GeForce GTX 760 Review: GK104 Shows Up (And Off) At $250

Page 4 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

mapesdhs

Distinguished
Mousemonkey writes:
> Send me the full versions of those programs ...

Yeah right, that's a good one. :D Just download and run the free editions, you don't need the full
versions. Just running in default Performance mode is sufficient to obtain useful data.


Novuake writes:
> You clearly have not had much to do with Mouse on the forum... You are just freaking out over nothing. Chill...

(it just irritates me that people post b.s which might result in someone making a poor purchasing decision)

Ah, so am I falling for a wind-up? And there I was assuming those posting here were sensible... :}

Btw, you're 100% right about my site, it has absolute diddly beyond basic text and images.


Dave Rainey writes:
> @mapesdhs: I was unable to locate on your sight where you tested 560ti SLI vs 680 OC. Please advise

See my earlier post, I gave some 3DMark11 links. And I was comparing to a typical 670/680, not
an oc'd version (at least, I don't think the 670 belonging to the person who sent me the data was an
oc'd edition). I don't yet own a 680; I consulted numerous site 670/680 reviews to find stock/oc
670/680 data for the various 3DMark tests, but I've not yet typed up my 3DMark Vantage/11/13
data, not had the time. Just search 3dmark.com and site reviews, plenty of examples, eg. for Fire
Strike Extreme see:

http://www.3dmark.com/fs/518352

and compare to:

http://www.legitreviews.com/article/2201/9/

Or for 3DMark11 see:

http://www.3dmark.com/3dm11/6035982
http://www.3dmark.com/3dm11/6037434

and compare to (note I think the toms page is just using Graphics scores, so careful
how you compare the numbers):

http://www.tomshardware.com/reviews/geforce-gtx-670-test-review,3217-11.html
http://www.guru3d.com/articles_pages/geforce_gtx_680_review,22.html

(is there a way of referencing all the results on 3dmark.com submitted by a particular user?)


matto17secs writes:
> A pair of stock GTX 560 Ti's in SLI is +11% over a GTX 580.

As I said before, it depends on the test, the platform, all sorts of things. Coming up with some kind
of overall average in that way really isn't that useful in answering real-world questions. It's fine as
a general guide, just like most site reviews, but in reality things are more complicated, eg. my 560Tis
are definitely not stock editions - in SLI they're almost 40% faster than a 797MHz GTX 580 for 3DMark11
Graphics score (see this compare), a difference which I expect to see reflected in tests like CoJ.

This is why I've been using a range of different tests that exhibit varying performance behaviours, to
show how each card responds in different ways depending on how the test settings are changed. I might
buy a newer card next month, perhaps a 770 or 780; not sure yet if I can justify the cost though as I
really need to obtain a newer AMD aswell.

Btw, I expect techpowerup is comparing with stock 822MHz 560Tis which in SLI are indeed I expect
slower than a stock 670. It's the better 560Tis like mine which can match or beat a 670. Indeed, toms
shows the Graphics score for a stock 560Ti is 4270 (don't get confused with the GTX 560Ti 448-core
which is a cut-down GTX 570), whereas one of my 900MHz Tis gives 4724; this difference scales nicely
with SLI. And btw, my system is only a P55, so I'm limited to x8/x8. I've not yet tested the cards with
my much quicker 2700K or 3930K rig (might make a difference to tests like Fire Strike Extreme, CoJ,
etc.), so it's not as if I've been running these 560 Ti tests with any kind of platform advantage; indeed,
if anything it shows just how good P55 was and still can be. :)

Ian.

 

kinggremlin

Distinguished
Jul 14, 2009
574
41
19,010
I don't know if it is a sign to come, but prices on AMD cards are quite volatile at the moment, and if you keep looking, you can catch some great deals that make most of these debates irrelevant.

I just picked up a Sapphire 7970 OC for $318 plus a $20 rebate from Newegg. That's a 7970 with a factory Ghz setting for under $300 plus it came bundled with 5 games (Borderlands 2, Crysis 3, Bioshock Infinite, Tomb Raider, and Far Cry 3 Blood Dragon). The 760 and anything in the $300-400 range from NVidia is decidedly less interesting if deals like this one become more common.
 

mapesdhs

Distinguished


Blimey! I wish they were that cheap here. Typical normal 7970 in the UK is more like the equivalent
of $440 US, while a GHz edition is in the $520 range.

Ian.



 

bustapr

Distinguished
Jan 23, 2009
1,613
0
19,780
seems like a great card at $250 and is the best choice at the price. but anyone who is interested in also buying one of the games included in the 7950boost bundle should definitely consider the 7950boost. I am seeing a good price drop on said card in the near future.
 

FormatC

Distinguished
Apr 4, 2011
981
1
18,990
For all the guys asked me about the GTX 580:

CUDA-Fluidmark-GTX-580.png


MSAA and PostFX are the death for the GTX 760. The Fluidmark is a composite benchmark and the half of the final result depends also on the graphics performance.
 

slicedtoad

Distinguished
Mar 29, 2011
1,034
0
19,360
I realise I'm not the average consumer but the thing about 6xx and 7xx nvidia cards that bugs me is their lack of overclocking ability. Not because they get too hot (I wc) but because they have locked voltage controllers.
 

svvparakala

Honorable
Jun 25, 2013
26
0
10,530


I'm new to the scene, so I don't know if a 25% increase is worth $60
 

The_Trutherizer

Distinguished
Jul 21, 2008
509
0
18,980
Sounds like great value. My concern at this point is that all things said and done it is still serious bucks for hardware (for me at least). And Maxwell is coming out early next year. I generally want my new purchases to be relevant for at least a year or two before I put my money down.
 

mapesdhs

Distinguished


Stock models, no; oc'd better models, yes. I've made this distinction clear several times, and unlike you,
I have the data to prove it, with numerous examples given.. However, your agenda is clear now; you've
provided no data, so you've lost the argument, end of discussion, QED. :D

Ian.

 

The data is all over the internet and some has even been provided by others.
 

mapesdhs

Distinguished


Yes, and yesterday I gave example links that proved both of the points in my previous post,
whereas you've provided absolutely nothing whatsoever. The more you post in this way,
the more I infer you really don't care about helping others make better buying decisions.
Beats me what your angle is, some kind of weird kick I guess.

Ian.

 

omgoblins

Honorable
Feb 25, 2012
20
0
10,510
I still enjoy reading these articles, but they are frustrating. As you can see from the [overclocking] section, the 600 and 700 series Nvidia cards are just not overclock friendly, because they are already "souped up" to a very high percentage of their max overclock by the reference card design and furthermore by the aftermarket designs. Then when the benchmarks are done, the specs for all cards are the base reference model numbers, not taking into account overclocking. Well, it has been proven again and again that the Radeon 7000 series, especially the 7950, is the opposite. It appears to set at a very low percentage of it's max overclock in the reference design which was a very bad move by AMD. The 7950 has shown to be quite the capable overclocker achieving over a 40% increase in clock speeds in max overclocks. If you were to compare a max overclocked 7950 versus a max overclocked 760, the benchmarks would paint a very different picture.

For instance, I will use the MSI cards as an example. I personally have the MSI R7950 3GB card, which is stock overclocked to 880Mhz compared to the reference 850 (with 925Mhz boost- and really can we stop even showing the non-boost 7950, what's the point? they cost the same now). While the MSI 760 has a slightly better cooler (Twin Frozr 4 versus 3 on the 7950) and a base overclock of 1020 (1085mhz w/ boost) versus the reference 980 (1033Mhz w/ boost).

So the MSI version of the cards boost base clock by 30Mhz on 7950 and 40Mhz on 760. Going from the information in this article the max overclock the 760 could attain was 1150 (1280Mhz w/ boost) or a gain of 7.84% or like they said in the article, 8%.

I was easily able to obtain a stable torture tested overclock on my MSI using Afterburner only, with no touching the default fan profile or voltage or any other settings of thus: Base clock: 1200Mhz Memory 1700 both of these settings are just shy of 40% increase, with no increase in voltage and staying within safe temperatures.

Now I am an enthuist, and this article and this card is supposedly pointed toward enthuists so why skew the debate and information by only using the reference numbers and not the real world potential numbers? Sure it's AMD's fault for going with such low reference clocks, but that doesn't mean these cards aren't capable of pushing a LOT more power and like I said I didn't raise the voltage at all, so while they will be using more power, it's not going to be an astronomical difference like it would be if you raised the voltage.

This is a flawed way to determine the actual difference, but if we assume that the 8% and 40% increase in speeds directly correlate to the same increase in FPS we end up with hugely different results. With the 7950 smashing the 760, every, time. We see the 7950 beating the 770 as well that has the same overclock ceiling of 1300mhz w/ boost as the GK104 chips (according to this article).

I'm not saying this is a perfect scientific argument, but there is ALWAYS the ignoring of the overclocking potential of the 7950 in these benchmarks, not saying the methods are biased, because I understand why they use reference and it's really AMDs fault, but it's not giving the readers the full picture. I can guarantee my R7950 overclocked will outbench and 770 or lower Nvidia card with the same components. I use an i5-3570k btw and a Z77 Black Edition motherboard with 16GB nm30 Samsung DDR3. My system is built to overclock, much like this benchmark test system only I am not afraid to show the real Radeon results and power.

Dont get me wrong though I LOVE that these new rebranded Nvidia cards are pushing AMD prices down because it just makes the deal that much sweeter! BTW tomshardware prices are jokes as well, they have to use the absolute worst prices I guess, but it just makes AMD look worse. $250 for a 7870???? I just bought an Asus 7870 Boost 2GB for $185 they go on sale from $175-200 many times in the past few months, mostly Sapphire which is a very solid brand.

Nvidia 600/700 only makes sense if you are afraid to overclock.
 


Best part is, they do not correlate to equal FPS, the HD7950 benefits A HELL OF A LOT more from a given "percentage" OC than any Kepler card. So 10% on a Kepler card is a lot less in actual performance gain than you would get from a 10% OC on an HD7950(or most GCN cards for that matter).

BUT keep in mind that very few people actually overclock to a significant degree, so the conclusion remains the same for the average user...
 

omgoblins

Honorable
Feb 25, 2012
20
0
10,510
These articles really suck when trying to educate someone who wants to be an "enthusiast"

They said it themselves in this article: 760 only overclocks 8% and GK104 chips (770, etc) hit a ceiling at 1300Mhz w/ boost.

AKA Nvidia set really good stock overclocks on the reference model, leaving NO room for user improvement. GREAT! if you are afraid to overclock.

While the 7950 (boost or no) is faster clock for clock, but it's reference speeds are WAY lower than it's max overclock and much lower than the reference nvidia speeds (850 versus 1033). That's 82% lower, yet they trade blow for blow with the 7950 coming out on top big in some games.

Yet I am able to easily, using afterburner software (NO VOLTAGE INCREASE), overclock my MSI R7950 to 1200 base clock, an increase of almost 40%! So using that same rate of 82.28% that means it's speed is equivalent to a GK104 card clocked at 1458Mhz. While the 7950 has 50% more VRAM, 384 vs 256 bit bus and performs really well in productivity apps as well.

don't get me wrong, I'm thrilled this thinly veiled GK104 garbage will continue pushing AMD prices down. I already got an Asus 7870 Boost 2GB for $180. I guarantee it overclocks better than the 760 to reach similar levels - $70 CHEAPER!
 

omgoblins

Honorable
Feb 25, 2012
20
0
10,510


That's awesome to hear (and what I was thinking) given how the 7950 has a better infrastructure in VRAM, Bus Width, etc.

I know that the average user wont overclock to extremes, but it's very easy to do simple overclocks and I think most readers of tomshardware articles like this should know how to do it (literally taking 1 minute tops with conservative speed overclocks).
 


You would be surprised how few people OC enough to make a difference. But yeah, I am still happy with my setup despite the nice price of this cards, it is almost a year later, so the pricing makes sense.
 

JPNpower

Honorable
Jun 7, 2013
1,072
0
11,360


If you go by straight maths, then a 25% increase in performance for 30% increase in price is not efficient. That is going by the $200 GTX660 on sale at my microcenter right now, and the $60 increase you claim with average 760 cards.
 

This is the comment section not the help section.
 

jrstriker12

Honorable
Jun 13, 2013
100
0
10,680


If you want a more powerfull card and have $60 to spare, then yes.

Also depends on the games you play, looks at the benchmarks for the games they tested and find out if the 760 gives you the boost you want at the level of detail and resolution that you plan to run.

BF3, it might be worth it. Far Cry 3 maybe not....
 
Status
Not open for further replies.