running at full boosts while in quiet mode.

 

[Erenhardt]

And this is why no one should defend the design decision at AMD to put that stupid thing on there. At stock voltage the 780ti and 290(x) are about even in terms of power draw (NOT overclocked or overvolted) yet the 780ti has way better acoustics. With this being the case it's pretty obvious that AMD could have and should have done better with their cooler, especially with the card not running at full boosts while in quiet mode.

It is not the place to discuss if amd should or should put better cooler.

780Ti seems like a very good card. Shame the price is not so good. I expected nv to pick up the ball and price 780ti in direct competition to 290x, forcing amd to react the same way amd made nvidia to react.

 

[PPB]

G-sync lowers frame latency to the eyes as well. I'm betting competitive players will care about that. With G-Sync, the image is drawn immediately when it reaches the monitor, not whenever the monitor feels like drawing it (how monitor refresh rates work now).

Gsync at best can have the same frame latency than a vanilla 144hz monitor running at as many fps. To achieve that, you have to reach steady 144fps.

Dont be naive to think that because your GPU could dump like 200 FPS in a game, suddenly your monitor would refresh at the same pace. There is a limit, and that limit is dictated by the current max refresh rate the monitor has before the mod. So, like I was saying, competitive players would rather dump that extra money required for having gsync in better hardware and assure constant 144 fps.

One of mantle's purposes is to eliminate the CPU bottleneck as much as possible. You can have as much GPU power you want, but at 144hz gaming, even with the deepest pockets in the world, no CPU would assure you +120fps minimums on AAA games. If mantle could aliviate that bottleneck, even if it doesnt boost GPU perf that much, it's a better sell for the competitive players than gsync.

 

[Cookie Monster]

$699 seems abit too much, I personally think $549~599 is the best price for this variant of the Ti. Instead nVIDIA is prepping up that "black" edition followed by rather large pricetag..

If they want to command a price of $699, perhaps they should of released a 6GB version of the GK110 with the TDP set at 300W and a higher operating temp instead of 83C.

This version of the GTX780 Ti I think is good all round, but if they've gone for a the best performance only with disregard to everything else it would most likely end distance itself alittle from the R290X.

 

[lavaheadache]

Paid the tax lava?

nope, I'm officially not gonna support Massachusetts new law about taxing Amazon purchases, even if it means I miss out on one of the major perks of my Prime account.

Bought from Tiger Direct with next day air ($20). Evga vanilla. I may gets ballsy when the card arrives and rip the bios from it and try to "stick" it on a 780. I'm 99.999999999999997% sure it won't work but without risk there is no reward.

I will definitely be tweaking the bios with Kepler Bios Tweaker as soon as I determine the card can oc a little.

 

[Subyman]

But is it more than just noise - is the problem that the 290 generates too much heat for any cooler to manage?

If it was just noise, why wouldn't he just have said they use a cheap fan and you need to buy an aftermarket one? Surely it's more than just a noisy fan that he is condemning the entire card over, I mean his review is completely negative, and he reminds you over and over using very extreme negative language repetetively across multiple pages from the first page to the last.

Personally I think it's nice to have options, where I can save $100 on the cost of the card because a cheap fan was installed, then spend what I decide on a fancier cooler if I want. I mean, if the video card comes with an expensive fan that forces the card to be more expensive, wouldn't I be upset if I'm simply removing that entire fan to put the card under water cooling?

The only reason to run the card in "quiet" mode was noise, hence the name "quiet" mode. Practically every other review site ran the card in uber because that's how an enthusiast would run it. Who buys a $550 card to run it on the lowest setting? Why compare nVidia's flagship to AMD's flagship on a neutered setting? Make note of the noise, but don't use that as the basis to run the card on the lowest performance setting when testing and providing analysis.

It's not like he is modding the card, its a simple toggle to go to the highest setting for testing.

 

[chimaxi83]

nope, I'm officially not gonna support Massachusetts new law about taxing Amazon purchases, even if it means I miss out on one of the major perks of my Prime account.

Bought from Tiger Direct with next day air ($20). Evga vanilla. I may gets ballsy when the card arrives and rip the bios from it and try to "stick" it on a 780. I'm 99.999999999999997% sure it won't work but without risk there is no reward.

I will definitely be tweaking the bios with Kepler Bios Tweaker as soon as I determine the card can oc a little.

Congrats man! I seriously debated picking one up because I do like PhysX sometimes. Two 290 for ~$760 won me though. Grats and have fun!

 

[Subyman]

Well we really don't know how the "majority" of users will use the stock card. If many pro reviewers find "Uber" mode to be annoyingly loud (referring specifically to the ref. design), then why would the "majority" of users not find that to be the case?

My opinion is that reviewers should run the cards at their maximum stock capabilities. We kind of went through this already when nvidia introduced Boost since the card was up clocking itself, but ti was agreed that since that was the stock option to maximize performance then it was kosher. IMO, cards should be tested in their best form: newest drivers, advertised features tested, and highest stock settings enabled. We care about the best possible settings. I can determine for myself from the noise charts if that is a problem, but analyzing a card at a lower setting because the noise is subjectively too high for the reviewer is strange. Make note of the noise in the analysis, but don't effectively run the card down clocked to meet some hypothetical noise threshold.

Anyway, I'm probably taking too hard of a stance on it, but it was strange reading the analysis of a "quiet mode" R9 290X versus the 780 TI when the graphs clearly showed the "uber" mode doing quite well.

 

[lavaheadache]

Congrats man! I seriously debated picking one up because I do like PhysX sometimes. Two 290 for ~$760 won me though. Grats and have fun!

that will be quite the rig too. Now what to do with these two 780's?

 

[skipsneeky2]

that will be quite the rig too. Now what to do with these two 780's?

Donate one to me.

 

[Enigmoid]

$699 seems abit too much, I personally think $549~599 is the best price for this variant of the Ti. Instead nVIDIA is prepping up that "black" edition followed by rather large pricetag..

If they want to command a price of $699, perhaps they should of released a 6GB version of the GK110 with the TDP set at 300W and a higher operating temp instead of 83C.

This version of the GTX780 Ti I think is good all round, but if they've gone for a the best performance only with disregard to everything else it would most likely end distance itself alittle from the R290X.

Way too low. With the better cooler and performance the 780 Ti is worth around $629-649. The cheapest R290X I can find is ~$565. With the better cooler, shield coupon and three game bundle (Worth currently around $50) the 780 Ti would be very well priced at $649.

 

[Subyman]

Way too low. With the better cooler and performance the 780 Ti is worth around $629-649. The cheapest R290X I can find is ~$565. With the better cooler, shield coupon and three game bundle (Worth currently around $50) the 780 Ti would be very well priced at $649.

The Ti can be priced high, its the flagship. What's kind of flopping in the wind is the 780 pricing $499 for R9 290 performance. People that are noise sensitive may jump for the $100 more, but once custom coolers come out, its harder to spend the extra.

 

[MrK6]

nope, I'm officially not gonna support Massachusetts new law about taxing Amazon purchases, even if it means I miss out on one of the major perks of my Prime account.

Bought from Tiger Direct with next day air ($20). Evga vanilla. I may gets ballsy when the card arrives and rip the bios from it and try to "stick" it on a 780. I'm 99.999999999999997% sure it won't work but without risk there is no reward.

I will definitely be tweaking the bios with Kepler Bios Tweaker as soon as I determine the card can oc a little.

I'm on board, I won't renew my Prime membership either. Good thing there's still Newegg! Looking forward to seeing what you can do with the card, lava.

 

[Silverforce11]

So its ~7-9% faster than Titan, spot on from the prior leaks of Firestrike and 3DMark benches. Uses similar power to the R290X, close enough in performance at 1600p. Again, the defining feature here is NV's excellent reference cooler. It allows the card to run cooler, and the chip running cool means less power leakage.

This will indeed be a battle between R290X custom cooled cards vs the 780Ti. Once R290X is running at 70C and constant 1ghz boost (as opposed to throttling to ~900mhz even in uber mode), its going to equal the 780Ti in performance and power use. Interesting times ahead!