WEBVTT

00:00:00.120 --> 00:00:05.759
this video is brought to you by CM Storm Preferred Choice of mice headsets and

00:00:03.919 --> 00:00:11.200
mechanical keyboards of curse gaming visit www. coolermaster usa.com teamor

00:00:09.760 --> 00:00:19.359
curse for more details now this review of the GTX 780

00:00:15.120 --> 00:00:21.720
at 2560 by 1440 really um was an

00:00:19.359 --> 00:00:27.119
emotional time for me because I used to run these benchmarks on my 30in Samsung

00:00:24.760 --> 00:00:32.399
monitor which my cat destroyed several months ago so we've obtained a new

00:00:29.519 --> 00:00:37.120
monitor which is 2560 x440 instead of 2560

00:00:33.680 --> 00:00:38.920
x600 and we are back to the races

00:00:37.120 --> 00:00:44.399
benchmarking graphics cards and telling you at high resolutions greater than

00:00:41.559 --> 00:00:48.680
1080p what is the best card for your dollar or in the case of something like

00:00:46.600 --> 00:00:54.039
a GTX Titan $1,000 now I'm going to get through this

00:00:51.600 --> 00:00:57.719
methodology stuff as quickly as I can we use a

00:00:54.840 --> 00:01:02.120
3960x overclock to 4 GHz cooled by an h100 we use a p9x 79 Deluxe motherboard

00:01:00.719 --> 00:01:08.080
we use an antech high current Pro thousand wat 128 gig SSD and we use a

00:01:06.040 --> 00:01:10.840
variety of graphics cards obviously for our testing we're running on Windows 7

00:01:09.640 --> 00:01:16.560
and we are running the latest beta drivers for both AMD and NVIDIA we run

00:01:13.640 --> 00:01:21.200
all graphics cards overclocked because overclocking is free I guess if your

00:01:19.159 --> 00:01:24.280
time's not worth anything but it really doesn't take that long because all you

00:01:22.799 --> 00:01:28.759
have to do is check out the overclocking settings we're using in the graph or

00:01:26.920 --> 00:01:31.759
spreadsheet that's Below in the video description and that'll give you you

00:01:30.159 --> 00:01:36.399
some idea where to start so it really shouldn't take you very long at all

00:01:34.320 --> 00:01:41.159
we're testing our games using fraps and because we're not doing any dual GPU or

00:01:38.720 --> 00:01:44.640
triple GPU setups it should still Fair be a fairly accurate representation of

00:01:43.200 --> 00:01:48.040
the performance of the cards however it should be noted that particularly if

00:01:46.560 --> 00:01:51.880
you're running SLI or Crossfire configurations it's important to use a

00:01:50.200 --> 00:01:56.759
frame capture setup to get the most realistic possible results now all of

00:01:55.040 --> 00:02:01.680
our games are not only being run at high resolutions but also at very high detail

00:01:59.399 --> 00:02:04.840
settings because well the justification for it is if you're going to spend $300

00:02:03.280 --> 00:02:09.360
on a graphics card you better be running at high details in this one we see

00:02:07.399 --> 00:02:13.319
what's going to become a recurring theme and that is that memory bandwidth and

00:02:11.200 --> 00:02:18.080
memory amount seem to be holding back the 660 Ti and the GTX 680 this is

00:02:16.280 --> 00:02:23.640
something we did not observe in our 1080p benchmarks so there is a very

00:02:21.239 --> 00:02:30.440
clear division where Titan and 780 are up here 7970 and 7950 are here and 680

00:02:27.280 --> 00:02:33.160
and 660 Ti are trailing at the bottom

00:02:30.440 --> 00:02:37.920
in Tomb Raider 2013 it really is men against boys with the Titan and the GTX

00:02:35.360 --> 00:02:41.360
780 coming out on top however of course they are priced much higher than the

00:02:39.720 --> 00:02:49.200
cards that they are now replacing from NVIDIA's own lineup and the 7950 and

00:02:45.519 --> 00:02:52.280
7970 Metro last Light is a game that uh

00:02:49.200 --> 00:02:55.080
actually kind of creamed all of these

00:02:52.280 --> 00:03:00.280
cards even the GTX Titan only ran at just over 20 FPS average however I mean

00:02:58.519 --> 00:03:06.799
okay so we learned that you need need multiple gk10 gpus in order to run this

00:03:03.440 --> 00:03:09.080
game smoothly at 2560 x440 we also

00:03:06.799 --> 00:03:13.480
learned that sometimes it's not feasible on current generation Hardware to turn

00:03:11.120 --> 00:03:18.400
games up to the very maximum but with all that said we still do see a clear

00:03:16.480 --> 00:03:22.480
difference in performance between that top tier gpus the mid- tier where those

00:03:20.920 --> 00:03:27.400
AMD cards are benefiting from their extra memory bandwidth and extra memory

00:03:24.560 --> 00:03:32.640
and then the older being phased out 600 series of GeForce graphics cards

00:03:30.439 --> 00:03:36.519
I'm going to do Far Cry 3 and Far Cry 3 blood Dragon together because even

00:03:35.040 --> 00:03:41.000
though the performance in these two games is quite different with Far Cry 3

00:03:38.959 --> 00:03:45.599
being much more demanding due to the different graphical style the way that

00:03:43.319 --> 00:03:49.760
the cards are segmented against each other in terms of performance is very

00:03:47.760 --> 00:03:54.720
similar and this is one of the ones where I mean NVIDIA is quoting around a

00:03:51.840 --> 00:04:01.760
35% Improvement on average going from a 680 to a 780 in this one we see that

00:03:57.519 --> 00:04:04.360
figure just destroyed at 2560 X 1440 a

00:04:01.760 --> 00:04:09.000
GTX 780 is going to give you almost double the performance of the 680 card

00:04:06.920 --> 00:04:13.519
that it's replacing and almost double the performance of something from the

00:04:11.239 --> 00:04:18.680
competing red team such as a 7970 it does cost significantly more

00:04:16.079 --> 00:04:25.280
than a 7970 but there you have it if you are a Far Cry 3 afficianado Guys the 780

00:04:23.080 --> 00:04:29.120
is worth a look see if you want to play at high

00:04:27.000 --> 00:04:35.680
resolutions let's wrap it up with the classic but can it run Crisis 3 and the

00:04:33.120 --> 00:04:40.280
answer is yes the gtx780 destroys its predecessor in Crisis 3 and we can

00:04:38.199 --> 00:04:47.720
attribute this in all likelihood to the fact that it has 50% more Cuda cor 50%

00:04:43.720 --> 00:04:51.039
wider memory bus and 50% more RAM giving

00:04:47.720 --> 00:04:53.240
it just 50% more muscle everall it is a

00:04:51.039 --> 00:05:00.199
true replacement for the 680 in this case and yeah again if you're a Crisis 3

00:04:57.160 --> 00:05:03.600
lover then the 780 is definit itely

00:05:00.199 --> 00:05:06.960
worth a look it is a significantly more

00:05:03.600 --> 00:05:09.199
powerful GPU with its Titan DNA and

00:05:06.960 --> 00:05:14.160
really much closer to Titan performance than anything else and of course it's

00:05:11.919 --> 00:05:18.919
much lower price point compared to the GTX

00:05:16.160 --> 00:05:23.039
Titan now the conclusion isn't really as cut and dried as this is the best

00:05:20.960 --> 00:05:26.520
performing GTX numbered series card so it's pretty much the one to buy however

00:05:25.400 --> 00:05:32.039
it's pretty clear that if you were shopping for a Titan the 780 is

00:05:29.000 --> 00:05:33.680
definitely worth having a look at it has

00:05:32.039 --> 00:05:37.479
most of the performance of a Titan particularly after you overclock it and

00:05:35.360 --> 00:05:42.919
comes in at a much lower price point it's an obvious successor for the GTX

00:05:40.360 --> 00:05:46.880
680 which looks distinctly last generation in terms of its performance

00:05:44.919 --> 00:05:50.840
compared to the 780 but things get a little bit muddier when you consider

00:05:48.759 --> 00:05:56.160
that something like a 7970 from the red team does offer up a lot of the

00:05:53.240 --> 00:06:00.720
performance that you'll need to run games at medium to high settings even at

00:05:58.520 --> 00:06:04.880
high resolutions it has three gigs of memory and because of the never settle

00:06:02.479 --> 00:06:09.400
bundle it comes with a bunch of really cool games whereas NVIDIA cards right

00:06:07.360 --> 00:06:13.440
now are limited to just coming with Metro last Light but then of course

00:06:11.560 --> 00:06:17.160
there's the intangibles to weigh in such as the fact that you can actually

00:06:14.919 --> 00:06:21.720
upgrade these gpus to multi-GPU configurations in SLI and get a tangible

00:06:20.160 --> 00:06:25.440
noticeable performance Improvement there's the fact that they run quieter

00:06:24.000 --> 00:06:30.400
um there's the fact that GeForce experience is going to make things a little bit more TurnKey so you don't

00:06:28.720 --> 00:06:35.720
have to go in and tweak things and tune them and remember to download your latest drivers which honestly for most

00:06:34.240 --> 00:06:39.479
of the people watching this video right now it's probably not a factor but if

00:06:37.880 --> 00:06:44.919
you're recommending a card to someone who is an avid gamer but not a hardware

00:06:41.800 --> 00:06:47.599
Enthusiast might be something to

00:06:44.919 --> 00:06:53.000
consider so there you have it guys thank you for checking out the Linus Tech tips

00:06:49.800 --> 00:06:55.919
2560 by 14440 performance review of the

00:06:53.000 --> 00:07:02.759
GeForce gtx780 don't forget to subscribe to lineus Tech tips for more unboxings

00:06:58.160 --> 00:07:02.759
reviews and other computer videos