WEBVTT

00:00:05.080 --> 00:00:11.480
large format highres monitors this is

00:00:08.679 --> 00:00:18.560
the 30in monitor that we used to get our 2560 x600 or 1600p performance results

00:00:16.279 --> 00:00:22.960
you can consider these pretty close to what you'd get if you ran at 2560 by

00:00:20.680 --> 00:00:30.199
1440 the difference actually isn't all that much and we are all about the

00:00:26.960 --> 00:00:33.879
GeForce GTX Titan today this is the

00:00:30.199 --> 00:00:36.480
fastest single GPU card Bar None So what

00:00:33.879 --> 00:00:44.719
that means is compared to dual GPU Solutions such as the 660ti SLI that we

00:00:41.200 --> 00:00:47.079
also ran and the ASUS Aries 2 which we

00:00:44.719 --> 00:00:53.359
also ran which basically amounts to 2 7970 GHz edition cards you will get

00:00:50.960 --> 00:00:58.719
consistently good performance so somewhere between a single GPU and two

00:00:56.480 --> 00:01:03.079
gpus because those ones rely on driver optimizations one of the other things

00:01:00.480 --> 00:01:07.360
that's interesting about Titan from a large format and high resolution

00:01:05.600 --> 00:01:11.759
perspective is the fact that it has six gigs of GPU memory which means not only

00:01:09.680 --> 00:01:16.720
will One Titan perform extremely well at high res but several Titans will perform

00:01:14.439 --> 00:01:21.640
extremely well at high res Titan seeks to do a few things keep Acoustics low

00:01:19.479 --> 00:01:26.759
keep temperatures reasonable deliver outstanding performance and do it all

00:01:23.439 --> 00:01:28.400
with one GPU and reasonable power

00:01:26.759 --> 00:01:32.159
consumption so we're going to be analyzing some of those other things in

00:01:29.960 --> 00:01:37.880
other videos and this video is focused on the performance at greater than 1080p

00:01:35.759 --> 00:01:45.640
resolution now I did a full 1080p review as well and there's a 5760 by 1080

00:01:42.560 --> 00:01:50.000
surround review as well so let's have a

00:01:45.640 --> 00:01:51.799
look at crisis so crisis was always a

00:01:50.000 --> 00:01:56.920
challenge this is Crisis one by the way not crisis 2 and it continues to be a

00:01:54.640 --> 00:02:03.439
challenge so our single GPU card however runs at almost 60 FPS average in Cris 1

00:02:00.680 --> 00:02:08.959
at 2560 x600 crisis has officially been conquered this is at directx10 very high

00:02:06.039 --> 00:02:14.840
presets eight times msaa with motion blur off the only thing that beats the

00:02:11.400 --> 00:02:17.440
Titan in crisis is the Aries 2 so that's

00:02:14.840 --> 00:02:22.959
the Dual 7970 solution and you can see that the Titan actually performs almost

00:02:19.599 --> 00:02:24.319
twice as well as a GTX 680 part of this

00:02:22.959 --> 00:02:27.920
could be due to the fact that it has a much larger frame buffer although I

00:02:26.120 --> 00:02:32.000
doubt it because crisis only used up about 1 gig of vram whereas as the rest

00:02:30.680 --> 00:02:38.879
of it is probably due to the fact that it has about 2.6 uh rather 7.1 billion transistors

00:02:37.599 --> 00:02:44.720
with 2600 Cuda course so moving right along

00:02:42.800 --> 00:02:48.400
yeah Titan performs pretty well but it's about the same as GTX 660 SLI which

00:02:46.840 --> 00:02:53.879
starts to look like a pretty good value here this is an example of how well

00:02:50.800 --> 00:02:55.440
optimized SLI scaling and games that

00:02:53.879 --> 00:02:59.920
don't demand too much in terms of video memory can perform really well compared

00:02:57.640 --> 00:03:04.360
to the Titan at a lower price crisis three was just released so this one is a

00:03:02.360 --> 00:03:09.319
very very interesting case study you can see here the Aries 2 demonstrates why

00:03:07.000 --> 00:03:14.120
sometimes dual GP Solutions are not the answer the Titan tops the performance

00:03:11.799 --> 00:03:20.239
charts beating out the 660 Ti SLI solution but not by much destroying the

00:03:17.319 --> 00:03:25.720
GTX 680 again performing almost D well not quite almost double but 75% better

00:03:22.920 --> 00:03:31.200
than a GTX 680 with the Aries 2 actually coming in behind a single 7970 now the

00:03:29.920 --> 00:03:35.640
reason for that can be any number of things the Aries 2 uses a special driver

00:03:33.280 --> 00:03:38.799
because it's not a standard AMD card so ASUS has their own driver so sometimes

00:03:37.439 --> 00:03:43.040
that can be a little bit behind in terms of optimization it could be Crossfire

00:03:40.959 --> 00:03:47.480
profiles I've seen online that people are finding better success with you know

00:03:45.360 --> 00:03:53.400
using random Crossfire profiles but we were using the default ones

00:03:50.239 --> 00:03:54.879
and yeah so sometimes a dual GPU

00:03:53.400 --> 00:03:59.760
solution can perform the same as a single one or even worse which is one of

00:03:57.560 --> 00:04:03.519
the reasons why I personally always recommend a single GPU solution when you

00:04:01.959 --> 00:04:07.000
can however when you look at the performance in this game 660ti looks

00:04:05.480 --> 00:04:13.799
like a pretty good value considering it performs 90% as well as a Titan but

00:04:09.560 --> 00:04:16.359
actually costs um about 65% as much and

00:04:13.799 --> 00:04:21.440
you can also get 660 Ti in 3 gig configurations Far Cry 3 another high

00:04:19.079 --> 00:04:28.280
performance title another high resolution result here so the Aries 2

00:04:24.520 --> 00:04:30.520
and 660 Ti SLI come out way ahead of any

00:04:28.280 --> 00:04:33.800
of the single GPU Solutions probably the most impressive thing on this on this

00:04:32.639 --> 00:04:41.680
particular chart right here I think Slick's waiting for a video card here have a 7970 is the 7970 that I just

00:04:38.240 --> 00:04:44.960
handed slick the 7970 beats the GTX 680

00:04:41.680 --> 00:04:47.160
by almost 30% only losing to the only

00:04:44.960 --> 00:04:53.000
other single GPU card in our test which is the Titan and that was only by about

00:04:50.080 --> 00:04:58.000
20% so it takes a dual GPU solution both NVIDIA and AMD scaled incredibly well in

00:04:55.520 --> 00:05:01.919
this game with NVIDIA getting about almost double the scaling and AMD

00:05:00.160 --> 00:05:07.680
getting about about 50 60% scaling so that's

00:05:05.320 --> 00:05:13.000
very very very cool stuff to see but definite nod to AMD and definite nod to

00:05:10.080 --> 00:05:18.240
dual GPU Solutions here again a game where perhaps the six gig frame buffer

00:05:15.320 --> 00:05:21.360
is not a benefit to the Titan Skyrim now Skyrim we're not running the same way we

00:05:19.880 --> 00:05:25.680
used to we've added a whole bunch of mods about 18 mods from the Steam

00:05:23.360 --> 00:05:28.919
Workshop making it much more demanding than it is in its bone stock

00:05:27.440 --> 00:05:31.840
configuration which is what we used to run and that's part of the reason why

00:05:30.479 --> 00:05:37.160
when I used to show these performance graphs you'd never really see any difference between the Cards now that

00:05:35.440 --> 00:05:43.199
has all changed the big winner here again is 660 Ti SLI with Titan coming up

00:05:40.280 --> 00:05:47.759
next GTX 680 coming up next and the AMD cards bringing up the rear so this is a

00:05:46.080 --> 00:05:51.160
game that seems to be better optimized for NVIDIA although that could be a

00:05:49.319 --> 00:05:54.319
function of the mods we're running actually running better on NVIDIA so we

00:05:53.039 --> 00:05:59.759
just selected mods that we thought looked really good and kind of went from

00:05:57.080 --> 00:06:03.840
there Battlefield 3 getting to be a bit of an older game but still very telling

00:06:01.400 --> 00:06:08.520
so the Aries 3 remember this one has three gigs per GP did I say Aries 3 I

00:06:05.960 --> 00:06:12.840
meant Aries 2 has three gigs of RAM per GPU so once again we're running into a

00:06:10.800 --> 00:06:18.000
situation where we don't need that 6 gigs yet however as 4K resolution starts

00:06:16.440 --> 00:06:22.360
becoming more popular or if you want to run surround that might be more of a

00:06:20.000 --> 00:06:27.400
factor but when you don't need the extra RAM more GPU horsepower seems to really

00:06:25.360 --> 00:06:33.440
work out for you and in this particular game the Aries 2 performed almost double

00:06:29.759 --> 00:06:35.520
as well as a single 7970 making it by

00:06:33.440 --> 00:06:40.199
far the ruler of the roost in Battlefield 3 best possible solution you

00:06:37.880 --> 00:06:44.440
could get very similar results by using dual 797s at a much more reasonable

00:06:42.800 --> 00:06:49.280
price as well which is part of the reason that you shouldn't just completely ignore the Aries 2 on the

00:06:47.120 --> 00:06:54.400
graphs because that solution is available for less than the cost of a

00:06:51.400 --> 00:06:57.360
Titan not necessarily more the Aries 2

00:06:54.400 --> 00:07:03.639
actually costs about $500 $600 more than a GeForce GTX Titan so the Tian comes in

00:07:00.240 --> 00:07:05.520
a tie with the 660 Ti SLI which again is

00:07:03.639 --> 00:07:10.680
because we didn't need the vram with the other two single GPU Solutions bringing

00:07:07.759 --> 00:07:15.560
up the rear once again Metro 2033 is a very interesting case study

00:07:13.599 --> 00:07:20.280
excellent scaling again double the performance of a single 7970 with the

00:07:17.560 --> 00:07:25.000
Aries 2 highlighting why dual GPU Solutions are often very awesome however

00:07:23.000 --> 00:07:30.759
looking at the 660 Ti compared to the 680 we see why dual GPU Solutions aren't

00:07:28.360 --> 00:07:36.280
always necessarily that great Titan puts up a great showing again just destroying

00:07:33.759 --> 00:07:42.120
the numbers from the GTX 680 beating it by about 50 60%

00:07:39.000 --> 00:07:44.120
here Witcher 2 is the last game in our

00:07:42.120 --> 00:07:46.879
suite and I made a bit of a mistake on the graph but you guys won't know that

00:07:45.680 --> 00:07:53.159
because I'll fix it before I show it to you so don't worry about it uh so Witcher 2 at 1600p Ultra preset

00:07:50.879 --> 00:07:57.840
including Uber sampling really didn't give as much of an edge to the Titan as

00:07:54.919 --> 00:08:02.000
I had thought because Titan relies on lots of vram and Uber sound Ling just or

00:08:00.240 --> 00:08:06.840
rather Titan relies on lots of vram to separate itself from Dual GPU Solutions

00:08:04.639 --> 00:08:11.159
um Witcher 2 I thought would demand more vram because of the fact that we had

00:08:08.759 --> 00:08:15.520
Uber sampling on but the Aries 2 pulls away from everything else in Witcher 2

00:08:13.240 --> 00:08:20.840
just proving that sometimes just more GPU horsepower is better than more

00:08:18.240 --> 00:08:25.000
memory or a larger single GPU however the Titan does put up a pretty good

00:08:22.479 --> 00:08:31.599
showing again tying the 660 Ti and beating the GTX 680 by about

00:08:28.039 --> 00:08:34.320
50% so what do all of this mean it means

00:08:31.599 --> 00:08:39.719
that looking at current resolutions like 2560 x600 if you're only running a

00:08:36.560 --> 00:08:43.440
single monitor you may actually not get

00:08:39.719 --> 00:08:44.600
$1,000 worth of benefit with a GTX Titan

00:08:43.440 --> 00:08:51.200
especially when you compare it to Solutions like two GTX 660 Ti in SLI or

00:08:47.920 --> 00:08:54.480
two competing cards running in crossfire

00:08:51.200 --> 00:08:56.360
however if you were to run a multi-way

00:08:54.480 --> 00:09:00.120
configuration Titan would be the solution to go with because running a

00:08:58.560 --> 00:09:03.240
multi-way configuration presumably you're running at a very high resolution

00:09:01.680 --> 00:09:07.120
whether it's 4K or whether it's in surround and that's where you'll

00:09:05.760 --> 00:09:11.279
hopefully start to benefit from the additional RAM and the other situation

00:09:09.600 --> 00:09:16.279
where Titan starts to make sense is in a small form factor configuration Aries 2

00:09:14.160 --> 00:09:20.640
pulled a good couple hundred more watts from the power supply than Titan that

00:09:18.560 --> 00:09:26.040
does show up on your power bill and that does affect both the heat output as well

00:09:23.040 --> 00:09:27.160
as the noise of your system so Titan's a

00:09:26.040 --> 00:09:31.640
great solution if you're looking to build a compact gaming rig so this is

00:09:29.440 --> 00:09:36.480
another situation where dual 660 Ti might not make sense because you need at

00:09:33.720 --> 00:09:41.360
least four expansion slots to even fit those so do you have to pay a premium

00:09:38.640 --> 00:09:46.959
for that single GPU single card high performance solution yes but if you're

00:09:44.519 --> 00:09:51.440
trying to work within a confined space if you're trying to get more vram or if

00:09:49.560 --> 00:09:57.040
you're a professional who wants to make use of the fact that the double

00:09:54.519 --> 00:10:01.519
Precision floating Point performance of this card is not crippled like it was on

00:09:59.480 --> 00:10:05.360
previous generation consumer models then something like a Titan might make sense

00:10:03.680 --> 00:10:08.720
thanks for checking out this video don't forget to subscribe to lus Tech tips for

00:10:07.000 --> 00:10:11.720
unboxings reviews and other computer videos