The GTX 780, 770, 760 ti Thread *First review leaked $700+?*

[toyota]

if its stable in a few loops of 3dmark11, AvP and Heaven then its stable in my expereince.

 

[24601]

if its stable in a few loops of 3dmark11, AvP and Heaven then its stable in my expereince.

Then you have not been through the nightmare that is quantifying stable clocks on a 7950/7970.

I had this exact same problem on my 460 which is why I have such experience with ROP issues.

ROP Bad. AMD need fix. AMD no money to refresh to fix. Communism sad.

 

[badb0y]

What kind of stress testing do you guys do? I wasn't able to hit 1300 core (I crashed on my 5-6th hour of Heaven) but when I overclock I run Heaven for 8 hours or so and that hasn't failed me yet. I used to use Furmark but it seems like the card throttles itself, I am also on water so maybe that has something to do with it. MrK6's 7970 is at 1350 core and I remember another member of our forums had his at 1375 core. All under water though I wouldn't attempt these types of overclocks with stock coolers.

 

[blackened23]

LOL another thread going to disaster-zone. This is just describing any overclocking. Stable in some cases, but not in others ? Your overclock is not stable, dial it down until it's stable in all cases.

Yeah, i'm anxious to talk about the GTX 700 again. I'm pretty happy if it's landing in the 550$ area, which - if it is indeed 500-600EU including VAT, it should fall in that 550USD price range, perhaps 600.

 

[badb0y]

At $500-$550 this really throws a wrench on my plans for waiting for 20nm.

:hmm:

 

[24601]

What kind of stress testing do you guys do? I wasn't able to hit 1300 core (I crashed on my 5-6th hour of Heaven) but when I overclock I run Heaven for 8 hours or so and that hasn't failed me yet. I used to use Furmark but it seems like the card throttles itself, I am also on water so maybe that has something to do with it. MrK6's 7970 is at 1350 core and I remember another member of our forums had his at 1375 core. All under water though I wouldn't attempt these types of overclocks with stock coolers.

Look at my previous posts in this thread for my specific ROP based testing for 7xxx series.

http://forums.anandtech.com/showpost.php?p=35024191&postcount=723

It's hard, but doable as long as you meet the conditions I laid out in that post.

The reason these crashes are so annoying is because of their almost Phantom nature, you can be running the games that don't stress ROPs and then maybe once every 24 hours you get a few seconds when the game does, and if this is a multiplayer game you are 100% screwed for that match/moment. Losing your progress in a single player game is maddening as well.

 

[Cloudfire777]

At $500-$550 this really throws a wrench on my plans for waiting for 20nm.

:hmm:

20nm won`t be here until Q1 2014 anyways

 

[tviceman]

At $500-$550 this really throws a wrench on my plans for waiting for 20nm.

:hmm:

Same here.

 

[rgallant]

Why in the world is it an even remotely close to fair comparison to compare one massively overclocked card to one at stock speeds? And the 7970HD was absolutely not a "huge increase" over an overclocked gtx580.

What in the world are you smoking??

not as bad as the gtx 6xx reviewers using cards that overclocks on auto against a fixed

clocked 580 and the reviews not showing or would not show the clock's of the 6xx ran at

because it came out of the box that way , but posted the un boosted stock clocks wtf,

-would not increase the clocks of the 580 by the same % , but said they could not stop

the 6xx boost ,might not look as good for the 6xxx .imo.

 

[omeds]

Because that is the default behaviour of GTX680 out of the box.. it would be worse to artificially cap the clocks, don't you think?..

 

[toyota]

Because that is the default behaviour of GTX680 out of the box.. it would be worse to artificially cap the clocks, don't you think?..

its cheating just like Intel does with the i7 and i5. <sarcasm>

funny now that AMD cpus and gpus do that too...

 

[3DVagabond]

its cheating just like Intel does with the i7 and i5. <sarcasm>

funny now that AMD cpus and gpus do that too...

What? Why are you posting stuff like this? You know that nVidia GPU boost isn't at all the same as what AMD does with their CPU/GPU boost. Nor Intel with theirs.

 

[toyota]

What? Why are you posting stuff like this? You know that nVidia GPU boost isn't at all the same as what AMD does with their CPU/GPU boost. Nor Intel with theirs.

yes its not exactly the same but the main point was that when Intel came out with the i7 so many morons called it cheating too. as for Kepler, anybody capable of reading and that keeps up with hardware knows max boost is going be higher than the advertised typical boost clock. and most cards with the same advertised clocks will not vary all that wildly for max boost. take the stock gtx680 which lists 1058 for typical boost. most cards have max boost of 1100-1150. big deal

 

[3DVagabond]

yes its not exactly the same but the main point was that when Intel came out with the i7 so many morons called it cheating too. as for Kepler, anybody capable of reading and that keeps up with hardware knows max boost is going be higher than the advertised typical boost clock. and most cards with the same advertised clocks will not vary all that wildly for max boost. take the stock gtx680 which lists 1058 for typical boost. most cards have max boost of 1100-1150. big deal

True enough about the complaining about Intel's turbo feature. Just like people complained about comparing 8 core FX CPU's against 4 core Intel CPU's.

Some review cards boosted much higher though, and that's the big difference. How many 680's do people own that will boost to 1300MHz? Intel and AMD's boost features boost to a predetermined clock which is the same across all chips.

 

[toyota]

True enough about the complaining about Intel's turbo feature. Just like people complained about comparing 8 core FX CPU's against 4 core Intel CPU's.

Some review cards boosted much higher though, and that's the big difference. How many 680's do people own that will boost to 1300MHz? Intel and AMD's boost features boost to a predetermined clock which is the same across all chips.

"Kyle saw a GTX 680 sample card reach over 1300MHz running live demos but it could not sustain this clock"

I agree 1300 does sound high but briefly hitting that in one game does not mean much though. we all know boost can vary a bit from game to game and max boost may not last or even be needed long. it is more confusing but you still get a minuscule overall average difference from cards based on the same clocks.

 

[ICDP]

"Kyle saw a GTX 680 sample card reach over 1300MHz running live demos but it could not sustain this clock"

I agree 1300 does sound high but briefly hitting that in one game does not mean much though. we all know boost can vary a bit from game to game and max boost may not last or even be needed long. it is more confusing but you still get a minuscule overall average difference from cards based on the same clocks.

Your definition of miniscule may be different than someone elses. All of the following were reference "stock" GTX680s with zero overclocking out of the box.

EVGA 2GB
Boost clock speed. 1120

Gainward 2GB
Boost clock speed: 1136

KFA 2GB
Boost clock speed: 1086

The KFA was ~5% slower than the Gainward card despite being stock cards. It might not seem like much but such margins have been used to judge GPUs in reviews.

 

[BallaTheFeared]

5% lower clock speed is probably a 2% difference in fps.


Dynamic boost is the future.

It's a better solution than Intel's, it needs work, but overall it is genius. Dynamically adjusting clock speeds based on a set power envelope... What's not to like? Hopefully they can work in vram speed as well!

If your i7 could self clock to 4.5GHz instead of 3.6 while using 77w of power instead of 56 in a dual threaded game wouldn't you be impressed? Any self respecting enthusiast who appreciates the technology aspect of this hobby would.

The problem with boost is it came packaged with strict voltage and TDP caps, what says it has to?

 

[ICDP]

5% lower clock speed is probably a 2% difference in fps.

Dynamic boost is the future.

It's a better solution than Intel's, it needs work, but overall it is genius. Dynamically adjusting clock speeds based on a set power envelope... What's not to like? Hopefully they can work in vram speed as well!

If your i7 could self clock to 4.5GHz instead of 3.6 while using 77w of power instead of 56 in a dual threaded game wouldn't you be impressed? Any self respecting enthusiast who appreciates the technology aspect of this hobby would.

The problem with boost is it came packaged with strict voltage and TDP caps, what says it has to?

That 5% difference was more than 2% in games, it scaled almost linearly to around 4% FPS loss. I agree not a lot but in some cases the difference in "stock" boost can be higher from card to card and from review to review.

My own experience of GTX680 boost was "nice but not great" because it actually hindered overclocking. I could tell my GTX680 to overclock to ~1280 core but it would frequently drop to 1230-1260ish as soon 70c was reached.

I can see the point of it to help prevent damage from overclocking beyond what a card can handle but for me 70c was just too low for the card to start down-clocking IMHO.

 

[blackened23]

Do you think we could lobby MSI to use larger fans with their twin frozr heatsink? 90mm fans make so much noise when spinning up. I swear on my life that the twin frozr series back in 2010 (with the gtx400 series cards) had larger fans, because I had a gtx465 that was dead quite up to 50% fan speed. My PE gtx670 is audible at 45% and above.

I think there is room for an easy redesign of their shroud to accommodate 100mm fans. Lets do it.

I'm not even sure if the GTX 700 series will allow custom boards. I really hope custom boards/PCBs are allowed - while the Titan cooler is fantastic, the reference PCB doesn't have the flexibility and versatility that custom boards / VRMs / coolers have. (and obviously custom PCBs are much better overclockers)

Anyway.....hopefully the 23rd will shed light on this and if custom boards are green lighted -- i'll be all about petitioning them on their forums.

 

[wand3r3r]

I wonder what ever happened to the MSI titan? There had been some hints of something and then ... nothing.

 

[blackened23]

I wonder what ever happened to the MSI titan? There had been some hints of something and then ... nothing.

Well, custom PCBs weren't allowed with the Titan. So it's pretty much a moot point to pair a TF4 on the Titan, since it's a reference PCB - doesn't make sense to create a lightning without the applicable additional power phases, higher quality VRM, etc.

Now, a few custom cooled Titans were released, yet they all still were using the reference PCB. I don't know if this will remain the case with the GTX 700 series.

 

[Face2Face]

Here you go guys,the latest news!

http://www.fudzilla.com/home/item/31429-nvidia-geforce-gtx-780-detailed

We managed to confirm the full spec of the upcoming Nvidia Geforce GTX 780 graphics card as well as some performance numbers, so we can see how well it measures up to the competition.

Judging by the slides that we saw, the GTX 780, based on Nvidia 28nm GK110 GPU, packs 2304 CUDA cores and works at 863MHz base GPU clock. It feature Nvidia's GPU Boost 2.0 that takes it up to 900MHz for the GPU. It will feature 3GB of GDDR5 memory clocked at 6000MHz and paired up with a 384-bit memory interface. The GTX 780 needs 6+8-pin PCI-Express power connectors and has 250W TDP. It comes with two DL-DVI, HDMI and DisplayPort outputs.

As noted earlier the GTX 780 reference graphics card will use the same cooler previously seen on the GTX Titan that should hold it at somewhere between 40 and 45 dBA making it much less noisy than the GTX 680.

According to an Nvidia-made performance slide that we had a chance to see, which pits the new card to AMD's Radeon HD 7970 GHz Edition, the GTX 780 should end up anywhere between 25 and 50 percent faster than the HD 7970 GHz Edition, with exception of the Tomb Raider where it is faster by around 20 percent.

In any case, we'll know for sure on the 23rd of May, when the GTX 780 is scheduled to launch. The rumored price puts it at anywhere between &#8364;500 and &#8364;600 but as always, the price is not carved in stone as Nvidia can easily change it.


This is interesting as well

http://wccftech.com/evga-teases-acx-cooler-geforce-gtx-780-geforce-gtx-770/

 

[wand3r3r]

So the specs are as follows with comparison to the titan

GeForce GTX TITAN
GK110
876 MHz boost (although usually 9x0 MHz)
2688
224
48
6GB GDDR5
384-bit
$999

GeForce GTX 780
GK110
900 MHz boost
2304
192
48
3-6GB GDDR5
384-bit
$699 (different price rumors)

 

[Smartazz]

I remember reading a guide by Kingpin about overvolting Titan. Could it work on the 780 too, especially if it's the same PCB?

 

[tviceman]

Performance looks to be exceptional. If on the miracle chance it comes in at $499 I'll definitely get one. $549 and I'll seriously consider it. $599 and up I'm waiting until Maxwell.